Brain correlates of fast and slow handwriting in humans:

a PET±performance correlation analysis

Hartwig R. Siebner,

1,2

Claus Limmer,

1

Alexander Peinemann,

1

Peter Bartenstein,

3

Alexander Drzezga

3

and

Bastian Conrad

1

1

Department of Neurology, Technische UniversitaÈt MuÈnchen, Moehlstrasse 28, D-81675 Munich, Germany

2

Sobell Department of Neurophysiology, Institute of Neurology, Queen Square, London WC1N 3BG, UK

3

Department of Nuclear Medicine, Technische UniversitaÈt MuÈnchen, Ismaninger Strasse 22, D-81675 Munich, Germany

Keywords: brain mapping, handwriting, humans, kinematics, velocity

Abstract

The present study examined the cerebral control of velocity during handwriting. We employed H

2

15

O positron emission

tomography (PET) to measure the regional cerebral blood ¯ow (rCBF) in 10 healthy subjects. Participants were required to write

the German verb `bellen' (`to bark') either at their normal speed (i.e. fast open-loop handwriting) or to write at approximately half

of their normal speed without visual feedback. The second task required a continuous modi®cation of the motor output according

to the kinaesthetic feedback from the hand (i.e. slow closed-loop handwriting). Pencil movements were recorded during PET

scanning and analysed off-line using a stroke-based analysing program. The mean number of inversions in velocity (NIV) per

stroke was used to quantify the mode of motor control during each PET scan. A NIV of 1 indicates fast open-loop processing,

whereas an increase in NIV re¯ects a shift towards slow closed-loop processing of handwriting. Foci in the left primary

sensorimotor cortex, the right lateral premotor cortex, the left anterior parietal cortex, the left anterior putamen, the left rostral

supplementary motor area and the right precuneus showed a graded increase in functional activation with the mean NIV per

stroke, suggesting that this set of brain regions is particularly involved in the processing of slow closed-loop writing movements.

No area showed a negative relationship between rCBF and the mean NIV per stroke, suggesting that fast open-loop handwriting

is achieved by an optimized cooperation of the manual sensorimotor network rather than by a selective activation of a distinct

network component.

Introduction

Handwriting is a highly over-learned ®ne manual skill. Under normal

conditions, central processing of writing movements operates in an

open-loop mode, which employs mainly feed-forward mechanisms to

adjust the kinematics of writing movements (Freund, 1986;

Plamodon, 1995). Sensory feedback is used only to monitor the

approximate range of the movement (i.e. `whole ®eld control') rather

than to scan a distinct detail during a single movement (Freund,

1986). However, when the subject is required to attentively track a

distinct variable such as writing velocity or letter size, writing

movements become critically dependent on sensory feedback, and

sensorimotor processing shifts from an open-loop mode towards a

closed-loop mode of handwriting (Marquardt et al., 1999). In contrast

to the single-peaked, bell-shaped velocity pro®les of fast automatic

writing movements, controlled closed-loop writing movements show

¯attened velocity pro®les with an increased number of inversions in

velocity (NIV) per single stroke and a reduction in mean writing

velocity (Fig. 1; Eichhorn et al., 1996; Marquardt et al., 1999). The

®ner the adjustment required, the slower the movement and the higher

the NIV per stroke. This inherent reciprocal relation between

movement velocity and the mode of motor control (i.e. mode of

sensorimotor processing) implies that changes in the speed of

movements do not re¯ect a simple scaling of a kinematic variable,

but are invariably associated with a change in the mode of motor

control (Freund, 1986; Kunesch et al., 1989).

Switching from open-loop to closed-loop processing has several

implications for motor control. First, the motor output changes

dramatically, as indicated by different velocity pro®les (Fig. 1).

Second, closed-loop handwriting necessitates a continuous integra-

tion of sensory input into appropriate motor output, whereas open-

loop movements are run off automatically. Third, subjects need to

pay more attention to handwriting. Thus, similar to early stages of

motor learning, brain areas involved in the generation of the motor

output, in sensorimotor integration, and in attention are likely to

increase their functional activity when closed-loop processing is

required.

The mechanisms involved in open-loop and closed-loop processing

of skilled hand movements need to be considered when functional

brain mapping is employed to study patients with manual motor

de®cits. In patients, automatic open-loop performance of manual

skills is commonly impaired and hand movements are slowed down

(Phillips et al., 1994; Eichhorn et al., 1996; Siebner et al., 1999a). A

disease-related shift from fast open-loop processing to slow closed-

loop processing is likely to have a considerable impact on the

movement-related cerebral activation pattern in patients with manual

motor de®cits. Therefore, a closer understanding of the mechanisms

being involved in fast open-loop and slow controlled-loop processing

of skilled hand movements in healthy subjects will help the

Correspondence: Dr Hartwig Roman Siebner,

2

Sobell Department of

Neurophysiology, as above.

E-mail: [e-mail]
Received 19 January 2001, revised 20 June 2001, accepted 21 June 2001

European Journal of Neuroscience, Vol. 14, pp. 726±736, 2001

ã Federation of European Neuroscience Societies

understanding of disease-related cerebral reorganization patterns in

patients (Weiller et al., 1992; Jahanshahi et al., 1995; Bartenstein

et al., 1997; Samuel et al., 1997; Catalan et al., 1999). By combining

H

2

15

O positron emission tomography (PET) and kinematic analysis

of handwriting movements, the purpose of the present study was to

identify those brain areas that are particularly involved in either fast

open-loop processing or slow closed-loop processing of handwriting

movements in healthy subjects.

Methods

Subjects
Ten healthy right-handers (two females) between 25 and 56 years of

age (mean age, 41.3 6 10.9 years) participated in this study.

Handedness was assessed by the Edinburgh Handedness Inventory

(Old®eld, 1971). Written informed consent for the procedure was

given by all subjects prior to the PET study. Permission to administer

radioactive substances was obtained from the German radiation

protection authorities and the study had the approval of the Ethics

Committee of the Faculty of Medicine of the Technische UniversitaÈt

MuÈnchen.

Experimental design
Each participant underwent nine sequential H

2

15

O-PET measure-

ments within a single session. All subjects were scanned in supine

position in a dimly lit room. Three experimental conditions were

investigated: two writing conditions (i.e. condition `A' and condition

`B') and a baseline condition (i.e. condition `C'). For each

experimental condition, three PET scans were performed. The

experimental conditions were repeated in an alternating order (A-B-

C-A-B-C-A-B-C), which was counterbalanced across subjects. In the

baseline condition, subjects held the pencil on a writing tablet without

writing. In the writing conditions, participants were required to

repeatedly write the German verb `bellen' (i.e. `to bark') without

visual feedback (Fig. 2). This rather simple word was selected to

minimize the cognitive effort during the writing task and to facilitate

¯uent handwriting. In the ®rst writing task, the subjects were asked to

write at their own size and speed (i.e. fast open-loop handwriting,

condition A). In the second writing task, subjects were instructed to

write at approximately half of their normal speed (i.e. slow closed-

loop handwriting, condition B). In both conditions, participants were

instructed to ensure a constant writing velocity. Each subject was

asked to reposition their hand to the starting point after having written

the word. In order to keep the number of writing movements constant

across handwriting conditions, writing was paced by a tone every 6 s.

The background noise and the pacing tone were kept constant

throughout the experimental conditions.

Prior to PET scanning, subjects were trained to perform both

writing tasks in a lying position without visual feedback. Training

was carried out in the PET scanner and subjects wrote the same word

(i.e. the word `bellen') as they actually had to produce during PET

scanning. Practice continued until kinematic analysis of the writing

movements indicated that subjects correctly performed the tasks at a

constant level. Stable motor performance was achieved after about

10±20 min of practice.

