Exp Brain Res (2000) 130:362–374
Digital Object Identifier (DOI) 10.1007/s002219900262
© Springer-Verlag 2000
R E S E A R C H A RT I C L E
M.R. Velicki · C.J. Winstein · P.S. Pohl
Impaired direction and extent specification of aimed arm movements
in humans with stroke-related brain damage
Received: 14 December 1998 / Accepted: 28 September 1999 / Published online: 10 December 1999
Abstract The role of sensorimotor (S-M) areas in the
specification of kinematic parameters for aiming movements was studied by comparing the performance of six
subjects with unilateral stroke to that of matched control
subjects. Rapid arm movements were made to one of
four targets by rotating the forearm in a short (20°) or
long (45°) arc of motion. Thus, the four targets represented two directions (flexion or extension) and two extents (short or long). Subjects with stroke used the arm
ipsilateral to the side of the lesion. A timed-response
paradigm was used to dissociate response initiation and
specification. Subjects initiated movements in concert
with the last of four regularly timed tones. A visual cue
of the designated target was presented during the preparation interval (400–0 ms) before the last tone. Targets
were presented in a fixed sequence (predictable condition) or a random sequence (unpredictable condition).
No significant differences in performance were found
between stroke and control groups in the predictable
condition. In the unpredictable condition, subjects with
stroke produced more direction errors and were less accurate in extent than the control subjects. As specification time increased to 400 ms, the frequency of direction
errors attenuated less for stroke than for control groups,
but the reduction in magnitude of extent errors was similar for the two groups. When specification was minimal
(i.e., <100 ms), default responses were distributed equally between directions and clustered around the short extent. Further, wrong direction responses did not converge
on the designated extent as specification time increased.
This pattern of findings is consistent with a view of parameterization of planning and executing movements, in
M.R. Velicki · C.J. Winstein (✉)
Department of Biokinesiology and Physical Therapy,
University of Southern California, 1540 East Alcazar St.,
CHP 155, Los Angeles, CA 90033, USA
e-mail: Winstein@hsc.usc.edu
Tel.: +1-323-442-2903, Fax: +1-323-442-1515
P.S. Pohl
Center on Aging and Department of Physical Therapy Education,
University of Kansas Medical Center, Kansas City, KA, USA
which direction and extent can be specified in parallel.
Our results suggest that ipsilateral S-M areas contribute
to the specification of an optimal motor program, particularly when imperative programming of unimanual goaldirected aiming movements is required.
Key words Human · Motor control · Arm movements ·
Stroke · Ipsilateral motor pathways
Introduction
The purpose of this investigation was to evaluate the role
of sensorimotor (S-M) areas in the specification of
movement parameters for visually cued goal-directed
aiming movements. Specifically, the ipsilateral arm
movements of subjects with unilateral cerebral damage
were compared with those of healthy matched control
subjects to provide insight into the role of ipsilateral S-M
areas in the specification of movement direction and extent.
Several neurophysiological studies have implicated
S-M areas in the specification of direction and extent
movement parameters. Georgopoulos et al. (1989) and
others have shown that cells in M1 are broadly tuned to
movement direction. Posterior parietal area-5 cells have
also been shown to be direction sensitive (Kalaska et al.
1983). Neurons in the basal ganglia (globus pallidus)
were modulated with movement extent (Georgopoulos et
al. 1983). More recently, Kurata (1993) recorded from
neurons in the premotor cortex that varied with changes
in both extent and direction. Finally, Fu and colleagues
(1993) showed that cells in M1 and the superior precentral premotor area modulate activity with extent and direction during arm-reaching movements (see also Fu et
al. 1995). Single-cell recordings of cell activity associated with the planning and execution of arm-aiming movements are almost always made from areas in the contralateral hemisphere. However, recent neural imaging
studies have shown that, for both healthy subjects and
those status-post unilateral stroke, there is increased met-
363
abolic activity in bilateral S-M areas during the performance of unimanual preplanned motor sequences or
complex sensory-motor transformations (Cheyne et al.
1995; Chollet et al. 1991; Deiber et al. 1991; Roland et
al. 1980; Winstein et al. 1997).
Subjects with unilateral cerebral hemisphere damage
involving S-M areas have demonstrated deficits in the
preparation and execution of programmed actions involving the ipsilateral limb, as evidenced by increased
RT (Dee and Van Allen 1973) and decreased accuracy of
ballistic or programmed aiming movements (Carmon
1971; Haaland et al. 1987; Winstein et al. 1999; Wyke
1967, 1971) or both (Haaland and Harrington 1989;
Tartaglione et al. 1991). These deficits do not simply reflect impaired response execution because performance
of the limb ipsilateral to the lesion is relatively normal in
conditions that do not require imperative programming.
Imperative programming involves the construction of a
motor plan under an imposed temporal constraint
(Haaland and Harrington 1989; see also Pohl et al. 1997
for a review of ipsilateral limb deficits).
Indeed, the neural activation pattern associated with
the planning of unimanual actions can involve S-M areas
in both hemispheres, while the pattern associated with
motor execution may only involve S-M areas in the contralateral hemisphere (Cheyne et al. 1995). It has been
suggested that the ipsilateral M1 is not directly involved
in activating spinal motoneurons (at least those required
to execute distal hand muscles), but is more involved in
the planning and higher-order organization of movements (Chen et al. 1997b). This idea is further supported
by the greater involvement of the ipsilateral motor areas
in complex compared with simple movement sequences,
as revealed by positron emission tomography (Dieber et
al. 1991; Winstein et al. 1997), functional imaging (Rao
et al. 1993; Shibasaki et al. 1993), repetitive transcranial
magnetic stimulation [rTMS, (Chen et al. 1997a)], and
movement-related cortical potential studies (Kitamura et
al. 1993). For example, Deiber and colleagues (1991)
found that the pattern of regional cerebral blood flow
(rCBF) differed between a simple conditional response
and one where imperative response programming was
required. In the imperative programming condition, subjects were required to move a joystick to one of four targets in response to an unpredictable auditory cue. Thus,
the movement planning and response preparation had to
be completed during the interval between the presentation of the auditory cue and movement onset. Regional
cerebral blood flow increased bilaterally in parietal and
frontal areas. In contrast, a primarily contralateral neural
activation pattern was seen when subjects moved a joystick to one target in response to a predictable auditory
cue.
