EXPERIMENTAL NEUROLOGY
ARTICLE NO.
146, 159–170 (1997)
EN976507
Parkinsonism Reduces Coordination of Fingers, Wrist,
and Arm in Fine Motor Control
Hans-Leo Teulings, José L. Contreras-Vidal, George E. Stelmach, and Charles H. Adler*
Motor Control Laboratory, Arizona State University, Tempe, Arizona 85287-0404; and *Mayo Clinic, Scottsdale, Arizona 85259
This experiment investigates movement coordination in Parkinson’s disease (PD) subjects. Seventeen
PD patients and 12 elderly control subjects performed
several handwriting-like tasks on a digitizing writing
tablet resting on top of a table in front of the subject.
The writing patterns, in increasing order of coordination complexity, were repetitive back-and-forth movements in various orientations, circles and loops in
clockwise and counterclockwise directions, and a complex writing pattern. The patterns were analyzed in
terms of jerk normalized for duration and size per
stroke. In the PD subjects, back-and-forth strokes,
involving coordination of fingers and wrist, showed
larger normalized jerk than strokes performed using
either the wrist or the fingers alone. In the PD patients,
wrist flexion (plus radial deviation) showed greater
normalized jerk in comparison to wrist extension (plus
ulnar deviation). The elderly control subjects showed
no such effects as a function of coordination complexity. For both PD and elderly control subjects, looping
patterns consisting of circles with a left-to-right forearm movement, did not show a systematic increase of
normalized jerk. The same handwriting patterns were
then simulated using a biologically inspired neural
network model of the basal ganglia thalamocortical
relations for a control and a mild PD subject. The
network simulation was consistent with the observed
experimental results, providing additional support that
a reduced capability to coordinate wrist and finger
movements may be caused by suboptimal functioning
of the basal ganglia in PD. The results suggest that in
PD patients fine motor control problems may be caused
by a reduced capability to coordinate the fingers and
wrist and by reduced control of wrist flexion. r 1997
Academic Press
INTRODUCTION
Parkinson’s disease (PD) is caused by the degeneration of nigrostriatal neurons resulting in a reduction of
the neurotransmitter dopamine (25, 26). Apart from
the well-known movement control problems, slowness,
reduced movement amplitudes, and prolonged decelera-
tion times, PD patients are hypothesized to suffer from
difficulties in the coordination and control of various
muscle systems (15). For example, PD patients show a
delay of the onset of the opening of the hand relative to
the initiation of the transportation of the forearm (6).
Whereas normal subjects can smoothly modify an
ongoing movement, PD patients initiate a corrective
movement only after completing the initial movement
(15). Temporal dissociation has also been observed
between the left and the right arms in PD patients (22).
When fingers and thumb of the same arm were analyzed, Benecke et al. (2) found that PD subjects showed
a substantial impairment when performing an isotonic
elbow flexion while isometrically squeezing a force
transducer. Similarly, Isenberg and Conrad (20) observed that PD subjects do not initiate components of
arm movements simultaneously, resulting in angular
or curved movement trajectories. These data suggest
that in PD patients coordination is reduced in movement patterns that require control of a large number of
muscles and joints.
PD leads to a disruption in the execution of practiced
skills such as handwriting (25, 41). Boisseau et al. (3)
observed that PD handwriting can be characterized by
various types of dysfluencies: lack of control, abrupt
changes of direction, tremor, slowness, hesitation, rigidity, variability of baseline, and, in some cases, micrographia. The breakdown of handwriting may come
from the inadequate control of acceleration amplitude
(41), which is also reflected by abnormal EMG amplitudes (42) and by reduced coordination of independent
movement components (20).
Handwriting-like movement patterns are ideal tasks
to study motor control because they are well learned.
Many potential artifacts due to muscle strength limitations, gravity, inertia, visual feedback, and insufficient
practice are absent. More importantly, coordination
complexity can be varied by using handwriting patterns involving different combinations of finger, wrist,
and arm movement components. In Western cursive
handwriting, finger movements (i.e., flexion and extension of the thumb and the opposing index and middle
fingers) generate primarily up-and-down strokes (i.e.,
159
0014-4886/97 $25.00
Copyright r 1997 by Academic Press
All rights of reproduction in any form reserved.
160
TEULINGS ET AL.
away and toward the body in the horizontal plane)
while wrist flexions and extension (plus radial and
ulnar deviations) generate primarily the small left-andright movements. Forearm movements generate primarily the large left-to-right progression (43). Therefore,
coordination complexity can be manipulated by selecting different handwriting patterns. For example, horizontal back-and-forth movements require mainly the
wrist joint, vertical back-and-forth movements require
mainly the finger joints, and oblique up-and-down
strokes require a combination of finger and wrist
movements that are in phase (e.g., 34). Circles require
independent (i.e., out of phase) control of the fingers
and wrist. Loops require an additional left-to-right
progression by the forearm, which further magnifies
the coordination complexity, irrespective of whether
the progression is continuous (e.g., 17, 35) or discontinuous (e.g., 4).
Reduced coordination is defined here as improperly
timed initiation and disproportionate activation of independent muscle systems resulting in multiple acceleration peaks which extend movement duration. Such
movement problems contribute to increased jerk levels
because jerk is the change of acceleration per time (e.g.,
15). However, because jerk level depends on the size
and the duration of the movements it needs to be
normalized. The advantage of normalized jerk is that
coordination difficulties in patterns of different shapes,
sizes, and durations can be compared.
We hypothesized that normalized jerk per handwriting stroke increases with the number of degrees of
freedom to be coordinated (e.g., wrist, fingers, wrist
plus fingers, and wrist plus fingers and arm), particularly in PD subjects relative to the elderly subjects. To
help understand our experimental data, writing patterns were also simulated by a biologically plausible
model of normal versus PD basal ganglia thalamocortical interactions (7–9).