Assessment of handwriting
During PET scanning, handwriting was recorded continuously using a

digitizing graphics tablet (UD-1212, Wacom Europe GmbH, Neuss,

Germany). The writing board was placed over the participant's legs

with the surface of the board angled at 45 degrees to the horizontal

plane. Care was taken that subjects could reach the digitizing tablet

without any effort. The ®rst pacing tone was given when starting the

injection of the radioisotope and participants continued to write until

the end of each 50-s PET scan. For each PET scan, only the last eight

words, which were written during the 50-s period of data acquisition,

were used for kinematic analysis. In order to minimize movements of

proximal joints and to match the posture of the hand as close as

possible to the normal position during handwriting, the right upper

limb was comfortably supported by foam plastic pads and partici-

pants were required to place the ulnar part of their right hand on the

writing board while they were writing.

Pen-tip position of an inking digitizing pen were sampled at a

frequency of 166 Hz with a spatial resolution of 0.05 mm and a

spatial accuracy of 0.025 mm. The CS software (MedCom, Munich,

Germany) was used for data collection and off-line analysis of the

writing movements. The curves of the vertical position, and vertical

velocity of the tip of the digitizing pen were calculated and smoothed

by nonparametric regression methods (Marquardt & Mai, 1994). The

CS software has been shown previously to be a suitable tool to assess

changes in kinematics during handwriting in patients with movement

disorders (Eichhorn et al., 1996; Siebner et al., 1999a, b). Kinematic

analyses focused on movement along the vertical axis (i.e. y-axis), as

this is the axis along which writing movements predominantly occur

(Fig. 2). Thus, for quantitative analysis, the writing movements were

segmented in subsequent vertical up- or down-strokes of the pencil,

which can be considered to be fundamental units of handwriting

movements (Hollerbach, 1981; Morasso & Mussa Ivaldi, 1982;

Plamodon, 1995). A single stroke is de®ned by the time segment

F

IG

. 1. Schematic illustration contrasting the characteristics of fast open-

loop movements with those of slow closed-loop movements. Note that a

change in velocity is not just a different scaling of a single kinematic

variable but involves a shift in the mode of motor control and a change in

the magnitude of sensorimotor guidance.

Central processing of writing velocity 727

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

between two subsequent changes in vertical direction of handwriting

(Fig. 2). Only up- and down-strokes that exceeded a duration of

50 ms and an amplitude of 1 mm were included in the kinematic

analysis. The following dimensions of writing performance were

calculated: mean vertical stroke length, mean vertical stroke duration,

mean vertical writing velocity and mean vertical writing pressure.

Furthermore, the mean NIV per single stroke was estimated to

quantify the degree of automation of the handwriting movements

(Marquardt & Mai, 1994). A NIV of 1 per stroke is characteristic of

fast open-loop performance, whereas an increase in mean NIV per

stroke indicates continuous adjustments of writing velocity to the

incoming feedback information during slow closed-loop handwriting

(Eichhorn et al., 1996; Marquardt et al., 1999).

PET scanning
Regional changes in neuronal activity were indexed by monitoring

relative changes in normalized regional cerebral blood ¯ow (rCBF).

Normalized rCBF was measured by recording the regional distribu-

tion of radioactivity following the intravenous injection of

15

O-

labelled water (Fox & Mintun, 1989). PET measurements were

conducted using a Siemens ECAT 951 R/31 PET scanner (CTI,

Knoxville, TN, USA) in three-dimensional mode. After corrections

for randoms, dead time and scatter, all emission data were

reconstructed by ®ltered back projection (Hanning ®lter, 0.5 cycles

per pixel cut-off frequency) to 31 consecutive axial planes with an

interplane separation of 3.375 mm. Reconstructed slices were

displayed in a matrix consisting of 128 3 128 voxels. The PET

scanner had a total axial view of 10.5 cm and no interplane dead

space, ensuring coverage of the upper two-thirds of the brain from the

vertex to the upper cerebellum. Because the ®eld of view of the PET

scanner did not suf®ciently cover the cerebellum, the present study

provides no information on the speci®c involvement of the

cerebellum in the performance of fast open-loop or slow closed-

loop handwriting movements.

For each measurement of rCBF, 250 mBq of H

2

15

O were

administered in the left cubital vein as a semibolus injection using

an infusion pump. A 50-s PET scan was initiated with the appearance

of the tracer bolus in the brain, approximately 30 s after the start of

the infusion (Fox & Mintun, 1989). This procedure was repeated for

each PET scan with 10 min between scans to allow for adequate

decay of radioactivity. A 20-min headholder transmission scan with a

rotating

68

Ge/

68

Ga source was obtained prior to the ®rst relative rCBF

measurement in order to correct the relative rCBF measurements for

effects of radiation attenuation.

PET image reconstruction
All calculations and image transformations were performed on Sun

SPARC 2 workstations (Sun Computers Europe, Surrey, UK). PET

data were analysed using statistical parametric mapping software

(SPM 96b, Wellcome Department of Cognitive Neurology,

University College of London, UK) implemented in the PRO

Matlab environment (Mathworks, Natic, MA, USA). The scans

F

IG

. 2. Recording of writing movements in a single subject during fast open-loop handwriting (left panel) and slow closed-loop handwriting (right panel).

The top row of each panel gives the handwriting movements in space when writing the German word `bellen' (the verb `to bark'). The middle row illustrates

the vertical writing movements (i.e. strokes, with vertical position is shown as a function of time), while the bottom row demonstrates the vertical velocity

pro®les. Note the different time-scales for the left and right panels. Automatic open-loop handwriting shows smooth single-peaked velocity pro®les, whereas

velocity pro®les during slow closed-loop handwriting are characterized by many irregularities (i.e. inversions in velocity), indicating corrective adjustments of

vertical writing velocity.

728 H. R. Siebner et al.

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

from each subject were realigned using the ®rst scan as a reference.

The six parameters of this rigid body transformation were estimated

using a least squares approach on a voxel-by-voxel basis (Friston

et al., 1995a). Following realignment, PET images were transformed

into standard stereotaxic space (Talairach & Tournoux, 1988). Spatial

normalization was performed using linear and nonlinear three-

dimensional transformations to match each scan to a reference image,

which already conformed to the standard stereotaxic space (Friston

et al., 1995a). As a ®nal preprocessing step, the normalized images

were smoothed using an isotropic Gaussian kernel of 12 mm full

width at half maximum (FWHM) for all directions to increase the

signal-to-noise ratio and reduce variance due to interindividual

differences in gyral anatomy (Friston et al., 1995a). Each voxel of the

resulting normalized and smoothed images was 2 3 2 3 4 mm in

size.

Data analysis
Statistical analysis of the kinematic data was performed with SPSS

Version 9 (SPSS Inc., Chicago, IL, USA). Analysis of variance

(

ANOVA

) for repeated measurements with `time' and `writing

condition' as within-subject factors was computed for each kinematic

variable to assess the effect of time and writing condition on motor

performance. Signi®cance was accepted at a P-value of 0.05.

Statistical analysis of the PET data was performed using statistical

parametric mapping software (SPM, Version 96b). To remove the

effect of variations in global cerebral blood ¯ow across subjects and

scans, an analysis of covariance (

ANCOVA

) was applied with global

cerebral blood ¯ow as the confounding variable (Friston et al., 1990).

Global blood ¯ow was normalized by scaling across the entire data

set to a global mean of 50 mL/100 mL/min. The adjusted voxel

values were then used for further statistical analysis.

In a ®rst set of analyses, a categorical comparison was performed

between the two writing conditions and the baseline condition in

order to delineate the brain areas involved in handwriting per se.

Using linear weighted contrasts, the main effects of both writing

conditions were estimated according to the general linear model and

the theory of Gaussian ®elds at each and every voxel (Friston et al.,

1991, 1995b; Worsley et al., 1992). The generated statistical

parametric [t] maps were subsequently transformed into normally

distributed statistical parametric [Z] maps (Friston et al., 1995b).