In neurologically normal subjects, a timed-response
paradigm (TRP) has been used to evaluate the specification of direction and extent of goal-directed actions
(Favilla and De Cecco 1996; Favilla et al. 1989, 1990;
Ghez et al. 1989; Hening et al. 1988a, 1988b). With this
paradigm, RT is manipulated as the independent variable
rather than measured as the dependent variable. This allows the specification process to be reflected solely by
the accuracy of rapid, goal-directed movement responses. The characteristics of responses produced following
various imposed RTs throughout the preparation interval
also provide insight into the nature of the parameterspecification process.
In these previous studies using a TRP, when targets
appeared in a fixed sequence (predictable condition),
healthy subjects prepared each response ahead of the
movement cue. Response accuracy was high and independent of preparation time. When targets appeared in a
random sequence (unpredictable condition), response accuracy was dependent on preparation time and responses
gradually became more accurate as preparation time increased. When preparation time was less than 100 ms
(forcing specification to be minimal), subjects produced
a default response. Using these methods, Favilla and colleagues (1989) demonstrated gradual and parallel specification of direction and extent for aimed force responses. There is some evidence that healthy subjects may not
be confined only to a parallel specification strategy.
Megaw (1972) noted that healthy subjects performing a
RT task appeared to use serial processing to transform
direction and extent information.
Damage to ipsilateral S-M areas may likely impair the
neural activity associated with the planning and higherorder organization of movement (Chen et al. 1997b). In
our experiment, a TRP procedure was used to evaluate
the role of ipsilateral S-M areas in the parameter-specification process by comparing the responses of subjects
with unilateral cerebral-hemisphere damage with those
of healthy control subjects. Specifically, we evaluated
how adults with chronic stroke, using the limb ipsilateral
to the lesion, specified the parameters of extent and direction during the preparation interval prior to the initiation of a rapid, visually cued, goal-directed arm movement. The use of an unpredictable and predictable condition allowed us to distinguish deficits in response specification from deficits in performance of the task itself,
which required visual cue identification and response execution. Thus, we investigated whether subjects with
stroke would perform differently from controls only in
the unpredictable condition, which would indicate a deficit in imperative motor planning, or whether they would
also perform differently in the predictable condition, a
result that would indicate a difficulty in simply performing the movement task adequately. The use of the TRP
also allowed us to evaluate the nature of the specification
process. Three possible alterations in the specification
process following ipsilateral S-M area damage were considered. First, would subjects with a S-M lesion demonstrate a greater magnitude of extent errors or a greater
frequency of direction errors? Second, would the stroke
group specify extent and direction in series rather than in
parallel? Third, would the stroke group require more
time to attain response accuracy than would matched
control subjects, thereby altering the time-course for parameter specification? Preliminary results of this work
364
have been reported previously (Velicki et al. 1993; Fisher
et al. 1996).
Materials and methods
Subjects
Twelve adults participated in this study: six with a unilateral S-Marea lesion secondary to a stroke involving the anterior circulation
system [age=62.7±8.9 years (mean±SD)], and six neurologically
healthy control subjects (aged 62.5±8.0 years). Table 1 summarizes the demographic and lesion information for each subject. Each
subject in the stroke group demonstrated some motor impairment
that affected their functional mobility. They all used an assistive
device to ambulate at a household or limited community level.
Further, all but subject 10 demonstrated a contralateral hemiparesis and hemisensory loss of varying degrees. The average length
of time from onset of the brain damage to participation in the
study was (mean±SD) 4.2±2.9 years. Lesion location was obtained
from a computer tomography (CT) or magnetic resonance imaging
(MRI) scan for each subject (Fig. 1). To determine lesion location,
the lesion site was mapped onto the appropriate view in the functional atlas of Domasio and Domasio (1989). Lesion locations
were identified for each subject in Table 1 using a classification
scheme modified from Kunesch et al. (1995). Areas of central-nervous-system involvement were consistent with the distribution of
the middle cerebral artery, including primary sensory and motor
cortex (S1, M1), surrounding association areas, underlying white
matter (internal capsule), and the subcortical basal ganglia in four
cases (see Table 1).
To obtain an index of lesion volume, the perimeter of each lesion site was outlined using a hand digitizer, and the area was
measured. This area was multiplied by the slice thickness to obtain
the volume of each lesion slice. These values were summed to obtain an index of lesion volume (Tartaglione et al. 1991). Each control subject was matched to a subject post-stroke by age, handdominance, and arm used for the task. All subjects passed screening tests for visual field, visual acuity, and hearing and completed
an extensive practice session to criteria. Subjects were excluded
from participation if they presented with evidence of posterior circulation involvement or if bilateral lesion areas were present on
CT or MRI scan. Each subject received $60 for participation.
Apparatus and task
The experimental apparatus consisted of an instrumented manipulandum, IBM compatible CPU, and monitor. The manipulandum
was a lightweight aluminum lever, which was mounted on the corner of a table with a near frictionless axle such that the lever was
free to move in the horizontal plane above the table surface. A
vertically oriented handgrip at the distal end of the lever was adjusted to accommodate the length of each subject’s forearm. A
wooden cover was mounted over the table and lever to prevent the
subject from visually monitoring the lever position.