EXPERIMENT
Subjects
Seventeen PD patients (14 males and 3 females; ages
42–78 years, mean age is 65 years) and 12 elderly
control subjects (7 males and 5 females; ages 53–78
years, mean age is 67) participated in this study after
providing informed consent. PD and elderly subjects
were paid for their participation. All PD subjects were
in mild stages of the disease. All but one reported that
they were unilaterally affected (8 right sided, 8 left
sided, and 1 bilateral). The median duration of PD since
diagnosis was 3 years. Ten PD patients reported having
bradykinesia and 9 reported having rigidity. Three PD
patients showed action tremors in their handwriting
with frequencies between 6.3 and 6.8 Hz. Nearly all PD
patients reported micrographic writing impairments; 2
did not. All subjects were right handed except 1 PD
patient, who wrote with the left hand. These selfreports were not significantly correlated with stroke
duration, size, or normalized jerk.
Apparatus
The subjects wrote on a digitizer–display (Wacom
PL-100V) controlled by an Intel 80486-based personal
computer. The digitizer sampled the x and y coordinates of the pen tip 200 times per second with a spatial
error of 0.05 mm. The digitizer–display (28 cm wide 3 23
cm high 3 1.5 cm thick) rested horizontally on top of a
table and was oriented to meet each subject’s preference.
Writing Patterns and Instructions
Nine writing patterns were used: back-and-forth
strokes in four different orientations—left-and-right
(i.e., horizontal), away-and-toward the body (i.e., vertical), and two oblique orientations (i.e., forward slanted
and backward slanted, respectively); continuous circles
in clockwise and counterclockwise directions; repetitive
loops in clockwise and counterclockwise directions (i.e.,
cursive ‘‘jjjjjjjj’’ and ‘‘llllllll’’); and a complex writing
pattern (i.e., cursive ‘‘ljielije’’). Subjects were asked to
suppress dotting of the ‘‘i’’ and ‘‘j.’’
The left-to-right progression is estimated in the loops
by comparing sequences of two loops (requiring no
progression), four loops, and eight loops (requiring
substantial progression). The effects due to different
numbers of strokes can be estimated separately by
comparing all nonprogressing patterns when performing two, four, and eight ‘‘letters.’’ The subjects were
instructed not to count the number of repeated strokes
while producing the patterns. As a result, some trials
could have fewer or more strokes than required. The
experimenter gave feedback when the number of strokes
was out of range so that the trial could be redone.
Once a comfortable position was adopted the subjects
were requested to keep the orientation of their forearm
and the digitizer constant. The angles of the forearm
and of the digitizer relative to the table front were
recorded. All subjects spontaneously positioned their
writing arm perpendicular (615°) to the horizontal
baseline. Therefore, the horizontal strokes are mainly
performed by the wrist joint and the vertical strokes by
the finger joints. Oblique strokes were produced by
both fingers and wrist.
The handwriting task was shown as standard-font
letters on the display screen to motivate subjects to
preprogram the movement patterns rather than copying. The subjects were asked to write at comfortable
161
COORDINATION IN PARKINSONIAN HANDWRITING
size and speed. No speed instruction was given because
elderly controls are likely to change their writing more
than PD subjects (41). The writing area of 18 by 10 cm
was centered on the digitizer and had a horizontal
guide line 3 cm from the bottom. Subjects were instructed to write on this guide line.
Procedure
Subjects performed three blocks of 4-, 8-, and 16stroke patterns in a random sequence. Within each of
these blocks the nine patterns were performed in a
random sequence. Each pattern was replicated at least
six times (three times by the three slowest parkinsonians) so that the duration of the experiment session
was about equal for all subjects (i.e., 1 h). Each time a
new pattern was to be produced for the first time the
experimenter showed an example of how the pattern
was to be performed. By blocking the replications and
pattern-length conditions, task switching was minimized. Task switching would have selectively disadvantaged PD patients (11). The subjects were familiarized
with the equipment by writing their name and a
sentence (‘‘we write llll in arizona’’) two times to check
for fatigue or medication effects during the experimental session.
Recording of the writing movements started as soon
as the pen touched the surface of the digitizer–display
and ended when the pen was lifted for more than 0.5 s
(2 s when writing the name or the sentence). The trace
of the pen position was made visible on the digitizer–
display in real time. A few seconds after the previous
trial, the subject started with the next trial. If the
experimenter or subject discovered that a trial was
inappropriate, it was redone.
Simulations
Handwriting patterns similar to those used in the
experiment were also generated using a neural network of parkinsonian and normal handwriting movements (7–9). In the handwriting simulations, a sequence of relative target position vectors was defined
according to the individual stroke direction and amplitude for each degree of freedom as follows: finger
flexion/extension producing vertical displacement
(TPVy ), forearm supination/pronation producing local
horizontal displacement (TPVx ), and radial flexion/
ulnar extension of the wrist joint producing the left-toright progression (TPVa ). At times of zero or peak
velocity the subsequent target position vector of the
motor program was fed into the VITE model, which
continuously computed the difference vector (DVx, DVy,
DVa ) between the target position (TPVx, TPVy, TPVa )
FIG. 1. Each basal ganglia–thalamocortical circuit controls 1 of
the 3 degrees of freedom in the handwriting simulations. The basal
ganglia modulate the dynamics of a central movement generator
(VITE) by gating difference vector (DV) computations between target
position vector (TPV) and present position vector (PPV) at a rate
specified by the pallidothalamic neurons. Dopamine depletion in
Parkinson’s disease has a differential effect on the direct and indirect
pathways of the basal ganglia. In particular, an overactivation of
striatal neurons in the indirect pathway causes inhibition of neurons
of the external segment of the globus pallidus (GPe) which in turn
disinhibits neurons in the subthalamic nucleus (STN). The disinhibition of STN cells causes an overactivation of neurons in the internal
segment of the globus pallidus (GPi). Furthermore, a reduced activation of striatal cells in the direct pathway further increases GPi
activity by disinhibition. This pathological increase in GPi activity
causes inhibition of thalamic neurons that project to motor cortical
areas necessary for movement production. During normal motor
behavior the direct pathway is activated to initiate movement at a
desired speed while the indirect pathway may be activated to
terminate the movement (7).