Signi®cance level was set at a P-value of 0.05 after correction for

multiple nonindependent comparisons; this corresponds to a Z-score

of 4.26. Brain areas showing increases in rCBF at an uncorrected

P-value of less than 0.001 (corresponding to a Z-score of 3.09), but

not surviving correction for multiple nonindependent comparisons,

were considered as trend activations. Given the aims of the study, the

subsequent analyses used these categorical comparisons as a mask

and restricted further interrogation of the data to brain areas showing

at least a trend activation during automatic or controlled handwriting

in comparison with the baseline condition.

In a second set of analyses, we directly compared the activation

patterns during fast open-loop and slow closed-loop handwriting. For

this purpose, baseline scans were not included. A categorical

comparison between both writing conditions was performed in

order to assess brain areas that were particularly activated during

either fast open-loop handwriting or slow closed-loop handwriting. A

categorical comparison is useful to demonstrate brain areas that

change their net activity in a stepwise fashion between two

experimental conditions. However, a categorical approach may fail

to detect those brain areas that gradually change their functional

activity according to the requirements of a given task. By contrast,

correlational analysis between regional functional activation and a

distinct variable of motor performance, such as velocity or com-

plexity, has been shown to provide a useful means to detect graded

changes in distinct brain areas related to certain aspects of motor

control (Boecker et al., 1998; Turner et al., 1998; van Mier et al.,

1998). Therefore, in addition to a categorical comparison between

fast and slow handwriting, an independent correlational analysis

between the NIV per stroke and the normalized rCBF was performed.

Because the mean NIV per stroke re¯ects the number of corrective

adjustments in velocity during a single up- or down-stroke, this

kinematic measure is thought to re¯ect directly the degree of closed-

loop processing during handwriting (Eichhorn et al., 1996; Marquardt

et al., 1999; Siebner et al., 1999a). Because data analysis was

restricted to brain areas that had already been identi®ed to participate

in handwriting, an uncorrected P-value of 0.001 was accepted as

threshold for signi®cance in the second set of analyses.

Results

Kinematic data
Mean total movement time for writing the target word was

2.24 6 0.64 s during fast handwriting, whereas mean total movement

time was 4.21 6 0.72 s during slow handwriting, indicating that

participants wrote about twice as long during the slow handwriting

condition. Figure 2 illustrates the recorded writing movements and

the corresponding velocity pro®les during fast open-loop handwriting

and slow closed-loop handwriting in a representative subject. Table 1

gives the averaged group values (mean 6 SD) of each kinematic

measure for both writing conditions.

ANOVA

for repeated measure-

ments revealed a highly signi®cant effect of `writing condition' on

mean stroke duration, mean vertical velocity and mean NIV per

stroke (Table 1). In accordance with the instruction, subjects wrote at

approximately half of their normal speed during slow, velocity-

controlled handwriting and velocity-controlled handwriting was

associated with a consistent increase in the mean NIV per stroke

T

ABLE

1. Group data of kinematic measurements during fast open-loop handwriting and slow closed-loop handwriting

Stroke-based kinematic measures of handwriting

Fast open-loop

handwriting

Slow closed-loop

handwriting

F

1,9

-value

P-value

Numbers of inversions in velocity (NIV) per stroke

1.38 6 0.36

3.62 6 1.06

51.5

P < 0.001

Vertical writing velocity (mm/s)

67 6 31

39 6 24

52.4

P < 0.001

Stroke duration (ms)

141 6 25

265 6 27

178.4

P < 0.001

Vertical writing pressure (N)

1.65 6 0.60

1.59 6 0.68

0.2

n.s.

Vetrical stroke length (mm)

7.1 6 3.2

8.4 6 4.7

2.9

n.s.

Data are given as mean 6 SD. The F- and P-values refer to multivariate analysis of variance for repeated measurements and describe the main effect of `writing

condition' on the dependent kinematic variable. n.s., not signi®cant (P > 0.05).

Central processing of writing velocity 729

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

(Table 1). The increase in NIV per stroke con®rmed that writing

velocity was continuously adjusted and thus, a closed-loop mode of

motor control was employed during the slow handwriting task. There

was a highly signi®cant positive correlation between the mean NIV

per stroke and the mean stroke duration (Spearman, r = 0.89,

P < 0.0001), indicating that slow writing movements were associated

strongly with a more irregular velocity pro®le.

There was no signi®cant effect of `writing condition' on mean

vertical stroke length and mean vertical writing pressure, con®rming

a selective modi®cation of writing kinematics related to velocity

without concurrent modulation of scaling the stroke length or vertical

force production. Furthermore,

ANOVA

for repeated measurements

showed no signi®cant main effect of `time' on any of the kinematic

variables of interest, indicating that there were no systematic changes

of motor performance during the entire experiment. Moreover, there

was no signi®cant interaction term between `time' and `writing

condition'.

PET data
Identi®cation of brain regions involved in handwriting per se
Categorical comparison between handwriting and the baseline

condition demonstrated that the activation patterns were very similar

during fast open-loop handwriting and slow closed-loop handwriting

(Fig. 3, Table 2). Regardless of the mode of motor control,

handwriting was associated with a signi®cant relative increase in

normalized rCBF in a large bihemispheric cortical cluster with a

prevailing activation of left hemispheric regions (Fig. 3). Within this

cluster, peak activation occurred in the hand representation of the left

primary sensorimotor cortex (SM1) and the adjacent dorsal premotor

cortex (PMD), as indicated by the highest Z-scores on categorical

comparison (Table 2). In addition, the cluster covered large areas of

the left parietal lobe and the left ventral premotor cortex (PMV) on

the left hemispheric surface (Fig. 3). Within the interhemispheric

®ssure, the cluster extended into the supplementary motor area

(SMA), the anterior cingulate cortex (ACC) and the left precuneus.

On the right hemispheric surface, the cluster included mainly the

right PMD. A second cortical cluster, which showed an activation

during both writing conditions, was located at the border between the

precuneus and the superior parietal lobule in the right parietal cortex.

Subcortically, handwriting was associated with a bilateral activation

of the thalamus.

Despite the substantial overlap in the cerebral activations during

fast open-loop handwriting and slow closed-loop handwriting, there

were some notable differences in the cerebral activation patterns. The

spatial distribution of peak activations within the large frontoparietal

cluster (indexed as `cluster 1' in Table 2) differed between condi-

tions. For instance, only fast open-loop handwriting gave rise to

F

IG

. 3. Surface rendering of colour-coded statistical parametric maps superimposed on stereotactically normalized (Talairach & Tournoux, 1988) T1-weighted

magnetic resonance images. Surface projections showing categorical increases of regional cerebral blood ¯ow (rCBF) during fast open-loop handwriting (left)

compared with holding a pencil. Surface projections showing categorical increases of rCBF during slow velocity-controlled handwriting (right) compared with

holding a pencil. Maps are thresholded at an uncorrected P-value of 0.001. Pixels with lower Z-scores are represented in red, and pixels with higher Z-scores

in yellow.

730 H. R. Siebner et al.

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

distinct peak activations in the left PMV and in the left anterior

inferior parietal cortex, whereas slow closed-loop handwriting caused

a selective peak of activation in the left lateral superior parietal

cortex. Moreover, slow closed-loop handwriting resulted in a larger

number of `activated voxels' and in a less lateralized activation

pattern with a relatively larger area of activation on the right

hemispheric surface. At a statistical threshold of P < 0.001

(uncorrected), » 10 000 voxels showed an increase in rCBF during

fast open-loop handwriting when compared with the baseline

condition, whereas » 15 000 voxels demonstrated a relative increase

in normalized rCBF during slow closed-loop handwriting (Table 2).

Only slow closed-loop handwriting caused a signi®cant relative CBF

increase in the right lateral prefrontal cortex and the right PMV, and a

trend activation in the left anterior putamen. By contrast, no

activation in the basal ganglia was observed during fast open-loop

handwriting, even at a low statistical threshold (P < 0.01, un-

corrected). Compared with slow closed-loop handwriting, no speci®c

cluster (i.e. brain region) was selectively activated in the `fast

handwriting' condition.