A computer monitor was positioned in front of the subject at
eye level. The distance between the computer monitor and the
subject ranged between 42 and 62 cm and was determined as the
Table 1 Subject and lesion characteristics. CVA Cerebrovascular accident, M male, F female, R right, L left
Subject
ID
Subject
match
Sex
Age
(years)
Hand
dominance
Etiology
Side
CVA
Lesion
duration
(years)
Lesion
size
(cm3)
Lesion location
7
8
9
10
11
12
F
F
M
F
M
F
M
54
61
67
56
61
76
51
R
R
R
R
L
R
R
–
–
–
–
–
–
Infarct
–
–
–
–
–
–
R
–
–
–
–
–
–
5.0
–
–
–
–
–
–
11.5
8
F
53
R
Infarct
R
2.8
50.6
9
M
71
R
Infarct
R
5.1
83.7
10
11
F
M
69
62
R
L
Hemorrhage
Infarct
L
L
0.8
2.2
2.2
64.2
12
M
70
R
Hemorrhage
L
9.3
51.6
–
–
–
–
–
–
Primary motor cortex with
subcortical frontal extensions
Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical extensions;
extensions into temporal
association area
Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical extensions;
some extension into prefrontal
and temporal association areas
Posterior parietal cortex only
Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical extensions;
minor extensions into temporal
and prefrontal association areas
Striatocapsular + cortex;
primary sensorimotor cortical areas
with subcortical frontal/parietal
extensions; extension into parietal
and temporal association areas
1
2
3
4
5
6
7
365
Fig. 1 Lesion location for each
stroke-group subject from magnetic resonance imaging (MRI)
or computer tomography (CT)
images. Each section shows the
lesion-site area of greatest extent using the functional atlas
of Domasio and Domasio
(1989)
distance where the subject could comfortably read messages displayed on the monitor.
The lever was interfaced with an A-to-D board (Keithly/Metrabyte) installed in a 486-based, 50 MHz IBM-compatible computer.
Custom ASYST software (Vanman 1992) was used for the control
of data acquisition, data analysis, and storage. The analog signal
from the lever potentiometer was sampled at 250 Hz and converted to digital without filtering. Each trial was 1000 ms in duration
and consisted of A-D sampling of the lever potentiometer beginning 100 ms before the movement cue. Movement time (MT) was
determined from movement onset to offset. Movement onset (start
of movement) was defined as the onset of a 20 ms interval (moving window) in which the A-D values were changed from baseline
(baseline break) and were consistently > or <3–5× the calibrated
resting SD, or approximately ±0.5° change. Movement endpoint
was defined as the onset of a 20-ms interval in which the A-D values did not change greater than 5× the calibrated resting SD.
Procedure
Each subject read and signed an Institutional Review Board-approved informed consent form. Subjects performing with the right
arm sat to the left of the lever axis and those performing with the
left arm sat to the right of the lever axis. Subjects post-stroke performed with the arm ipsilateral to the lesion.
When the lever was placed in the home (or start) position, the
subject was positioned in approximately 45° shoulder flexion and
abduction and 70° elbow flexion. The task involved moving the
lever from this home position to one of four target locations produced by mixing two directions and two extents. Subjects flexed
or extended the elbow in either a short arc of motion (20° from the
home position) or a long arc of motion (45° from the home position).
The TRP is summarized in Fig. 2A. The onset of lever movement was to be timed with the last in a series of four tones. The
computer presented a series of four audible tones, each with a progressively higher pitch, increasing from 625 to 1300 Hz. The
tones each had 50 ms duration and were spaced by an inter-tone
interval of 500 ms. Prior to the fourth tone, a yellow arrow appeared and pointed to the designated target. This movement cue
was randomly presented before the fourth tone within one of
the four preparation time bins (1=0–99 ms, 2=100–199 ms,
3=200–299 ms, and 4=300–400 ms). Preparation time bins were
used to ensure that an adequate number of acceptable trials was
collected throughout the preparation interval for subsequent analysis (see Data analysis). The first bin was selected because the default response was produced following these latencies in previous
TRP studies (Favilla and De Cecco 1996; Favilla et al. 1989,
1990; Ghez et al. 1989; Hening et al. 1988a, 1988b). The remaining time was divided into three bins to allow for an assessment of
change over the full preparation interval and to ensure that each
bin had an adequate number of trials to be representative of the
time period (see Data analysis).
Prior to data collection, all subjects completed a three-phase
practice session. Subjects were trained to: (1) accurately move to
each target location without time constraints, (2) initiate responses
within a timing window surrounding the fourth tone, and (3) move
the lever quickly without final adjustments. Subjects practiced until 70% of the trials to each target were performed within preset
criteria. The criteria for accuracy, initiation time, and MT were
gradually narrowed over the practice phase to challenge the subject’s capability. At the completion of the practice phase, all subjects were able to: (1) move to within 5° of each target, (2) move
within a timing window of at most 95 ms before and 95 ms after
the tone, and (3) perform a rapid arm movement to each target
with a MT of less than 500 ms. To complete the practice period,
subjects required between 1 and 6 days and between 1.4 and 11.2 h
of practice. The number of practice trials was greater for the
stroke group, but this difference was not reliable between stroke
(2282±873) and control groups (data available for only five of the
six control subjects: 1301±832), Student’s t-test (t=–1.9, P=0.09).
The trial sequence depicted in Fig. 2A was used throughout the
third phase of practice and the subsequent experimental test phase.
At the beginning of a trial, the subject viewed the instructions:
“Return to home” displayed on the monitor. The subject used the
real-time position of a red pointer on the monitor as feedback to
move the lever to the home position. When the subject held a
steady arm position within 2.5° of the home position for 1 s, the
red pointer disappeared and a “Ready” cue appeared on the monitor. Then, the series of four tones began. Sometime between the
third and fourth tone, a yellow arrow (movement cue) appeared
and pointed to one of the four targets.
Throughout the movement, the yellow arrow continued to
point to the correct target. Following the movement, the red pointer reappeared and was fixed at the end position of the movement
366
Fig. 2 A The modified timedresponse paradigm. Movement
initiation was to be synchronized with the last in a series of
four tones (1–4). The movement cue was presented at variable times between 400 and
0 ms before the fourth tone.
The time-course of movement
cue and feedback presentation
is shown below. The stimulustone (S-T) interval, the stimulus-response (S-R) interval,
and the timing error [(S-R
interval) – (S-T interval)] are
shown above. B The written
feedback displayed when
movement was initiated at various times throughout the trial
sequence. This example shows
a timing window of 95 ms before and 95 ms after the fourth
tone
response. Visual feedback was provided regarding movement initiation (“too early”, “too late”) and MT (“too slow”) in relation to
pre-established criteria (see below for explanation). Following this
feedback, the cue, “Return to home” was displayed on the monitor, and the next trial sequence began. Subjects controlled the inter-trial interval by the rate at which they returned to the home position.