(see Appendix) and the present position vectors (PPVx,
PPVy, PPVa ). The sequence of present position vectors
forms the trajectory of the pen tip.
The horizontal and vertical coordinates of the pen tip
are derived from the present position vector (PPVx,
PPVy, PPVa ) as follows (4),
x 5 PPVx 3 cos (PPVa)
1 (length 1 PPVy) 3 sin (PPVa),
y 5 2PPVx 3 sin (PPVa)
1 (length 1 PPVy) 3 cos (PPVa),
162
TEULINGS ET AL.
FIG. 2. Examples of handwriting patterns used in the experiment, performed by an elderly control. The writing patterns were (a)
horizontal back-and-forth strokes, (b) back-and-forth strokes slanted forward, (c) vertical back-and-forth strokes, (d) back-and-forth strokes
slanted backward, (e) counterclockwise circles, (f ) clockwise circles, (g) a sequence of counterclockwise loops, ‘‘llllllll,’’ (h) a sequence of
clockwise loops, ‘‘jjjjjjjj,’’ and (i) a complex writing pattern, ‘‘ljielije.’’ The dots mark the segmentation points between strokes. For each of the
patterns we show the vertical (in the back-and-forth movements, upward or leftward along the orientation) velocity V as a function of time and
the normalized jerk level per stroke. The horizontal back-and-forth movements are performed by alternating flexion plus ulnar deviation and
extension plus radial deviation of the wrist joint. The normalized jerk is rather constant for successive strokes. The patterns were clipped after
2.5 s.
where length 5 7.5 cm, represents the distance between wrist joint and pen tip.1
The repetitive patterns are generated by repeating
the basic sequence of relative target positions. Withinsubject motor variability was simulated by adding
random noise to the target positions. The noise was
equal for both simulated PD and control, namely
uniformly distributed noise between minus and plus
0.025 cm (for TPVx and TPVy ) and between plus and
1 The biomechanical models used in the simulation and in Wing
(43) are not substantially different. In the simulation model the wrist
flexion and extension has the same function as the arm movement in
Wing (43). Wrist rotation has the same function as wrist extension
and flexion in the aforementioned model. Note that both models
include 3 degrees of freedom, which are involved in a similar
hierarchy with coordination complexity.
minus 3° (for TPVa ). The dynamics of the central
pattern generator are modulated by a neural model of
the basal ganglia–thalamus circuit which gates the
output of the difference vector (see Fig. 1). Therefore,
the network specifies the onsets and the speeds of each
of the degrees of freedom. In the simulation of the PD
subject, the dopamine depletion was set to 15%, corresponding to preclinical PD. In the simulation of the
control subject the dopamine depletion was set to 0%.
The reason that, in the model, a small dopamine
depletion shows a measurable effect, whereas in PD
patients depletion levels of 80% are required to show
any effects is probably because the model lacks the
compensatory mechanisms of the nervous system such
as presynaptic overactivity and supersensitivity of
receptors (32). The dynamical equations specifying the
COORDINATION IN PARKINSONIAN HANDWRITING
163
FIGURE 2—Continued
network model are explained in detail in ContrerasVidal and Stelmach (7). One hundred twenty trials of
back-and-forth strokes were generated using the PD
model and as many using the elderly control model.
Half of them started with an upstroke and the other
half started with a downstroke. This represents the
amount of data generated by 20 PD subjects and 20
elderly subjects.
Analysis
The handwriting recordings and simulated patterns
were filtered using a frequency-domain attenuation
filter with a pass band between 0 and 3 Hz and a cosine
transition band between 3 and 11 Hz in all subjects
(39). Proper filtering was required for calculating higher
order time derivatives such as jerk. To verify that the
results were not caused by a particular choice of filter
frequency all data were processed using a higher
(3.5–12.5 Hz) and a lower (2.5–9.5 Hz) transition band.
The qualitative results of this study appeared independent of the low-pass filter frequency. After filtering, the
patterns were segmented into alternating up- and
downstrokes where the interpolated vertical velocity
crosses 0. Spurious segmentation points, such as segmentation points spaced less than 0.5 mm, were removed. Back-and-forth movement patterns were rotated to a vertical orientation: 90° counterclockwise for
horizontal strokes and 45 or 245° for oblique forward
or backward strokes, respectively. Subsequently, a pattern verifier was used to check the correct performance
and the segmentation of each individual pattern. The
first stroke was removed if it was a downward or a
leftward stroke or an incomplete stroke, so that all
processed patterns were aligned with a beginning
164
TEULINGS ET AL.
FIG. 3. The same as Fig. 2 but from a PD patient instead of an elderly control. The wrist flexions show greater normalized jerk levels in
nearly all PD subjects in comparison to the wrist extensions. The patterns were clipped after 5 s.
upward or a rightward stroke. Trials with less than 2,
6, or 12 strokes in the 4-, 8-, and 16-stroke conditions,
respectively, were discarded. Trials were also discarded
when ‘‘e’’ or ‘‘l’’ had no counterclockwise loop or ‘‘j’’ had
no clockwise loop, or when the loop areas were smaller
than one of the spurious loops at the movement reversals of ‘‘i’’ and ‘‘j.’’ Finally, the complex pattern ‘‘ljielije’’
was checked for the proper sequence of stroke sizes.
Replications in which the vertical sizes of ‘‘l’’ or ‘‘j’’ were
smaller than the vertical sizes of a subsequent ‘‘i’’ or ‘‘e’’
were discarded. In both the elderly controls and the PD
subjects 10% of the trials were discarded, leaving 5.8
and 5.5 appropriate replications per condition, respectively.