Fast open-loop handwriting vs. slow closed-loop handwriting
As outlined earlier, the search volume was limited to brain areas that

have shown at least a trend activation during either fast or slow

handwriting compared with holding the pencil. Please note that this

`mask' is not overly restrictive, as it incorporates the entire set of

brain areas that have previously been shown as being implicated in

the generation of skilled writing movements (Ceballos-Baumann

et al., 1997; Seitz et al., 1997; IbaÂnÄez et al., 1999). Within the regions

showing at least a trend activation during handwriting per se,

categorical comparison between fast open-loop handwriting and

slow closed-loop handwriting detected differences in rCBF in a

distinct set of brain areas (Table 3). Because fast and slow

handwriting resembles similar `activated states' of the cerebral

motor control system, the overall Z-scores were much lower than

those for the contrasts comparing each writing condition with the

baseline condition. The only area showing a signi®cant activation

during fast open-loop handwriting as compared with slow closed-loop

handwriting was a focus in the right SM1 dorsomedially to the

primary sensorimotor hand area (Table 3). Bilateral foci in the

inferior parietal lobule, the right PMD and the left anterior rostral

putamen were more active during slow closed-loop handwriting when

compared with fast open-loop handwriting (Table 3). The areas in the

left dorsolateral prefrontal cortex and the right PMV, which were

activated exclusively during slow handwriting but not during fast

handwriting (when compared with the baseline condition), did not

show signi®cant differences when performing a direct categorical

comparison between fast and slow handwriting. This was due to

subtle, nonsigni®cant rCBF increases in these areas during fast open-

loop handwriting compared with holding the pen.

Correlational analysis revealed a somewhat different regional

pattern of cerebral clusters, which demonstrated a linear relationship

between the rCBF and the degree of closed-loop performance, as

indexed by the mean NIV per stroke (Table 4). No cerebral area

within the ®eld of view of the PET scanner showed a negative linear

T

ABLE

2. Clusters of brain regions showing a signi®cant relative increase in normalized rCBF, indicating a functional activation during fast open-loop

handwriting or slow closed-loop handwriting compared with holding the pencil

Brain regions

(BA)

Voxels

per cluster

Z-score of

peak activation

Coordinates of peak activation

x

y

z

Areas showing relative increases in rCBF during automatic handwriting (condition A)

Cluster 1

9730

Left dorsal premotor cortex

(6)

7.24

±22

±10

66

Left primary sensorimotor cortex

(3/4)

7.22

±38

±24

58

Left ventral premotor cortex

(6)

5.97

±62

±12

38

Caudal supplementary motor area

(6)

5.87

±8

±8

56

Left anterior inferior parietal cortex

(2/40)

5.59

±54

±26

44

Right dorsal premotor cortex

(6)

(3.72)

30

±10

62

Cluster 2, left thalamus

±

28

(3.91)

±12

±22

4

Cluster 3, right inferior parietal cortex

(40)

93

(3.85)

38

±44

52

Cluster 4, right precuneus

±

77

(3.66)

16

±66

62

Cluster 5, right thalamus

±

30

(3.49)

14

±16

10

Areas showing relative increases in rCBF during controlled handwriting (condition B)

Cluster 1

14005

Left primary sensorimotor cortex

(3/4)

7.83

±38

±22

60

Left dorsal premotor cortex

(6)

7.55

±26

±14

62

Left lateral superior parietal cortex

(7)

6.87

±32

±52

60

Right dorsal premotor cortex

(6)

5.86

30

±12

60

Caudal supplementary motor area

(6)

5.55

0

±4

56

Cluster 2, right precuneus

(7)

440

5.40

14

±64

60

Cluster 3, right ventral premotor cortex

(6/44)

198

4.49

62

10

24

(3.67)

52

6

28

Cluster 4, right lateral prefrontal cortex

(9)

177

4.29

34

38

34

Cluster 5, left thalamus

±

29

(3.59)

±10

±20

4

Cluster 6, right thalamus

±

16

(3.57)

8

±2

22

Cluster 7, left anterior putamen

±

48

(3.55)

±24

6

16

Cluster 8, right anterior claustrum/putamen

±

43

(3.51)

32

16

2

A Z-score of 4.26 corresponds to a P-value of 0.05 after correction for multiple comparisons. Z-scores ranging from 3.09 to 4.26 were considered as trend

activations. Trend activations in this table are those in parentheses. Coordinates express the position of activation foci within the cluster relative to the anterior

commissure in the stereotactic space of Talairach & Tournoux (1988).

Central processing of writing velocity 731

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

relationship between rCBF and the mean NIV per stroke. In contrast,

several regions within the distributed network, which participates in

handwriting, showed a positive relationship between rCBF and the

mean NIV per stroke, indicating a gradual increase in functional

activation with close-loop handwriting. These areas included distinct

foci in the left SM1, the left anterior inferior parietal cortex, the right

PMD, the left rostral SMA, the right posterior superior parietal

cortex, the right anterior inferior parietal cortex, and the left anterior

putamen (Fig. 4, Table 4).

Discussion

By combining PET scanning and continuous recording of writing

movements, the present study yielded two main results. First, a

distinct set of cortical and subcortical areas demonstrated a gradual

increase in functional activation with slow handwriting, involving

closed-loop control of writing velocity. Second, no cerebral region

within the ®eld of view of the PET scanner showed a graded

activation with fast automatic handwriting movements when match-

ing the numbers of movements per PET scan between `fast open-loop

handwriting' and `slow closed-loop handwriting'.

Methodological considerations
Automatic open-loop performance is characterized by fast move-

ments with a higher movement rate, whereas controlled closed-loop

performance is restricted to the range of slower movements with a

lower movement rate (Fig. 1; Freund et al., 1986; Kunesch et al.,

1989). The close relationship between movement velocity and the

mode of sensorimotor processing implies a methodological dilemma

in terms of matching total motor output between fast open-loop

handwriting and slow closed-loop handwriting. On one hand, when

matching the number of movements between writing conditions, total

movement time will differ, because subjects need less time to carry

out fast writing movements. On the other hand, by keeping total

movement time constant between writing conditions, the total motor

output (e.g. the length of the writing trace and the number of strokes)

will differ between writing conditions.

Previous studies on velocity control of skilled hand movements

(Turner et al., 1998; van Mier et al., 1998) investigated continuous

movements, keeping total movement time constant across PET scans.

Because an increase in movement velocity is invariably associated

with an increased number of submovements (i.e. switches in the

motor programme), as well as in an increase in the total length of the

T

ABLE

3. Clusters of brain regions showing signi®cant differences in normalized rCBF during both writing conditions when applying a categorical comparison

(i.e. fast open-loop vs. slow closed-loop handwriting)

Brain regions

(BA)

Voxels

per cluster

Z-score of

peak activation

Coordinates of peak activation

x

y

z

Areas showing relative rCBF increases during automatic handwriting compared with controlled handwriting

Right primary sensorimotor cortex

(3/4)

21

3.56

22

±20

76

Areas showing relative rCBF increases during controlled handwriting compared with automatic handwriting

Left inferior parietal cortex

(40)

153

3.95

±36

±40

54

Left rostral putamen

±

55

3.80

±24

10

14

Right inferior parietal cortex

(40)

93

3.73

38

±44

36

Right lateral premotor cortex

(6)

84

3.73

28

0

48

The data are presented as in Table 2. Please note that an uncorrected P-value of 0.001 was accepted as the statistical threshold for those brain areas, as data analysis

was restricted to those brain areas, which had shown at least a trend activation during either fast or slow handwriting as compared with holding the pencil.