Because it is difficult to initiate movement exactly with the
fourth tone, acceptable trials were those initiated within an acceptable timing window (Fig. 2A). The stimulus-tone (S-T) interval
was the time between movement cue and the onset of the fourth
tone; it was the preparation time forced by the paradigm. The
stimulus-response (S-R) interval was the time between movement
cue and movement onset, the preparation time actually used by the
subject. The timing error [(S-T) – (S-R)] reflected the accuracy of
movement initiation time. Subjects were provided with visual
feedback to facilitate the timing of movement initiation within the
acceptable timing window (Fig. 2B). A trial was not included in
the analysis if the subject anticipated and moved before the movement cue (i.e., a negative S-R interval). This kind of timing error
constituted less than 9% of all acceptable responses per subject.
In addition, acceptable trials had to be completed within a preestablished MT criteria. A hierarchy of trial errors was used such
that “too slow” was provided only after trials in which movements
were initiated within the timing window criteria, but the MT was
greater than the MT criteria. The initial MT criterion for an acceptable trial was <450 ms. As each subject demonstrated the capability
to move faster, (i.e., very few “too slow” errors) the criterion MT
was decreased. Actual MTs were much shorter than criterion (control: 262±52 ms, stroke: 315±63 ms). As seen in Fig. 3, in three of
the four different preparation time bins, subjects in both groups
were able to produce rapid responses without overt corrections
during the movement. Sixty-four acceptable trials constituted a
“block”. Unacceptable trials were tallied, but not used for analyses.
Trial blocks were differentiated by two conditions. In the predictable condition, targets 1, 2, 3, and 4 were presented in serial
order, and, in the unpredictable condition, targets were presented
in a pseudo-randomized order (e.g., target 2, 4, 1, 3). Initially,
each subject performed 11 trial blocks: three predictable condition
blocks (number 1, 6, 11) and eight unpredictable condition blocks
(number 2–5, 7–10).
Inherent in the TRP is the precise control over the preparation
interval. This is accomplished by imposing a movement start time
(fourth tone). However, precise timing of movement onset is a difficult task for most subjects, even for those without neural deficits.
Therefore, we established an acceptable window around the fourth
tone (timing window) that was used during data collection (see explanation above). Subjects may, however, systematically adjust the
movement start time within the acceptable timing window. For example, subjects could move late when the preparation interval is
short, or move early when the preparation interval is long, resulting
in a high negative correlation between timing error and S-T interval. To control for this, after each successfully completed block,
the subject was shown a “timing error” × “S-T interval” plot along
with the Pearson correlation coefficient. The goal was to complete
each block with the smallest correlation possible. Subjects repeated
trial blocks until they produced 11 blocks similar to their best (i.e.,
lowest) correlation achieved during the testing period (average
block timing-error correlation range by subject r=0.41 to –0.07).
In order to ensure that subjects were not disregarding response
accuracy to meet these timing demands, a second plot of “absolute
error (AE)” × “S-T interval” was displayed. Acceptable trial
blocks were those with a significant accuracy correlation (r>0.20)
in the unpredictable condition (average block accuracy-correlation
range by subject r=–0.53 to –0.28). Trial blocks chosen for analysis where those that met the timing and accuracy criteria for the
unpredictable condition (eight “best” blocks) and the timing criteria for the predictable condition (three “best” blocks).
367
Fig. 3 Individual trial responses for one control subject (A)
and one stroke subject (B). For
all plots, the four targets (long
dashed lines) at ±20° and ±45°
are as displayed in the top plot
in A; extension movements are
positive, and flexion movements are negative. Time 0 is
the onset of trial sampling
100 ms before movement-cue
presentation. The three rows represent responses in three of
the four preparation intervals.
Top rows are responses to
each of the four targets with
<100 ms preparation interval;
middle rows are responses with
between 100–199 ms preparation interval; bottom rows are
responses with between
300–400 ms preparation interval. Short dashed lines are the
short target movements, and
solid lines are the long target
movements
Note that the aforementioned procedures included modifications to the standard TRP protocol (e.g., Favilla et al. 1990) necessary to assure that acceptable trials were those where the preparation interval was most effectively controlled. In summary, (1) for
the practice phase, a minimum criteria were used for accuracy and
timing error; (2) for the test phase, specific criteria were used for
trial acceptance; (3) timing-error and -accuracy scatter plots and
correlations were displayed as feedback after each completed
block of trials (J. Gordon, personal communication); and (4) these
correlation coefficients were used to choose trial blocks for analysis.
Data analysis
Three dependent variables were computed: (1) constant error
(CE), (movement offset – target position), (2) absolute error (AE),
i.e., |CE|, and (3) direction error, i.e., the number of responses in
the opposite direction of the intended target. For wrong direction
responses, AE was recalculated using the correct target extent in
the direction of the response (i.e., the virtual target). This was
done in order to dissociate the error due to direction specification
from that due to extent specification. Each block had 16 different
trial types, resulting from a combination of the four targets, and
the four S-R interval bins. Each block contained approximately
four trials of each trial type. The number of trials included in each
S-R bin varied somewhat as a result of acceptable timing errors.
For each subject, the mean AE was calculated for acceptable trials
with the same trial type and block condition. The mean AE of
each trial type was calculated from approximately 32 trials across
the eight blocks in the unpredictable condition and approximately
12 trials across the three blocks in the predictable condition.
To produce a full complement of acceptable test trials (i.e., 704
trials with specific S-T interval, target, and condition characteristics), each subject performed between 12 and 21 blocks. Each subject returned to the lab between 4 and 8 days and performed between two and four blocks each day. In all, each subject worked
between 6.3 and 13.4 h and performed between 2196 and 4455 trials. There were no differences between stroke and control group
means for either number of hours or trials in which they participated (Students t test: t=–0.62, –1.93, P>0.05, respectively).
Statistical analyses were used to evaluate the time course of
extent and direction specification for the stroke and control
groups. To evaluate extent specification, a mixed model analysis
of variance (ANOVA) was used with AE as the dependent variable. For analysis, the between-subject factor was group (stroke,
control) and the two within-subject factors were target (1, 2, 3, 4)
and S-R bin (1, 2, 3, 4). Post-hoc orthogonal contrasts were performed on significant effects from the ANOVA to determine the
locus of the effect. To determine direction specification, a stepwise logistical regression with main (group and S-R bin) and interaction terms (group × S-R bin) was used with the frequency of
wrong direction responses as the dependent variable.