The measure for fine motor control impairment
adopted here is based on the integrated squared jerk
(e.g., 15). Jerk is the change of acceleration, which is
the third time derivative of position, and possesses a
unit of length/duration3. The time-integrated squared
jerk is minimal in smooth movements (19). Because
jerk varies dramatically with duration and size of the
stroke (e.g., 33) and because the present study attempted to measure movements under preferred stroke
durations and sizes, jerk has to be normalized for
different stroke durations and sizes. This was done by
dividing integrated squared jerk by length2/duration5
per stroke (e.g., 21). Subsequently, the square root was
taken so that normalized jerk is proportional with
absolute jerk:
Œ(1⁄2 e dt j 2(t) 3 duration5/length2).
Normalized jerk is a unit-free measure and can be
compared between strokes of different sizes and durations. For example, a harmonic back-and-forth movement yields a normalized jerk score equal to p3/22 5
7.75. This is smaller than the normalized jerk of a
half-constant-velocity circular movement, which is p3/
21.5 5 10.96, thus reflecting the higher cost of producing
a curved stroke than a straight stroke.
Data across replications within subjects were averaged. This generated better results than medians,
COORDINATION IN PARKINSONIAN HANDWRITING
165
FIGURE 3—Continued
probably because the number of replications is small
(e.g., 5) so that the medians per stroke may not follow
the general trend. The resulting data per subject were
in general non-Gaussian (e.g., outliers tended to be far
too large) and showed nonuniform variances. Therefore, the nonparametric Mann–Whitney U or Wilcoxon
test between elderly and PD groups was used (e.g., 5,
13). For tests between conditions within groups the
sign test for paired data with a priori probability of 0.5
was performed.
RESULTS
The left-hand side of Fig. 2 shows the actual handwriting-like patterns produced by a control subject: backand-forth movements in (a) horizontal, (b) forward
oblique, (c) vertical, and (d) backward oblique direc-
tions, counterclockwise and clockwise (e and f ) circles
and (g and h) loops, and (i) the complex writing pattern.
The scales are calibrated in centimeters. The right side
of Fig. 2 shows the vertical velocity V (or the velocity
along the upward or rightward directions in the straightstroke patterns) as a function of time and the normalized jerk per stroke (N. Jerk). Only the first 2.5 s of the
movement are shown. The positive phases of the velocity refer to upward or rightward movements and the
negative phases to downward or leftward movements.
Each positive or negative phase is called a stroke (see
Fig. 2 legend). The normalized jerk is plotted as a bar
chart in which the bar widths are equal to the stroke
durations so that the bar chart is aligned with the
velocity patterns per stroke. The graph shows that
multimodal velocity patterns, which we think signify
coordination deficits, yield high values for normalized
166
TEULINGS ET AL.
FIG. 4. Normalized jerk per stroke in back-and-forth movements as a function of stroke direction (a) for PD and elderly subjects and (b) for
a simulated PD and a control subject. Vertical bars indicate the standard deviations of the means. To discriminate the vertical bars per subject
group, the curves have been slided in horizontal direction. For each direction the predominant finger and wrist activities are given. The PD
subjects show higher normalized jerk, especially in the oblique stroke directions requiring the coordination of wrist and fingers. Furthermore,
PD subjects show higher normalized jerk in wrist flexions than in wrist extensions.
jerk. The straight-stroke patterns (a–d) show uniformly
low normalized jerk levels close to the minimum of
7.75. The circular patterns (e–i) show slightly higher
normalized jerk levels of about 10.96. The complex
pattern (i) shows normalized jerk levels which are
generally higher.
Figure 3 shows the same patterns executed by a PD
subject. It can be seen that the PD subject produced the
movements at a lower rate. Therefore, the time scale
was extended to 5 s. The PD subject generally shows
greater normalized jerk per stroke than the elderly
subject. Furthermore, the normalized jerk per stroke
differed largely between successive strokes. Specifically, Fig. 3a shows that normalized jerk levels in the
wrist flexions were systematically higher than in the
wrist extensions.
The normalized jerk of the back-and-forth strokes
are depicted in Fig. 4a as a function of stroke direction.
In the PD subjects, strokes with oblique orientations
have greater normalized jerk than the horizontal and
vertical strokes. This effect cannot be explained by
different stroke durations as the stroke orientation
effect consists only of the backward oblique strokes
being much slower than other strokes in both parkinsonians and the elderly (see the mean durations in the
back-and-forth movements in Table 1). The orientation
effect upon normalized jerk cannot be explained by
different stroke sizes, either, as the stroke size effect
consists of the vertical strokes being smaller while the
horizontal strokes are larger than the oblique strokes
in both PD and the elderly subjects (see the mean
TABLE 1
Means and Standard Deviations of Duration, Size, and
Normalized Jerk per Stroke in the Back-and-Forth Strokes in
the Elderly and PD Subjects
Duration (ms) Size (cm) Normalized jerk
Pattern/Stroke
Mean
SD
Mean SD
Mean
Elderly control subjects
(a) Horizontal back-and-forth strokes
Left
247
117 1.01 0.31 14.9
Right
255
106 1.03 0.31 14.6
(b) Oblique slanted forward back-and-forth strokes
Upward
240
106 0.96 0.24 14.1
Downward
247
116 0.96 0.24 15.9
(c) Vertical back-and-forth strokes
Upward
247
127 0.81 0.27 14.4
Downward
237
125 0.82 0.27 13.1
(d) Oblique slanted backward back-and-forth strokes
Upward
294
144 0.96 0.32 17.2
Downward
268
132 0.96 0.33 15.8
PD subjects
(a) Horizontal back-and-forth strokes
Left
286
166 1.11 0.42 17.3
Right
306
182 1.11 0.42 19.8
(b) Oblique slanted forward back-and-forth strokes
Upward
288
168 0.99 0.28 20.0
Downward
303
191 1.00 0.27 19.6
(c) Vertical back-and-forth strokes
Upward
282
169 0.93 0.34 16.9
Downward
286
178 0.95 0.33 16.8
(d) Oblique slanted backward back-and-forth strokes
Upward
341
221 0.98 0.39 23.1
Downward
346
236 0.99 0.38 24.6
SD
10.8
9.5
8.2
11.9
7.5
7.1
11.2
12.9
14.4
16.3
16.0
16.4
16.5
15.3
23.8
28.5
COORDINATION IN PARKINSONIAN HANDWRITING
stroke sizes in the back-and-forth movements in Table
1). Therefore, the normalization of jerk for stroke size
and duration differences is appropriate.