T

ABLE

4. Clusters of brain regions showing a signi®cant correlation between normalized rCBF and the mean numbers of inversions in velocity (NIV) per

stroke (P < 0.001, uncorrected)

Brain regions

(BA)

Voxels

per cluster

Z-score of

peak activation

Coordinates of peak activation

x

y

z

Areas showing a negative correlation between rCBF and the mean NIV per stroke during handwriting

No cluster within the ®eld of view of the scanner

Areas showing a positive correlation between rCBF and the mean NIV per stroke during handwriting

Left primary sensorimotor cortex

(3/4)

258

4.57

±56

±14

52

Right lateral premotor cortex

(6)

208

4.34

3.59

16

20

±12

±8

58

48

Left inferior parietal cortex

(5/40)

208

3.99

±44

±34

58

Left rostral putamen

±

43

3.55

±22

8

12

Left supplementary motor area

(6)

33

3.37

±12

2

56

Right precuneus

(7)

15

3.37

20

±56

50

The mean NIV per stroke was taken as a kinematic measure to quantify the degree of closed-loop performance during handwriting. A mean NIV of 1 corresponds

to automatic open-loop handwriting, whereas an increase in mean NIV per stroke indicates a shift towards closed-loop performance. The higher the mean NIV per

stroke the greater the magnitude of closed-loop processing. The data are presented as in Table 3. Please note that an uncorrected P-value of 0.001 was accepted as

the statistical threshold for correlational analysis, as data analysis was restricted to those brain areas, which had shown at least a trend activation during either fast

or slow handwriting as compared with holding the pencil.

732 H. R. Siebner et al.

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

movement trace, the reported differences in functional activation

were not exclusively related to movement velocity. In the present

study, we decided to keep the number of movements constant across

PET scans rather than matching total movement time. The rationale

for matching the number of writing movements was twofold. First, it

has been shown that differences in the number of movements per scan

have a profound effect on neuronal activation in executive motor

areas, especially in the primary sensorimotor cortex (Blinkenberg

et al., 1996; Sadato et al., 1997; Jancke et al., 1998; Kawashima et al.,

1999). Second, previous functional imaging studies on differences in

the motor activation pattern between `fast-moving' healthy controls

and `slow-moving' patients have generally matched the number of

movements per PET scan rather than total movement time (Weiller

et al., 1992; Jahanshahi et al., 1995; Bartenstein et al., 1997; Samuel

et al., 1997; Catalan et al., 1999). However, we like to emphasize that

total movement time was not matched between writing conditions, as

participants were engaged in slow closed-loop handwriting longer

than in fast open-loop handwriting.

Sensorimotor control of writing movements may involve different

sensory modalities and may focus on different aspects of handwriting,

such as velocity, stroke size or shape. The present study focused on

the kinaesthetic control of the velocity of handwriting. In order to

ensure that only kinaesthetic feedback was used to monitor writing

velocity, participants were deprived of visual feedback. In accordance

with Marquardt et al. (1999), kinematic analysis revealed that the

lack of visual feedback did not hamper open-loop processing of

handwriting movements. Moreover, the functional activation pattern

during handwriting was not in¯uenced by different levels of motor

learning, as stable task performance was con®rmed by kinematic

analysis during PET scanning.

Brain activation during handwriting per se
The cerebral activation pattern related to handwriting in comparison

with baseline condition was quite similar during fast open-loop and

slow-closed-loop performance, showing a widespread bilateral

increase in normalized rCBF in the SM1, the lateral premotor cortex,

the SMA, adjacent ACC, and anterior and posterior parts of the

parietal cortex with a left-hemispheric preponderance. Subcortically,

handwriting was associated with a consistent bilateral thalamic

activation. This cerebral activation pattern is in accordance with

previous functional imaging studies on healthy volunteers, which

have con®rmed the participation of these brain regions in the

generation of handwriting movements (Ceballos-Baumann et al.,

1997; Seitz et al., 1997; IbaÂnÄez et al., 1999). The close spatial

correspondence in writing-related cerebral activation patterns across

studies is somewhat surprising, given the fact that there are notable

differences in the writing tasks employed in these studies (Ceballos-

Baumann et al., 1997; Seitz et al., 1997; IbaÂnÄez et al., 1999). In the

study by Seitz et al. (1997), subjects were required to continuously

write letters or `nonsense' letters and visual feedback was provided,

whereas in the other studies subjects repeatedly wrote either a word

(`dog') paced by an auditory cue every 4 s (Ceballos-Baumann et al.,

1997) or a sentence (`the book is on the desk') in a self-paced manner

(IbaÂnÄez et al., 1999) without visual feedback. With the exception of

the study by Ceballos-Baumann et al. (1997), the number of writing

movements was not matched across PET scans. Moreover, the

instructions differed considerably across the studies. In the study by

Seitz et al. (1997), subjects were asked to write the letters either as

fast as possible or as exact as possible with respect to letter size,

requiring either an automatic open-loop mode of handwriting during

`fast writing' or a closed-loop mode during `exact writing'. Please

note that closed-loop performance in the study by Seitz et al. (1997)

referred to size control (i.e. spatial aspect of handwriting), whereas, in

the present study, closed-loop performance referred to velocity

control (i.e. temporal aspect of handwriting). In the study by IbaÂnÄez

et al. (1999), subjects were asked `not to write quickly' at their own

pace (presumably in an open-loop mode), whereas in the study by

Ceballos-Baumann et al. (1997), subjects were required to write

continuously and adapt their speed of handwriting to the interstimulus

interval of the cueing tone, demanding a continuous closed-loop

adjustment of writing velocity comparable with the `slow writing

condition' in the present study. With regard to the considerable

differences in the writing task within and between these studies, the

substantial overlap in the spatial activation pattern strongly suggests

that the generation of handwriting movements per se activates a

distinct set of cortical and subcortical regions.

Activation pattern associated with slow velocity-controlled

handwriting
Extending previous ®ndings by Seitz et al. (1997) on visually guided

handwriting, we found also notable differences in the cerebral

activation pattern during fast open-loop and slow closed-loop right

handwriting. During handwriting, the magnitude of closed-loop

control (as indexed by the mean NIV per stroke) was associated

with a graded increase in neuronal activity in a distinct subset of

cortical and subcortical motor control areas, including foci in the left

F

IG

. 4. Surface rendering of statistical parametric maps superimposed on

stereotactically normalized T1-weighted magnetic resonance images. The

white areas represent brain regions that showed a positive correlation

between normalized regional cerebral blood ¯ow (rCBF) and the mean

number of inversions in velocity (NIV) per stroke (see also Table 4),

indicating a weighted regional activation of these areas depending on the

amount of closed-loop processing. Maps are thresholded at an uncorrected

P-value of 0.001.

Central processing of writing velocity 733

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

SM1, the right PMD, the left anterior superior parietal lobule, the left

anterior putamen, the left rostral SMA, and the right precuneus. As

already outlined in the introduction, these brain regions may be

involved in the generation of the motor output, in sensorimotor

integration or in attentional processes, when closed-loop processing is

required.

Left primary sensorimotor cortex (SM1)
Categorical and correlational comparison revealed an increased

neuronal activity in the left SM1 during slow velocity-controlled

handwriting. At a glance, this ®nding seems to be at odds with

previous imaging studies that have shown increasing movement

velocity results in a gradual increase in activation in SM1 (Turner

et al., 1998; van Mier et al., 1998). It needs to be pointed out that

these studies investigated continuous movements (Turner et al., 1998;

van Mier et al., 1998). As already outlined earlier, increasing the

velocity of continuous movements entails an increase in movement

rate and the total amount of ankle displacement per scan. This

increase in net motor output and net sensory feedback renders it

dif®cult to ascribe gradual changes in SM1 activation exclusively to

changes in movement velocity.

Three mechanisms may have contributed to the linear relationship

between functional activation in the left SM1 and closed-loop

performance in the present study. First and most importantly, subjects

were engaged for about twice as long in task performance during

slow closed-loop handwriting, resulting in a longer neuronal

processing time. Second, an increase in attention paid to kinaesthetic

input from the writing hand during closed-loop processing might have

contributed to increased functional activation of the left SM1 (Meyer

et al., 1991). Third, because the SM1 has been shown to be

particularly active during ®nger movements guided by somatosensory

information (Geyer et al., 1996), the gradual increase in SM1 activity

might also re¯ect increasingly higher demands for online processing

of kinaesthetic feedback during slow velocity-controlled handwriting.