368
Fig. 4 A Bar graph of mean
absolute error (AE) (±SEM) for
responses within each stimulusresponse (S-R) interval for each
group in the predictable condition. B Bar graph of mean AE
(±SEM) for responses within
each S-R interval for the unpredictable and predictable conditions. Data are averaged across
12 subjects and four targets.
C Bar graph of the mean AE
(±SEM) for responses to the
long and short extent targets
within each S-R interval in the
unpredictable condition. Data
are averaged across 12 subjects. D Bar graph of mean AE
(±SEM) for responses to the
long and short extent targets in
the unpredictable and predictable conditions. Data are averaged across 12 subjects. Mean
AE (±SEM) long targets: unpredictable condition =12±1.0°,
predictable condition =9.0±0.9°;
short targets: unpredictable
condition =5.4±0.4°, predictable condition =5.3±0.2°
Results
Target effect in predictable and unpredictable conditions
Predictable-condition responses are described first by
comparing between groups and then for all subjects. This
is followed by a description of the unpredictable-condition responses first for correct and then for wrong direction responses. General response characteristics are described first for all subjects, followed by the group differences that highlight the contribution of the ipsilateral
S-M areas in the movement parameter-specification process.
A target effect was obtained for AE in both the predictable condition, P=0.047, and the unpredictable condition, P=0.0002. As expected, responses to the shortextent targets were more accurate [range: 4.6±0.4°
(mean±SEM) to 6.2±0.5°] than those to the long-extent
targets (11.2±1.0° to 13.1±1.1°) for both correct and
wrong-direction responses. Post-hoc orthogonal contrasts revealed that mean AE was similar for targets of
the same extent, P>0.05, and different for targets of different extents, P<0.05, in the unpredictable condition.
Therefore, for all further analyses, responses to each of
the two similar-extent targets were combined.
Predictable-condition responses
The predictable condition was the control condition providing a measure of how well subjects performed the
task without the imperative programming demand. The
stroke group demonstrated slightly higher extent errors
(~3°) than the control group across all response intervals
(Fig. 4A); however, this difference was not reliable for
response accuracy, (P=0.12). All subjects preplanned
their responses before the movement cue, therefore, response accuracy (P=0.22) was not dependent on preparation time (Fig. 4B, striped bars). Direction errors were
rare, contributing to only 1% of the total responses in
this condition.
Unpredictable condition
Extent specification of correct direction responses
All subjects. In the unpredictable condition, as the S-R
interval increased, all subjects were able to use movement-cue information to improve response accuracy. A
conclusion that can be drawn from this is that, with longer preparation intervals, subjects planned their responses within the S-R interval. For correct direction responses, accuracy was dependent on the imposed preparation
time, P<0.00001, (Figs. 3, 4B, dark bars). Post-hoc orthogonal contrasts revealed that accuracy increased be-
369
Fig. 6 A Bar graph of mean absolute error (AE) (±SEM) for responses to the long and short extent targets for the stroke and control groups. Mean AE (±SEM) long targets: stroke =15.4±2.4°,
control =8.8±1.5°; short targets: stroke =5.5±0.8°, control
=5.3±1.0°. B Bar graph with group mean AE (±SEM) within each
stimulus-response (S-R) interval for correct direction responses
Fig. 5 A Scatter plot of constant error (CE) for a full complement
of correct direction responses for S1 (control subject) and (B) S8
(stroke subject) in the unpredictable condition. The horizontal line
at 0° represents the target. For these two subjects, the default position was the short target extent. C Scatter plot as in A and B for
S4 (control subject) in the unpredictable condition. For this subject, the default position was between the short and long target extents. S-R Stimulus-response interval
tween the first and second S-R intervals (P=0.001) and
between the second and third S-R intervals (P=0.002),
but not between the third and fourth S-R intervals
(P=0.32) (Figs. 3, 4B). Thus, response accuracy did not
improve when preparation time was 300 ms or greater.
For all subjects, responses to the short targets were
relatively accurate regardless of preparation time, whereas responses to the long targets increased in accuracy as
preparation time increased (Figs. 3, 4C). This resulted in
a significant target × S-R-interval effect for AE,
P=0.004. These findings suggest that the short target extent was programmed in advance as the default and the
long target extent was specified from the default extent
as the S-R interval increased. In fact, AE for responses
to the short extent targets was similar to that found in the
predictable condition, whereas AE for responses to the
long extent targets was somewhat higher in the unpredictable condition than in the predictable condition
(Fig. 4D). Indeed, 10 of the 12 subjects used the short
target extent as the default and specified only the long
target extent during the S-R interval (see, for example,
Fig. 5A, B). The remaining two subjects (S4 and S5,
both control) used a default position in-between the short
and long target extents and appeared to specify both
short and long target extents during the S-R interval (see,
for example, Fig. 5C).
Group differences. Absolute error to the long targets was
greater for the stroke group than the control group,
P<0.05, whereas AE to the short targets was similar between the groups, P=0.05 (Fig. 6A). This finding suggests that subjects in the stroke group were as capable as
control subjects of specifying the default extent in advance of the stimulus, but had difficulty scaling that default response to the long extent within the imposed S-R
interval (Fig. 3). However, the time course of extent
370
Fig. 7 A Scatter plot of constant error (CE) for a full complement
of responses to the long extent targets in the unpredictable condition for S2 (control subject) and (B) S7 (stroke subject). The horizontal line at 0° represents the target. Note that wrong-direction
responses (WDR bracketed) are clustered at the default extent
(~–65°). Only correct-direction long-extent responses are progressively scaled from the default to the correct target extent as stimulus-response (S-R) interval increases
specification was similar for both groups, P=0.75 (Fig.
6B).
Extent specification of wrong direction responses
All subjects. Wrong direction responses for both the
short and long targets were positioned near the short target extent, resulting in AE to the virtual target of
6.4±0.24° (mean±SEM) for responses to short targets
and 21.5±0.33° for responses to long targets; the latter
approximated the difference between the short and long
targets.