Strokes in oblique directions that require the coordination of fingers and wrist had significantly greater
normalized jerk than the strokes using only the wrist
joint or only the finger joints, sign test(16) 5 4, P ,
0.05. This especially holds for the stroke directions in
which fingers flex while wrist extends and vice versa. In
the elderly controls, the normalized jerk difference
between oblique and other directions was much less
pronounced and nonsignificant. In addition, PD patients showed significantly greater normalized jerk in
the single-joint wrist flexions (plus ulnar deviation)
than in the extensions (plus radial deviation), sign
test(16) 5 3, P , 0.05 (see also Fig. 3a). This is
supported by the observation that one of the PD
subjects included in the data who wrote with his left
hand showed greater normalized jerk in flexion than
extension as wrist flexion in the left hand results in a
stroke in the direction opposite to that of the right
hand. In contrast to the PD subjects, the elderly control
subjects showed no differences between wrist flexion
and wrist extension (see Fig. 3a).
The additional effect of the left-to-right progression
component was estimated by comparing circles and
loop patterns. Table 2 summarizes stroke duration,
size, and normalized jerk in the circular patterns. As
can be seen in Fig. 5, the counterclockwise circles and
the ‘‘llllllll’’ patterns did not differ for both PD and
elderly control groups, whereas the ‘‘jjjjjjjj’’ pattern
showed greater jerk than the clockwise circles and
this was true for both PD and elderly subjects, sign
test(16) 5 1, P , 0.01 and sign test(10) 5 1, P , 0.05,
TABLE 2
Means and Standard Deviations of Duration, Size, and
Normalized Jerk per Stroke in the Curved Patterns in the
Elderly and PD Subjects
Duration
(ms)
Pattern
Elderly control subjects
(e) Circles counterclockwise
(f ) Circles clockwise
(g) Loops counterclockwise
(h) Loops clockwise
(i) Complex writing pattern
PD subjects
(e) Circles counterclockwise
(f ) Circles clockwise
(g) Loops counterclockwise
(h) Loops clockwise
(i) Complex writing pattern
Size
(cm)
Mean SD Mean SD
Normalized
jerk
Mean
SD
247
262
257
332
297
131
165
122
144
177
0.62
0.67
0.68
0.86
0.50
0.36
0.36
0.24
0.36
0.30
15.1
16.9
13.7
22.9
30.9
6.7
12.6
6.4
14.6
64.0
278
321
306
417
335
128
170
167
202
158
0.72
0.74
0.74
0.88
0.48
0.24
0.24
0.22
0.34
0.27
17.8
23.1
19.6
35.6
32.6
12.3
21.1
16.5
29.4
49.4
167
FIG. 5. Normalized jerk per stroke for counterclockwise and
clockwise circles and counterclockwise and clockwise loop sequences
(i.e., cursive ‘‘llllllll’’ and ‘‘jjjjjjjj,’’ respectively) and the complex
pattern ‘‘ljielije’’ for PD subjects and elderly controls. Vertical bars
indicate the standard deviations of the means. To discriminate the
vertical bars per subject group, the curves have been slided in
horizontal direction.
respectively. Therefore, there is no systematic additional effect of left-to-right progression of the forearm.
In a separate analysis, instead of averaging across
the first 12 strokes, the first up-and-down strokes in
‘‘llllllll’’ and in ‘‘ljielije’’ (i.e., ‘‘l’’) were compared. Similarly, the second down-and-up strokes in ‘‘jjjjjjjj’’ and
‘‘ljielije’’ (i.e., ‘‘j’’) were compared. The first ‘‘l’’ showed
greater normalized jerk in the complex pattern than in
the repetitive pattern both in the elderly subjects, sign
test(11) 5 1, P , 0.05, and in the PD subjects, sign
test(15) 5 3, P , 0.05, but the first ‘‘j’’ did not show any
difference. Therefore, the PD subjects showed no differences with increasing task complexity. Detailed analysis showed that serial stroke position (i.e., 1 to 12) in
the repetitive patterns did not affect normalized jerk
nor size or duration per stroke. Furthermore, pattern
length (i.e., 4, 8, or 16 strokes) had no effect upon the
duration, size, or normalized jerk of the first 2 strokes
in either PD or elderly group.
Back-and-forth movements were also simulated using a biologically plausible model of movement production in normal and PD subjects (see Simulations; 7). The
movements produced by the simulations have been analyzed analogously to the experimental data. The results are
shown in Fig. 4b. As observed in the experimental data, the
PD simulation showed markedly elevated normalized jerk
in oblique strokes requiring coordination between wrist
and fingers. However, the asymmetry between wrist flexion and extension observed in the PD subjects but not
in the elderly was not replicated in the simulations as
the model is concerned with differences between extension and flexion and therefore the outputs were perfectly symmetric for flexion and extension.
168
TEULINGS ET AL.
DISCUSSION
This research suggests that some of the fine motor
control problems in PD patients are caused by a
reduced capability to coordinate the fingers and wrist
and by reduced control of wrist flexion. The control
problems occur in handwriting strokes that require the
coordination of wrist and fingers (i.e., strokes in oblique
orientations) compared to strokes executed by the wrist
or the fingers alone (i.e., strokes in left–right or up–
down orientations). Coordination difficulties were expressed in terms of normalized jerk, which is a measure
of the rate of change in acceleration. The integrated
squared jerk increases with excessive accelerations and
decelerations as occurring in suboptimal coordination.