Frontal premotor areas
Two circumscribed areas in the frontal premotor cortex showed a

graded increase in rCBF with an increase in sensorimotor guidance,

namely the right PMD and the left rostral SMA anterior to the vertical

anterior commissural plane. These premotor areas have shown a

progressive increase in rCBF with movement complexity (i.e. the

right PMD with increasing sequence length and left rostral-SMA with

complexity of sequence pattern) when subjects performed a sequence

of auditory paced ®nger movements with their dominant right hand,

suggesting a role of these areas in the control of sequential ®nger

movements (Sadato et al., 1996; Boecker et al., 1998; Catalan et al.,

1998). In the present study, subjects wrote the same word in both

writing tasks. Hence, differences in task complexity did not account

for the increased activity in these two premotor areas during slow

handwriting. Given that slow velocity-controlled handwriting de-

manded a high degree of temporal accuracy, the increased activation

in the left pre-SMA and right PMD suggests an involvement of these

areas in precise timing of over-learned movement sequences

(Halsband et al., 1993). The increased activation of the pre-SMA

during slow handwriting is probably related to additional timing

operations during handwriting at a frequency slower than the

subject's normal pace (Kawashima et al., 1999).

It has been suggested that the caudal part of the PMD is concerned

with online correction of movement execution (see Wise et al., 1997

for review, Grafton et al., 1998). In keeping with this notion, the

activation peak in the PMD, showing a graded increase in activation

with the amount of closed-loop handwriting, was located within the

more caudal part of the PMD (x, y and z of 16, ±12 and 58).

Moreover, in accordance with studies by Sadato et al. (1996) and

Catalan et al. (1998) on motor sequence control, activation of the

PMD was observed within the right hemisphere ipsilaterally to the

moving hand. This is most likely due to a right hemispheric

dominance for spatial attention (Gitelman et al., 1996; Winstein et al.,

1997).

Parietal cortex
An area in the left anterior inferior parietal cortex Brodmann area

(BA) 5, extending into BA 40, showed a graded functional activation

with slow closed-loop handwriting. It has been reported that the

activity in this anterior part of the parietal cortex gradually increases

with the degree of exerted force during ®nger movements (Dettmers

et al., 1995) and that this area is selectively active during execution

but not during preparation or imagination of freely selected joystick

movements (Stephan et al., 1995), indicating a role of this parietal

area in processing somatosensory feedback during hand movements.

This parietal area may correspond to the parietal area PE (area 5) in

monkeys, which is thought to be a higher-order somatosensory area

mostly devoted to the analysis of proprioceptive information (for

review see Rizzolatti et al., 1998). Area PE receives proprioceptive

and cutaneous input from the limb and responds to somaesthetic

stimuli, but it does not receive visual input (Sakata et al., 1973;

Mountcastle et al., 1975). Many neurons in area PE encode

movement kinematics of upper limb movements (Kalaska et al.,

1990) and the location of the arm in space in a body-centred

coordinate system (Lacquaniti et al., 1995). Furthermore, the PE area

is richly interconnected with the primary motor hand area of the

monkey (F1). These features of area PE and the PE±F1 circuits have

led Rizzolatti et al. (1998) to suggest that the `main role of the PE±F1

circuits appears to be that of providing F1 with information on the

location of body parts necessary for the control of movement'.

Because the velocity-controlled writing task involved continuous

sensorimotor processing of kinaesthetic feedback from the moving

hand, we would like to put forward the hypothesis that the increase in

the level of activation in the left SM1 and anterior parietal cortex

during controlled writing re¯ects the activation of parietofrontal

loops involved in kinaesthetic control of skilled hand movements,

which may be the human homologues to the PE±F1 circuits in the

monkey.

A gradual increase in the level of activation with the amount of

closed-loop handwriting was also observed in the right precuneus.

Apart from the importance of the precuneus in spatial awareness

(Corbetta et al., 1993), the precuneus is thought to be an integrative

transmodal area providing a sensory representation of extrapersonal

space (for review see Mesulam, 1998). Our data suggest that the

precuneus contributes to the integration of kinaesthetic information

into a complex movement sequence presumably by storing not only

spatial representations (Seitz et al., 1997), but also kinaesthetic

representations (Sirigu et al., 1996).

Basal ganglia
Subcortically, controlled handwriting was associated with a focus of

increased activation in the left rostral putamen anterior to the vertical

anterior commissural plane. The rostral putamen is thought to belong

to the `association striatal territory', which receives projections from

various frontal, temporal, and parietal areas, whereas the `sensori-

motor' striatal territory comprises the post-commissural portion of the

putamen, receiving projections from the somatosensory, motor and

premotor areas (Kunzle, 1975; Parent & Hazrati, 1995). The anterior

parts of the basal ganglia have been shown to be active during new

734 H. R. Siebner et al.

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

learning or random generation of a ®nger sequence (Jenkins et al.,

1994; Jueptner et al., 1997), as well as during externally paced, but

not self-paced, sequences of arm movements (Menon et al., 1998).

The graded activation of the anterior putamen during slow closed-

loop handwriting demonstrates that the basal ganglia activity related

to motor performance is in¯uenced by the contextual setting and

plays a role in `optimizing the pattern of muscular activity in the light

of sensory feedback' (Brooks, 1997). Furthermore, the present ®nding

is in concordance with the view that the anterior part of the putamen

is activated during movement sequences with higher cognitive

demands (Jueptner et al., 1997), whereas the posterior parts may be

more related to motor performance itself (LeheÂricy et al., 1998).

Brain activation during fast open-loop handwriting
Fast open-loop handwriting was associated with a smaller extent of

brain activation compared with slow closed-loop handwriting.

Furthermore, the brain activation pattern was more lateralized,

showing a more preponderant activation of left-hemispheric parieto-

frontal motor areas. We attribute the smaller magnitude in brain

activation to several factors. Fast open-loop handwriting required a

lower level of sensorimotor integration and less attentional control

than slow closed-loop handwriting. More importantly, the manual

motor network needed only half of the neuronal processing time to

complete the writing task as opposed to slow velocity-controlled

handwriting.

Correlational analysis revealed no brain area within the motor

network related to handwriting, which showed a graded increase in

rCBF with the magnitude of movement automation of handwriting

movements. The basal ganglia, especially, did not show a graded

activation during automated handwriting. The lack of signi®cant

activation within the basal ganglia during automatic open-loop

handwriting is in accordance with recent PET studies by Seitz et al.

(1997) and IbaÂnÄez et al. (1999), who have observed no basal ganglia

activation during ¯uent automated writing in healthy subjects.

Interestingly, closed-loop velocity-controlled handwriting move-

ments resulted in a writing-related activation of the basal ganglia

(Ceballos-Baumann et al., 1997). Moreover, a recent fMRI study

found a consistent activation of the basal ganglia during self-paced

signing and zigzagging with the ®nger or toe (Rintjes et al., 1999).

However, Rintjes et al. (1999) performed a categorical comparison of

self-paced signing (and zigzagging) with rest, whereas the present

study compared externally paced handwriting with holding the pencil.

Thus, differences in the experimental conditions may explain the

discrepant activation pattern in the basal ganglia. The lack of an

automation-related increase in functional activity in the basal ganglia

during fast open-loop handwriting is in line with the notion that the

corticobasal ganglia±thalamocortical motor loops are not essential for

fast automated running of complex motor programmes in healthy

subjects once motor pro®ciency has been achieved (Brooks, 1997).