Wrong direction responses did not converge on the
correct target extent as preparation time increased. Responses to long target extents in the correct direction
were progressively scaled to the correct target extent,
while responses in the wrong direction remained near the
default extent regardless of preparation time (Fig. 7A,
B). For all subjects, mean AE remained relatively stable
across preparation intervals, P=0.14 (Fig. 8A).
Group differences. Overall, the stroke group produced
responses with a larger AE [15.5±0.42° (mean±SEM)]
Fig. 8 A Bar graph of mean absolute error (AE) (±SEM) to the
virtual target for wrong-direction responses within each stimulusresponse (S-R) interval. Values represent mean responses of all 12
subjects in the unpredictable condition. B Bar graph of mean AE
(SEM) to the virtual target for each target (short, long) and group
(C control, S stroke) within each S-R interval. C Frequency of direction errors as a percent of the total responses within each S-R
interval for the stroke and control groups in the unpredictable condition
than did the control group (12.2±0.34°), P<0.0001. Further, as the preparation interval increased, control subjects showed minimal scaling of extent, for only long target responses. In contrast, subjects with stroke did not
scale response extent as preparation interval increased
for either target (Fig. 8B). This group difference in
wrong-direction extent errors was reliable, as evidenced
by a triple interaction between group, target, and preparation interval (P<0.03). Although the control group
showed some capability to scale these wrong-direction
responses to the long target extent, that response was
markedly truncated (~4.7° change across preparation interval, see open bars in Fig. 8B) compared with that for
correct direction responses (~10.7° change across preparation intervals).
371
Direction specification
All subjects. Overall, approximately one-fourth of all responses were in the opposite direction of the designated
target. The frequency of direction errors decreased as
preparation time increased from 0 to 400 ms, χ23=534.13,
P<0.0001 (Fig. 8C). If direction was not specified, the
chances of moving in either of the two possible directions should have been equal. In fact, when preparation
time was <100 ms and specification was expected to be
minimal, the percentage of responses in the wrong direction was close to 50% for both groups (Fig. 8C). Alternatively, when preparation time was greater than
300 ms, the frequency of wrong direction responses for
all subjects was less than 10% of all responses within
that S-R interval.
Group differences. Overall, across all preparation intervals, the stroke group was 20% more likely to make an
error in direction than the control group, (χ2=7.287,
P<0.008). Figure 8C shows the group differences in direction specification across the four preparation intervals. For the shortest interval (<100 ms), the stroke
group was 1.2× (20%) more likely to make a direction
error than the control group. For the longest interval
(300–400 ms), the group differences were greater, with
the stroke group 1.9× (90%) more likely to make a direction error than the control group. More importantly and
in contrast to extent specification, the time course for direction specification was significantly different for the
two groups [χ2=34.14, P<0.0001). Figure 8C shows that
the reduction in frequency of direction errors proceeded
more gradually for the stroke group (solid bars) than for
the control group (open bars) over the preparation interval. The group difference in the rate of decrease in the
frequency of direction errors was greatest between the
second (100–199 ms) and third (200–299 ms) preparation intervals (Fig. 8C). The control subjects showed a
30% reduction in the number of direction errors, while
those with stroke only showed a 19% reduction in the
number of direction errors over the same preparation
time (i.e., 100–299 ms). This group difference in the
time course shows that the added preparation time (especially >200 ms) was more beneficial for control subjects
than for those with stroke of the ipsilateral hemisphere.
Discussion
The TRP allowed an examination of the nature of the
processes by which visually cued, goal-directed arm
movements are specified in advance of movement. With
the TRP approach, we found clear evidence of programming deficits in individuals with ipsilateral S-M area
brain damage; these deficits impaired both movement
extent and direction accuracy. Subjects with unilateral SM area damage were less accurate primarily in the unpredictable condition where imperative planning was required. In the predictable condition when specification
time was not constrained, movements using the ipsilateral limb were nearly as accurate as those of control subjects. Therefore, deficits in movement-parameter specification observed in the unpredictable condition were related to the preparation phase and not a reflection of impairments in visual cue identification, movement initiation/execution, or stimulus-response transformation processes.
These findings will be discussed first in relation to our
three proposed alterations in the specification process
and, second, with respect to the role of the ipsilateral S-M
areas in programming of goal-directed actions. Consistent
with the first suggested alteration, when time to prepare
was limited, subjects with stroke produced more direction
errors and were consistently less accurate at specifying
the long target extent. When the time for imperative response planning was constrained to <400 ms, both extent
and direction specification were compromised for those
with ipsilateral S-M area damage.
Our second proposed alteration suggested that, if imperative planning challenges the resources for parameter
specification after ipsilateral S-M area damage, one expected and parsimonious solution would be to invoke serial rather than parallel specification of direction and extent. In general, both groups showed a similar and gradual specification of extent and direction (see Figs. 6B and
8C), which suggests a predominantly parallel specification process. One surprising finding in this regard was
that, for both groups, extent specification of wrong direction responses was either truncated or aborted, as
these responses did not fully converge on the correct extent as preparation time increased. While this could be
interpreted as suggesting a serial specification strategy, it
alone does not suffice to dismiss a parallel specification
scheme. Our results suggest that at least partial specification of extent occurred before direction was completely
specified (i.e., wrong direction responses were specified
at the short target extent, see Fig. 7). Indeed, it is possible that parallel specification is the default control strategy for both groups and that it is aborted when a direction
error is detected (see Fig. 8B).
In previous TRP investigations, young, healthy adults
showed a parallel specification of direction and extent of
isometric force pulses (Favilla et al. 1989, 1990; Ghez et
al. 1989; Hening 1988a). Responses in both the correct
and incorrect directions converged on the target extent as
preparation time increased. Recently, Favilla and De
Cecco (1996) showed that the time course of direction
specification of a reaching movement (two directions
and two extents) in healthy adults was not influenced by
the need to simultaneously specify extent. While their results support a model of parallel and independent channel processing for direction and extent, it does not preclude the possibility that extent scaling was completed
after movement onset. Indeed, Fu et al. (1995) recently
showed a temporal parcellation of direction and extent
specification in the activity of M1 and PM cortex cells in
primates. Consistent with Fu et al. (1995) and others,
there is support in the literature for a serial and hierarchi-
372
cal movement-parameter selection process (Fu et al.