Integrated jerk was normalized to allow comparison
between strokes of various durations and sizes.
Superimposed on the stroke direction effect, PD
subjects show greater normalized jerk in wrist flexions
plus ulnar deviation than in wrist extensions plus
radial deviation, whereas elderly control subjects show
no difference. The difference between extension and
flexion may be due to increased tonic activation of flexor
muscles in PD patients (28).
Marginal evidence supported the notion that integrating forearm movements with wrist and finger movements causes increase of normalized jerk. Forearm
movements are required in the left-to-right progression
of series of loops. This conclusion was based upon the
observation that normalized jerk of nonprogressing
circles and progressing loops showed only effects for
clockwise circles and loops. Loops do require additional
coordination skills compared to circles. This was shown
by Mai and Marquardt (24) who observed that specific
brain-damaged patients perform pairs of circles well,
whereas pairs of loops were substantially impaired.
The present results indicated no change of coordination during repetitive patterns. A change of normalized
jerk would have been expected based on Martin et al.
(27). They investigated 10-cm-large zigzags on smooth
sheets with and without targets visible and found a
small progressive slowing and variability increase in
the PD subjects. They observed an increase of the serial
position feedback in conditions where the targets are
not visible, suggesting that the PD subjects are forced
to rely more on position feedback, which is impaired.
Stelmach and Castiello (36) studied PD and control
subjects performing sequences of cursive ‘‘e’’ and ‘‘l’’ and
observed that both stroke size and stroke duration
progressively decreased. Stelmach et al. (38) found that
PD subjects showed progressively increasing errors. In
contrast, reports exist that PD subjects may also speed
up repetitive movements. Finger tapping seems to
cause movement hastening in PD subjects (29, 30). A
mixture of opposing effects may have canceled a serial
position effect. The number of strokes of a pattern (i.e.,
4, 8, or 16) and the context (e.g., ‘‘l’’ in a repetitive
versus nonrepetitive context) had little influence upon
normalized jerk. Possibly, both PD and elderly subjects
preprogrammed only 1 or 2 strokes ahead independently of the number of strokes to follow or the complexity of the strokes.
The handwriting patterns generated by the simulations show effects qualitatively similar to the observed
experimental data. The neural network used only
information about the beginning and target positions of
each stroke and its speed so that the simulations were
not expected to show any differences between wrist
flexion and extension. Although the amount of noise
added to the target positions was the same for PD and
normal movements, the simulations replicated the
experimental finding that normalized jerk (i.e., motor
output variability) was larger in PD than in the normal
case. In the present simulations, each degree of freedom (i.e., finger, wrist, and arm) was controlled by
separate basal ganglia–thalamocortical circuits (1, 7)
(see Fig. 1). This is consistent with Hoover and Strick
(18) who found that separate regions of the internal
segment of the globus pallidus (GPi) project to distinct
cortical motor areas, namely, the primary motor cortex,
the supplementary motor area, and the ventral premotor area. The model presented here assumes an even
greater level of segregation: parallel channels from the
GPi to the motor cortex controlling individual degrees
of freedom. This assumption is corroborated by singlecell studies of the primate motor cortex and putamen,
showing that neurons in these areas are related to
active rather than passive movements in one direction
for a single joint movement (12). This suggests that the
basal ganglia may also be involved in joint coordination. In the present simulations, small levels of dopamine depletion caused slightly smaller and shorter
than normal pallidothalamic signals which in turn
modulate the dynamics of a central movement generator (the VITE model). Due to the altered neurotransmitter dynamics in the PD model, the basal ganglia output
signals show greater than normal variability causing
an increased jerk in strokes requiring multiple degrees
of freedom (7).
In the design of the handwriting tasks and in our
handwriting model, we assumed that fingers, wrist,
and arm generate specific components of the writing
movements. The fingers produce the vertical movement
as in up–down strokes, toward and away from the body
in the horizontal plane. The wrist produces the local
horizontal movement as in left–right strokes. The
forearm produces the left-to-right horizontal progression as in extended horizontal lines of writing (40).
Some handwriting theorists have proposed that the
left-to-right progression in loops occurs steadily during
169
COORDINATION IN PARKINSONIAN HANDWRITING
a circular motion (e.g., 17, 35), whereas others have
suggested a discontinuous progression that takes place
mainly during the connecting stroke (e.g., 4). Nevertheless, it is safe to suppose that adding more components
to the handwriting movement should increase the
complexity.
In several instances the present experiment showed
no significant effects where the previous literature
would have predicted effects. The absence of effects
may be a result of the design of the experiment which
was geared toward eliminating potential artifacts. Three
of those potential artifacts are discussed here. (a) The
subjects were not artificially limited in their degrees of
freedom because this would be confounded with the
experimental patterns having different numbers of the
degrees of freedom. The disadvantage is that no certainty exists that the subjects used fingers, wrist, and
arm exactly as assumed in the model. However, nearly
all subjects spontaneously positioned their forearm
approximately vertically relative to the baseline of the
digitizer so that wrist and finger movements play roles
as assumed. This assumption is supported by the large
consistency among subjects in terms of larger normalized jerk in oblique directions. (b) In the present
experiment no additional instructions or guidelines
were presented because they would require extra attention to visual feedback resulting in additional on-line
corrections. This could affect the PD and the elderly
subjects differently because visual feedback is thought
to play a greater role in PD subjects (16), because PD
patients have greater uncertainty of limb position (10)
than the elderly, and because PD patients have more
difficulty initiating and executing nonrepetitive patterns (37). (c) Replications of trials were performed in a
blocked fashion. This minimizes task switching which
is a known problem specifically in PD subjects (11).