However, there are two points worth making about the lack of

negative covariance between NIV and rCBF. First, the cerebellum

was outside the ®eld of view of the PET scanner. Hence, we cannot

exclude that distinct parts of the cerebellum may show a signi®cant

negative covariance between NIV and rCBF. Second, a `ceiling'

effect may account for the absence of negative covariance with rCBF,

as handwriting per se resulted in a considerable activation of the

entire motor network regardless of the mode of motor control. Thus,

when comparing fast automatic with slow closed-loop handwriting, it

is conceivable that slight additional increases in functional activity in

a distinct motor control area related to fast open-loop performance

might have been missed in the present study. For instance, this may

apply to the caudal SMA, which is thought to be especially involved

in automatic open-loop motor performance (Jenkins et al., 1994).

Acknowledgements

The authors would like to express their gratitude to Ms C. Kruschke and Ms G.

Dzewas for their assistance during PET acquisition, to Ms N. Nguyen for

careful review of the manuscript, and to the staff of the Radiochemistry

Section. The study was supported by the German Research Council

(Sonderforschungsbereich `Sensomotorik' SFB 462, Teilprojekt C3).

Abbreviations

ACC, anterior cingulate cortex; NIV, number of inversions in velocity; PET,

positron emission tomography; PMD, dorsal premotor area; PMV, ventral

premotor area; rCBF, regional cerebral blood ¯ow; SMA, supplementary

motor area.

References

Bartenstein, P., Weindl, A., Spiegel, S., Boecker, H., Wenzel, R., Ceballos-

Baumann, A.O., Minoshima, S. & Conrad, B. (1997) Central motor

processing in Huntington's disease. Brain, 120, 1553±1567.

Blinkenberg, M., Bonde, C., Holm, S., Svarer, C., Andersen, J., Paulson, O.B.

& Law, I. (1996) Rate dependence of regional cerebral activation during

performance of a repetitive motor task: a PET study. J. Cereb. Blood Flow

Metab., 16, 794±803.

Boecker, H., Dagher, A., Ceballos-Baumann, A.O., Passingham, R.E., Samuel,

M., Friston, K.J., Poline, J., Dettmers, C., Conrad, B. & Brooks, D.J. (1998)

Role of human rostral supplementary motor area and the basal ganglia in

motor sequence control: investigations with H215O PET. J. Neurophysiol.,

79, 1070±1080.

Brooks, D.J. (1997) Neuroimaging of movement disorders. In Watts, R.L.,

Koller, W.C. (eds), Movement Disorders: Neurologic Principles and

Practice. McGraw-Hill, New York, NY, pp. 31±48.

Catalan, M.J., Honda, M., Weeks, R.A., Cohen, L.G. & Hallett, M. (1998) The

functional neuroanatomy of simple and complex sequential ®nger

movements: a PET study. Brain, 121, 253±264.

Catalan, M.J., Ishii, K., Honda, M., Samii, A. & Hallett, M. (1999) A PET

study of sequential ®nger movements of varying length in patients with

Parkinson's disease. Brain, 122, 483±495.

Ceballos-Baumann, A.O., Sheean, G., Passingham, R.E., Marsden, C.D. &

Brooks, D.J. (1997) Botulinum toxin does not reverse the cortical

dysfunction associated with writer's cramp. A PET study. Brain, 120,

571±582.

Corbetta, M., Miezin, F.M., Shulman, G.L. & Petersen, S.E. (1993) A PET

study of visuospatial attention. J. Neurosci., 13, 1202±1226.

Dettmers, C., Fink, G.R., Lemon, R.N., Stephan, K.M., Passingham, R.E.,

Silberzweig, D., Holmes, A., Ridding, M.C., Brooks, D.J. & Frackowiak,

R.S. (1995) Relation between cerebral activity and force in motor areas of

human brain. J. Neurophysiol., 74, 802±815.

Eichhorn, T.E., Gasser, T., Mai, N., Marquardt, C., Arnold, G., Schwarz, J. &

Oertel, W.H. (1996) Computational analysis of open loop handwriting

movements in Parkinson's disease: a rapid method to detect dopamimetic

effects. Mov. Disord., 11, 289±297.

Fox, P.T. & Mintun, M.A. (1989) Noninvasive functional brain mapping by

change-distribution analysis of averaged PET images of H215O tissue

activity. J. Nuclear. Med., 30, 141±149.

Freund, H.J. Time control of hand movements. Prog. Brain Res., 64, 287±294.

Friston, K.J., Ashburger, J., Poline, J.B., Frith, C.D., Heather, J.D. &

Frackowiak, R.S.J. (1995a) Spatial registration and normalization of

images. Hum. Brain Mapp., 2, 1±25.

Friston, K.J., Frith, C.D., Liddle, P.F., Dolan, R.J., Lammertsma, A.A. &

Frackowiak, R.S.J. (1990) The relationship between global and local

changes in PET scans. J. Cereb. Blood Flow Metab., 10, 458±466.

Friston, K.J., Frith, C.D., Liddle, P.F. & Frackowiak, R.S.J. (1991) Comparing

functional (PET) images: the assessment of signi®cant change. J. Cereb.

Blood Flow Metab., 11, 690±699.

Friston, K.J., Holmes, A., Worsley, K.J., Poline, J.B., Frith, C.D. &

Frackowiak, R.S.J. (1995b) Statistical parametric maps in functional

imaging: general linear approach. Human Brain Mapp., 2, 189±210.

Central processing of writing velocity 735

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736

Geyer, S., Ledberg, A., Schleichter, A., Kinomura, S., Schormann, T., Burgel,

U., Klingberg, T., Larsson, J., Zilles, K. & Roland, P.E. (1996) Two

different areas within the primary motor cortex of man. Nature, 382, 805±

807.

Gitelman, D.R., Alpert, N.M., Kosslyn, S., Daffner, K., Scinto, L., Thompson,

W. & Mesulam, M.M. (1996) Functional imaging of human right

hemispheric activation for exploratory movements. Ann. Neurol., 39,

174±179.

Grafton, S.T., Fagg, A.H. & Arbib, M.A. (1998) Dorsal premotor cortex and

conditional movement selection: a PET functional mapping study. J.

Neurophysiol., 79, 1092±1097.

Halsband, U., Ito, N., Tanji, J. & Freund, H.J. (1993) The role of the premotor

cortex and the supplementary motor area in the temporal control of

movement in man. Brain, 116, 243±266.

Hollerbach, J.M. (1981) An oscillation theory of handwriting. Biol. Cybern.,

39, 139±156.

IbaÂnÄez, V., Sadato, N., Karp, B., Deiber, M.P. & Hallett, M. (1999) De®cient

activation of the motor cortical network in patient's with writer's cramp.

Neurology, 53, 96±105.

Jahanshahi, M., Jenkins, I.H., Brown, R.G., Marsden, C.D., Passingham, R.E.

& Brooks, D.J. (1995) Self-initiated versus externally triggered movements.

I. An investigation using measurement of regional cerebral blood ¯ow with

PET and movement-related potentials in normals and Parkinson's disease

subjects. Brain, 118, 913±933.

Jancke, L., Specht, K., Mirazade, S., Loose, R., Himmelbach, M., Lutz, K. &

Shah, N.J. (1998) A parametric analysis of the `rate effect' in the

sensorimotor cortex: a functional magnetic resonance imaging analysis in

human subjects. Neurosci. Lett., 252, 37±40.

Jenkins, I.H., Brooks, D.J., Nixon, P.D., Frackowiak, R.S.J. & Passingham,

R.E. (1994) Motor sequence learning: a study with positron emission

tomography. J. Neurosci., 14, 3775±3790.

Jueptner, M., Frith, C.D., Brooks, D.J., Frackowiak, R.S.J. & Passingham,

R.E. (1997) Anatomy of motor learning. II. Subcortical structures and

learning by trial and error. J. Neurophysiol., 77, 1325±1337.

Kalaska, J.F., Cohen, D.A.D., Prud'Homme, M. & Hyde, M.L. (1990) Parietal

area 5 neuronal activity encodes movement kinematics, not movement

dynamics. Exp. Brain Res., 80, 351±364.