1993; Megaw 1972; Riehle and Requin 1989). Neural
activity was recorded as monkeys performed arm-aiming
movements to targets with different direction and extent
characteristics. Directionally selective neurons in PM
cortex were active during the preparation period prior to
movement. During this preparation period, very few neurons were recorded which clearly processed prior information about movement extent. During movement, M1
and PM activity was correlated with both direction and
extent (Fu et al. 1993). In addition, when the monkeys
were provided with advance information about the correct movement direction, RT decreased more than when
they were provided with information about the correct
movement extent (Riehle and Requin 1989). One critical
issue and possible explanation for the discrepancy in
these findings is the MT across these studies. If MT is
quite long, extent can be scaled and corrections made
during the movement with only a rudimentary or partial
specification made during the preparation interval. In
contrast, if MT is quite short, a more complete specification is necessary during the imperative preparation interval. In our case, the fact that all subjects were hypometric with short S-R intervals (see Fig. 5) suggests that only partial specification was made during the imperative
interval.
Although parallel specification of kinematic parameters is possible and most studies support such a strategy,
it might not be optimal under all task conditions. In our
study, the perceived penalty for direction errors may
have been greater than extent errors when subjects were
producing observable arm movements (as opposed to
isometric force pulses). Also, the MT was considerably
longer (~300 ms) under these movement conditions than
for the force-pulse conditions (~100 ms) used in previous work. In a subsequent paper, we have applied an early measure of response magnitude similar to that used by
Ghez and colleagues to ferret out the degree to which
preplanned specification and later compensatory adjustments contribute to final position (Fisher BE, Winstein
CJ, Velicki MR, submitted). We find that those with ipsilateral S-M area deficits are more dependent on the preplanned specification of extent than are the healthy control subjects, who can also correct on-line to achieve a
more accurate final position (Fisher et al. 1998).
The group difference in the time course for direction,
but not extent specification is an important finding for
two reasons. First, it provides further support for the independent specification of extent and direction, and, second, it highlights a potential role of the ipsilateral S-M
areas in the movement parameter specification process.
This first point will be discussed in relation to our third
possible alteration in the parameter-specification process. The potential role of the ipsilateral S-M areas in
this process will be discussed in the last section.
Our third proposed alteration was that the stroke group
would demonstrate a different time course compared with
that for the control group. As described above, this prediction held up for direction specification, but not extent
specification. Our results show that the time course for
direction specification was different between groups,
while that for extent specification was similar. Such a difference in time course for extent and direction provides
further support for a model of semi-independent processing of movement parameters (cf. Favilla and De Cecco
1996). Using the TRP paradigm, the time course naturally
involves the pre-movement interval. If direction specification primarily utilizes this pre-movement time frame
and extent specification can also use the interval after
movement onset, then a group difference in direction-,
but not extent-parameter specification for the pre-movement interval would be more likely than if both parameters were fully specified during the exact same time interval. The more gradual specification time course of direction for the stroke group, especially during the 100–
300 ms pre-movement interval, compared with that for
the control group suggests a potential role for the ipsilateral S-M areas in the chronometrics of imperative programming (see Fig. 8C). Chen et al. (1997a) induced timing errors in both simple and complex piano sequences
using rTMS of the ipsilateral M1 on either side, suggesting involvement of the ipsilateral motor cortex in processing of complex motor programs.
We now focus the discussion on the role of the ipsilateral S-M areas in the planning and organization of goaldirected actions. Independent of the time course, the error in extent and the frequency of direction errors were
greater in the stroke group than the control group. What
could possibly explain these findings in imperative programming in the arm ipsilateral to the lesion? While the
answer is speculative, we offer a few logical and testable
hypotheses. The first explanation is based on the assumption that the ipsilateral hemisphere is important for
imperative programming and considers what the nature
of the contribution might be. The second explanation is
based on the assumption that the contribution of the ipsilateral hemisphere to imperative programming is disturbed as a result of the negative effects (i.e., artifact) of
plasticity triggered by the brain damage itself.
First, anatomical data indicate that the ipsilateral motor pathway from M1 to the spinal cord projects primarily to axial and proximal muscles (Colebatch and Gandevia 1989). Indeed, there is little evidence for direct ipsilateral connections from M1 to distal hand muscles.
However, functional imaging and physiological studies
including rTMS show that ipsilateral M1 is involved in
the performance of complex finger movements (e.g.,
Chen et al. 1997a; Kim et al. 1993). One interpretation
of this apparent discrepancy is that ipsilateral M1 is not
directly involved in activating spinal motoneurons required to execute these movements, but is more involved
in the planning and the higher-order organization of
movements (Chen et al. 1997b; Haaland and Harrington
1990). This role for the ipsilateral hemisphere is consistent with our findings of significant group differences for
the unpredictable (complex, imperative) condition and
not the predictable (simple) condition where imperative
planning is likely not required.
373
A second possibility is that the ipsilateral effect is a
by-product of plasticity in the unaffected hemisphere. If
the unaffected hemisphere after stroke shows plastic
changes in motor output organization, one negative effect
of this plasticity could be a decreased or disordered organization of the contralateral (affected) or ipsilateral (unaffected) S-M areas involved in motor planning and higherorder organization. Such a possibility could explain the
findings in this study and might be the result of cortical
reorganization similar to that described in cases of repetitive strain injury (e.g., Byl et al. 1996). Indeed, there is
some suggestion that ipsilateral motor pathways may play
a role in functional recovery from stroke (Chollet et al.
1991; Lee and van Donkelaar 1995; Lemon 1993). In two
positron emission tomography (PET) studies, the stroke
group had significantly increased rCBF in the ipsilateral
SM1 with movement of the affected hand, but not the unaffected hand (Chollet et al. 1991; Weiller et al. 1992).
However, in a later report from the same group, the patients with the ipsilateral SM1 activation also had mirror
movements in the unaffected hand when they moved the
affected hand (Weiller et al. 1993). Several studies using
TMS found no evidence that the ipsilateral fast corticospinal tract was responsible for recovery of affected limb
function (Netz et al. 1997; Palmer et al. 1992; Turton et
al. 1996). Indeed, ipsilateral responses with TMS of the
undamaged hemisphere were more common among patients who had poor recovery than in those patients with
good recovery (e.g., Turton et al. 1996). While there is
evidence that the motor outputs in the unaffected hemisphere are significantly changed after stroke (e.g., Netz et
al. 1997), these pathways seem to be of little significance
for recovery, as the existence of these ipsilateral responses are not correlated with clinical improvement.