Movement proficiency measures other than normalized jerk exist but they have several disadvantages. For
example, the number of acceleration peaks per velocity
peak (23) or the number of acceleration zero crossings
(14, 31) can only assume integer values and do not
exploit acceleration amplitude. Some other coordination measures do not yield a simple score but rather a
pictorial characteristic, such as velocity versus acceleration diagrams (24). Timing and amplitude differences
between the coordinated components yield nonstraight
trajectories so that straightness error could also be
used as a measure for coordination problems. Isenberg
and Conrad (20) found that PD patients initiate arm
movement components nonsimultaneously, resulting in
curved and angular movement trajectories. Duration
per stroke may be used as a measure of movement
difficulty. Duration per stroke is correlated with normalized jerk because a slow movement is likely to have
multiple acceleration peaks which result in a greater
normalized jerk. However, in contrast to jerk effects, no
significant stroke duration effects were observed.
In summary, our data indicate that PD patients,
when compared to the elderly controls, show reduced
capability to coordinate the wrist and fingers in handwriting-like tasks. In addition, wrist flexion showed
more irregular acceleration profiles than wrist extension. Therefore, we conclude that coordination impairments in PD patients can be detected in finger and
wrist movements and even in flexions and ulnar deviations of the wrist which may contribute to the handwriting impairments observed in PD patients.
APPENDIX
The sequences of relative target position vectors
(TPVx, TPVy, TPVa) for the simulations of the straight
back-and-forth strokes were:
(a) horizontal (0.47, 0.003, 0) (0, 0, 0) (20.47, 0.003, 0)
(0, 0, 0),
(b) oblique slanted forward (0.33, 0.33, 0) (0, 0, 0)
(20.33, 0.33, 0) (0, 0, 0),
(c) vertical (0.003, 0.47, 0) (0, 0, 0) (20.003, 20.47, 0)
(0, 0, 0), and
(d) oblique slanted backward (20.33, 0.33, 0) (0, 0, 0)
(0.33,20.33, 0) (0, 0, 0).
ACKNOWLEDGEMENTS
This research was supported by NIH R01 NS 33173-01. The
authors thank the Parkinson patients and the elderly subjects from
Arizona, The Barrow’s Neurological Institute, and the Mayo Clinic
Scottsdale for their willingness to participate and to cooperate.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
Alexander, G. E., M. R. DeLong, and P. L. Strick. 1986. Parallel
organization of functionally segregated circuits linking basal
ganglia and cortex. Annu. Rev. Neurosci. 9: 357–381.
Benecke, R., J. C. Rothwell, J. P. R. Dick, B. L. Day, and J. C.
Marsden. 1986. Performance of simultaneous movements in
patients with Parkinson’s disease. Brain 109: 739–757.
Boisseau, M., G. Chamberland, and S. Gauthier. 1986. Handwriting analysis of several extrapyramidal disorders. [Presented at
the meeting of the CSFS, Niagara Falls, Canada, September
1986]
Bullock, D., S. Grossberg, and C. Mannes. 1993. A neural
network model for cursive script production. Biol. Cybernet. 70:
15–28.
Cassell, K., K. Shaw, and G. Stern. 1973. A computerized
tracking technique for the assessment of parkinsonian motor
disabilities. Brain 96: 815–826.
Castiello, U., K. M. B. Bennett, and M. Scarpa. 1994. The reach
to grasp movement of Parkinson’s disease subjects. Pages
215–237 in K. M. B. Bennett and U. Castiello, Eds., Insights into
the Reach to Grasp Movement. North-Holland, Amsterdam.
Contreras-Vidal, J. L., and G. E. Stelmach. 1995. A neural
network model of basal ganglia–thalamocortical relations in
170
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
TEULINGS ET AL.
normal and parkinsonian movement. Biol. Cybernet. 73: 467–
476.
Contreras-Vidal, J. L., H. L. Teulings, and G. E. Stelmach. 1995.
A network model of pathological cortical-subcortical interactions in abnormal handwriting. Pages 38–39 in the Proceedings
of the 7th International Graphonomics Society Conference. London, Ontario. [August 1995, ISBN 0-921121-14-8]
Contreras-Vidal, J. L., H. L. Teulings, and G. E. Stelmach. 1995.
Micrographia in Parkinson’s disease. NeuroReport 6: 2089–
2092.
Cooke, J. D., J. D. Brown, and V. B. Brooks. 1978. Increased
dependence on visual information for movement control in
patients with Parkinson’s disease. Can. J. Neurol. Sci. 5:
413–415.
Cools, A. R., R. Jaspers, M. Schwarz, K. H. Sonntag, M.
Vrijmoed-de Vries, and J. Van den Bercken. 1984. Basal ganglia
and switching motor programs. Pages 513–543 in J. S. McKenzie, R. E. Kemm, and L. N. Wilcock, Eds., The Basal Ganglia:
Structure and Function. Plenum, New York.
Crutcher, M. D., and M. R. DeLong. 1984. Single cell studies in
the primate putamen. Exp. Brain Res. 53: 233–243.
Daniel, W. W. 1978. Applied Nonparametric Statistics. Houghton Mifflin, Boston.
Eichhorn, T., G. Arnold, N. Mai, T. Gasser, and W. H. Oertel.
1994. Disturbance in the execution of automated handwriting
movements: A sensitive marker for early Parkinsonism. Mov.
Disord. 9: 121.
Flash, T., R. Inzelberg, and A. D. Korsczyn. 1992. Quantitative
methods for the assessment of motor performance in Parkinson’s disease. Pages 87–106 in F. Clifford Rose, Ed., Parkinson’s
Disease and the Problems of Clinical Trials. Smith-Gordon,
London.
Flash, T., R. Inzelberg, E. Schechtman, and A. D. Korczyn. 1992.
Kinematic analysis of upper limb trajectories in Parkinson’s
disease. Exp. Neurol. 118: 215–226.