Kawashima, R., Inoue, K., Sugiura, M., Okada, K., Ogawa, A. & Fukuda, H.

(1999) A positron emission tomography study of self-paced ®nger

movements at different frequencies. Neuroscience, 92, 107±112.

Kunesch, E., Binkofski, F. & Freund, H.J. (1989) Invariant temporal

characteristics of manipulative hand movements. Exp. Brain Res., 78,

539±546.

Kunzle, H. (1975) Bilateral projections from the precentral motor cortex to the

putamen and other parts of the basal ganglia. An autoradiographic study in

Macaca fascicularis. Brain Res., 88, 195±209.

Lacquaniti, F., Guigon, E., Bianchi, L., Ferraina, S. & Caminiti, R. (1995)

Representing spatial information for limb movement: role of area 5 in the

monkey. Cereb. Cortex, 5, 391±409.

LeheÂricy, S., Van de Moortele, P.F., Lobel, L., Paradis, A.L., Vidalhet, M.,

Frouin, V., Neveu, P., Agid, Y., Marsault, C. & Le Bihan, D. (1998)

Somatotopical organization of striatal activation during ®nger and toe

movement: a 3-T functional magnetic resonance imaging study. Ann.

Neurol., 44, 398±404.

Marquardt, C., Gentz, W. & Mai, N. (1999) Visual control of automated

handwriting movements. Exp. Brain Res., 128, 224±228.

Marquardt, C. & Mai, N. (1994) Computational procedures for movement

analysis in handwriting. J. Neurosci. Meth., 52, 39±45.

Menon, V., Glover, G.H. & Pfefferbaum, A. (1998) Differential activation of

dorsal basal ganglia during externally paced and self paced sequences of

arm movements. Neuroreport, 9, 1567±1573.

Mesulam, M.M. (1998) From sensation to cognition. Brain, 121, 1013±1052.

Meyer, E., Ferguson, S.S., Zatorre, R.J., Alivisatos, B., Marrett, S., Evans,

A.C. & Hakim, A.M. (1991) Attention modulates somatosensory cerebral

blood ¯ow response to vibrotactile stimulation as measured by positron

emission tomography. Ann. Neurol., 29, 440±443.

Morasso, P. & Mussa Ivaldi, F.A. (1982) Trajectory formation and

handwriting. Biol. Cybern., 45, 131±142.

Mountcastle, V.B., Lynch, J.C., Georgopoulos, A., Sakata, H. & Acuna, C.

(1975) Posterior association cortex of the monkey: Command functions for

operations in extrapersonal space. J. Neurophysiol., 38, 871±908.

Old®eld, R.C. (1971) The assessment and analysis of handedness: the

Edinburgh inventory. Neuropsychologia, 9, 97±113.

Parent, A. & Hazrati, L.N. (1995) Functional anatomy of the basal ganglia. I.

The cortico-basal ganglia-thalamo-cortical loop. Brain Res. Rev., 20, 91±

127.

Phillips, J.G., Bradshaw, J.L., Chiu, E. & Bradshaw, J.A. (1994)

Characteristics of handwriting of patients with Huntington's disease. Mov.

Disord., 9, 521±530.

Plamodon, R. (1995) A kinematic theory of rapid human movements, part I.

Movement representation and generation. Biol. Cybern., 72, 295±307.

Rintjes, M., Dettmers, C., BuÈchel, C., Kiebel, S., Frackowiak, R.S.J. &

Weiller, C.A. (1999) Blueprint for movement: functional and anatomical

representations in the human motor system. J. Neurosci., 15, 8043±8048.

Rizzolatti, G., Luppino, G. & Matelli, M. (1998) The organization of the

cortical motor system: new concepts [Review]. Electroencephalogr. Clin.

Neurophysiol., 106, 283±296.

Sadato, N., Campbell, G., IbaÂnÄez, V., Deiber, M. & Hallett, M. (1996)

Complexity affects regional cerebral blood ¯ow change during sequential

®nger movements. J. Neurosci., 16, 2691±2700.

Sadato, N., IbaÂnÄez, V., Campbell, G., Deiber, M.P., Le Bihan, D. & Hallett, M.

(1997) Frequency-dependent changes of regional cerebral blood ¯ow during

®nger movements: functional MR compared to PET. J. Cereb. Blood Flow

Metab., 17, 670±679.

Sakata, H., Takaoka, Y., Kawarasaki, A. & Shibutani, H. (1973)

Somatosensory properties of neurons in the superior parietal cortex (area

5) of the rhesus monkey. Brain Res., 64, 85±102.

Samuel, M., Ceballos-Baumann, A.O., Blin, J., Uema, T., Boecker, H.,

Passingham, R.E. & Brooks, D.J. (1997) Evidence for lateral premotor and

parietal overactivity in Parkinson's disease during sequential and bimanual

movements. Brain, 120, 963±976.

Seitz, R., Canavan, A.G., Yaguez, L., Herzog, H., Tellmann, L., Knorr, U.,

Huang, Y. & Homberg, V. (1997) Representations of graphomotor

trajectories in the human parietal cortex: evidence for controlled

processing and automatic performance. Eur. J. Neurosci., 9, 278±289.

Siebner, H.R., Ceballos-Baumann, A., Standhardt, H., Auer, C., Conrad, B. &

Alesch, F. (1999a) Changes in handwriting due to bilateral high-frequency

stimulation of the subthalamic nucleus in Parkinson's disease. Mov. Disord.,

14, 964±971.

Siebner, H.R., Tormos, J.M., Ceballos-Baumann, A.O., Auer, C., Catala,

M.D., Conrad, B. & Pascual-Leone, A. (1999b) Low-frequency repetitive

transcranial magnetic stimulation of the motor cortex in writer's cramp.

Neurology, 52, 529±537.

Sirigu, A., Duhamel, J.R., Cohen, L., Pillon, B., Dubois, B. & Agid, Y. (1996)

The mental representation of hand movements after parietal cortex damage.

Science, 273, 1564±1568.

Stephan, K.M., Fink, G.R., Passingham, R.E., Silbersweig, D., Ceballos-

Baumann, A.O., Frith, C.D. & Frackowiak, R.S. (1995) Functional anatomy

of the mental representation of upper extremity movements in healthy

subjects. J. Neurophysiol., 73, 373±386.

Talairach, J. & Tournoux, P. (1988) Co-Planar Stereotaxic Atlas of the Human

Brain. Thieme, Stuttgart.

Turner, R.S., Grafton, S.T., Votaw, J.R., Delong, M.R. & Hoffman, J.M.

(1998) Motor subcircuits mediating the control of movement velocity: a

PET study. J. Neurophysiol., 80, 2162±2176.

van Mier, H., Tempel, L.W., Perlmutter, J.S., Raichle, M.E. & Petersen, S.E.

(1998) Changes in brain activity during motor learning measured with PET:

Effects of hand of performance and practice. J. Neurophysiol., 80, 2177±

2199.

Weiller, C., Chollet, F., Friston, K.J., Wise, R.J. & Frackowiak, R.S.J. (1992)

Functional reorganization of the brain in recovery from striatocapsular

infarction in man. Ann. Neurol., 31, 463±472.

Winstein, C.J., Grafton, S.T. & Pohl, P.S. (1997) Motor task dif®culty and

brain activity: investigation of goal-directed reciprocal aiming using

positron emission tomography. J. Neurophysiol., 77, 1581±1594.

Wise, S.P., Boussaoud, D., Johnson, P.B. & Caminiti, R. (1997) Premotor and

parietal

cortex:

corticocortical

connectivity

and

combinatorial

computations. Annu. Rev. Neurosci., 20, 25±42.

Worsley, K.L., Evans, A.C., Marrett, S. & Neelin, P. (1992) A three-

dimensional statistical analysis for CBF activation studies in human brain.

J. Cereb. Blood Flow Metab., 12, 900±918.

736 H. R. Siebner et al.

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 14, 726±736