Hollerbach, J. M. 1981. An oscillation theory of handwriting.
Biol. Cybernet. 39: 139–156.
Hoover, J. E., and P. L. Strick. 1993. Multiple output channels in
the basal ganglia. Science 259: 819–821.
Hogan, N., and T. Flash. 1987. Moving gracefully: Quantitative
theories of motor coordination. Trends Neurosci. 10: 170–174.
Isenberg, C., and B. Conrad. 1994. Kinematic properties in slow
arm movements in Parkinson’s disease. J. Neurol. 241: 323–
330.
Kitazawa, S., T. Goto, and Y. Urushihara. 1993. Quantitative
evaluation of reaching movements in cats with and without
cerebellar lesions using normalized integral of jerk. Pages 11–19
in N. Mano, I. Hamada, and M. R. DeLong, Eds., Role of the
Cerebellum and Basal Ganglia in Voluntary Movement. Elsevier,
Amsterdam.
Lazarus, A., and G. E. Stelmach. 1992. Interlimb coordination in
Parkinson’s disease. Mov. Disord. 7: 159–170.
Maarse, F. J., R. G. J. Meulenbroek, H. L. Teulings, and
A. J. W. M. Thomassen. 1987. Computational measures for
ballistic handwriting. Pages 16–18 in R. Plamondon, C. Y. Suen,
J. G. Deschenes, and G. Poulin, Eds., Proceedings of the 3rd
International Symposium on Handwriting and Computer Applications. Ecole Polytechnique, Montreal. [ISBN 2-553-00197-5]
Mai, N., and Ch. Marquardt. 1993. Analysis of handwriting
movements in brain damaged patients. Pages 240–242 in 6th
International Conference on Handwriting and Drawing. Telecom, Paris.
Marsden, C. D. 1984. Which motor disorders in Parkinson’s
disease indicate the true motor function of the basal ganglia?
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
Functions of the basal ganglia. Pages 225–241 in Ciba Foundation Symposium, Vol. 107. Pitman, London.
Marsden, C. D., and J. A. Obeso. 1994. The function of the basal
ganglia and the paradox of stereotaxic surgery in Parkinson’s
disease. Brain 117: 877–897.
Martin, K. E., J. G. Phillips, R. Iansek, and J. L. Bradshaw.
1994. Inaccuracy and instability of sequential movements in
Parkinson’s disease. Exp. Brain Res. 102: 131–140.
Mink, J. W., and W. T. Thach. 1991. Basal ganglia motor control.
III. Pallidal ablation: Normal reaction time, muscle concentration, and slow movement. J. Neurophysiol. 65: 330–351.
Nagasaki, H., and R. Nakamura. 1982. Rhythm formation and
its disturbance: A study based upon periodic response of the
motor output system. J. Hum. Ergol. 11: 27–142.
Nakamura, R., H. Nagasaki, and H. Narabayashi. 1978. Disturbances of rhythm formation in patients with Parkinson’s disease. Part I. Characteristics of tapping response to the periodic
signals. Percept. Mot. Skills 46: 63–75.
Phillips, J. G., J. L. Bradshaw, E. Chiu, and J. A. Bradshaw.
1994. Characteristics of handwriting of patients with Huntington’s disease. Mov. Disord. 9: 521–530.
Riederer, P., E. Sofic, C. Konradi, J. Kornhuber, H. Beckmann,
M. Dieti, G. Moll, and G. Hebnestrait. 1989. The role of
dopamine in the control of neurobiological functions. Pages 1–17
in P. Riederer et al., Eds., The Role of Brain Dopamine. Springer,
Berlin.
Schneider, K., and R. F. Zernicke. 1989. Jerk-cost modulations
during the practice of rapid arm movements. Biol. Cybernet. 60:
221–230.
Schomaker, L. R. B., A. J. W. M. Thomassen, and H. L. Teulings.
1989. A computational model of cursive handwriting. Pages
153–177 in R. Plamondon, C. Y. Suen, and M. Simner, Eds.,
Computer Recognition and Human Production of Handwriting.
World Scientific, Singapore.
Singer, Y., and N. Tishby. 1994. Dynamical encoding of cursive
handwriting. Biol. Cybernet. 71: 227–237.
Stelmach, G. E., and U. Castiello. 1992. Functional force control
in Parkinson’s disease patients. Pages 401–423 in G. E. Stelmach and J. Requin, Eds., Tutorials in Motor Behavior II.
North-Holland, Amsterdam.
Stelmach, G. E., A. Garcia-Colera, and Z. E. Martin. 1989. Force
transition control within a movement sequence in Parkinson’s
disease. J. Neurol. 236: 406–410.
Stelmach, G. E., C. J. Worringham, and E. A. Strand. 1987. The
programming and execution of movement sequences in Parkinson’s disease. Int. J. Neurosci. 36: 55–65.
Teulings, H. L., and F. J. Maarse. 1984. Digital recording and
processing of handwriting movements. Hum. Mov. Sci. 3: 193–
217.
Teulings, H. L., A. J. W. M. Thomassen, and F. J. Maarse. 1989. A
description of handwriting in terms of main axes. Pages 193–
211 in R. Plamondon, C. Y. Suen, and M. Simner, Eds., Computer
Recognition and Human Production of Handwriting. World
Scientific, Singapore.
Teulings, H. L., and G. E. Stelmach. 1991. Control of stroke size,
peak acceleration, and stroke duration in parkinsonian handwriting. Hum. Mov. Sci. 10: 315–333.
Wierzbicka, M. M., A. W. Wiegner, E. L. Logigian, and R. R.
Young. 1991. Abnormal most-rapid contraction in patients with
Parkinson’s disease. J. Neurol. Neurosurg. Psychiatry 54: 210–
216.
Wing, A. M. 1978. Response timing in handwriting. Pages
153–172 in G. E. Stelmach, Ed., Information Processing in
Motor Control and Learning. Academic Press, New York.