In summary, the role of the ipsilateral S-M areas in the
parameter specification process was revealed in a condition of imperative programming using the TRP protocol.
Subjects with unilateral stroke-related brain damage were
consistently less accurate than age-matched control subjects in the specification of movement extent and direction for ipsilateral arm aiming movements. In addition,
the time course for direction, but not extent specification
was altered in the stroke group compared with the control
group. Together, the group differences here and in previous work (cf. Winstein and Pohl 1995; Winstein et al.
1999) strongly suggests that the ipsilateral S-M areas
contribute both to the specification of an optimal plan or
task-relevant parameter set (i.e., motor program) and, to
some degree, the chronometrics (i.e., time-course) of the
specification process.
Acknowledgements We thank Karen Altman, MD for reading the
MRI and CT scans; Jim Gordon for valuable comments on an earlier version of this paper and for suggesting the particular timing
error by S-R interval feedback plots; John L. Morrison for statistical consultation; Beth Fisher for characterizing the lesion locations from the brain scans; and all the participants, especially the
stroke group, for their time and effort toward this study. Funding
for this project was provided in part from a grant to CJW and
MRV from the California Physical Therapy Fund.
References
Byl NN, Merzenich MM, Jenkins WM (1996) A primate genesis
model of focal dystonia and repetitive strain injury. I. Learning-induced dedifferentiation of the representation of the hand
in the primary somatosensory cortex in adult monkeys. Neurology 47:508–520
Carmon A (1971) Sequenced motor performance in patients with
unilateral cerebral lesions. Neuropsychologia 9:445–449
Chen R, Gerloff C, Hallett M, Cohen LG (1997a) Involvement of
the ipsilateral motor cortex in finger movements of different
complexities. Ann Neurol 41:247–254
Chen R, Cohen LG, Hallett M (1997b) Role of the ipsilateral motor cortex in voluntary movement. Can J Neurol Sci 24:284–
291
Cheyne D, Weinberg H, Gaetz W, Jantzen KJ (1995) Motor cortex
activity and predicting side of movement: neural network and
dipole analysis of pre-movement magnetic fields. Neurosci
Lett 188:81–84
Chollet F, DiPiero V, Wise RJ, Brooks DJ, Dolan RJ, Frackowiak
RS (1991) The functional anatomy of motor recovery after
stroke in humans: a study with positron emission tomography.
Ann Neurol 29:63–71
Colebatch JG, Gandevia SC (1989) The distribution of muscular
weakness in upper motor neuron lesions affecting the arm.
Brain 112:749–763
Dee HL, Van Allen MW (1973) Speed of decision-making processes in patients with unilateral cerebral disease. Arch Neurol
28:163–166
Deiber MP, Passingham RE, Colebatch JG, Friston KJ, Nixon PD,
Frackowiak RSJ (1991) Cortical areas and the selection of
movement: a study with positron emission tomography. Exp
Brain Res 84:393–402
Domasio H, Domasio AR (1989) Lesion analysis in neuropsychology. Oxford University Press, New York
Favilla M, De Cecco E (1996) Parallel direction and extent specification of planar reaching arm movements in humans. Neuropsychologia 34:609–613
Favilla M, Hening W, Ghez C (1989) Trajectory control in targeted force impulses. VI. Independent specification of response
amplitude and direction. Exp Brain Res 75:280–294
Favilla M, Gordon J, Hening W, Ghez C (1990) Trajectory control
in targeted force impulses. VII. Independent setting of amplitude and direction in response preparation. Exp Brain Res
79:530–538
Fisher BE, Winstein CJ, Velicki MR (1996) Evidence for feedforward processing in the control of rapid aiming when time to
program is constrained. Soc Neurosci Abstr 22:2041
Fisher BE, Winstein CJ, Velicki MR (1998) Amplitude scaling in
rapid aiming movements is impaired after unilateral sensorimotor area damage. Soc Neurosci Abstr 24:1663
Fu QG, Suarez JI, Ebner TJ (1993) Neuronal specification of direction and distance during reaching movements in the superior precentral premotor area and primary motor cortex of monkeys. J Neurophysiol 70:2097–2116
Fu QG, Flament D, Coltz JD, Ebner TJ (1995) Temporal encoding
of movement kinematics in the discharge of primate primary
motor and premotor neurons. J Neurophysiol 73:836–854
Georgopoulos AP, DeLong MR, Crutcher MD (1983) Relations
between parameters of step-tracking movements and single
cell discharge in the globus pallidus and subthalamic nucleus
of the behaving monkey. J Neurosci 3:1586–1598
Georgopoulos AP, Crutcher MD, Schwartz AB (1989) Cognitive
spatial-motor processes. 3. Motor cortical prediction of movement direction during an instructed delay period. Exp Brain
Res 75:183–194
Ghez C, Hening W, Favilla M (1989) Gradual specification of response amplitude in human tracking performance. Brain Behav Evolut 33:69–74
Haaland KY, Harrington DL (1989) Hemispheric control of the
initial and corrective components of aiming movements. Neuropsychologia 27:961–969
374
Haaland KY, Harrington DL (1990) Complex movement behavior:
toward understanding cortical and subcortical interactions in
regulating control processes. In: Hammond GE (ed) Cerebral
control of speech and limb movements. Elsevier Science,
Amsterdam, pp 169–200
Haaland KY, Harrington DL, Yeo R (1987) The effects of task
complexity on motor performance in left and right CVA patients. Neuropsychologia 25:783–794
Hening W, Favilla M, Ghez C (1988a) Trajectory control in targeted force impulses. V. Gradual specification of response amplitude. Exp Brain Res 71:116–128
Hening W, Vicario D, Ghez C (1988b) Trajectory control in targeted force impulses. IV. Influences of choice prior experience
and urgency. Exp Brain Res 71:103–115
Kalaska JF, Caminiti R, Georgopoulos AP (1983) Cortical mechanisms related to the direction of two-dimensional arm movements: relations in parietal area 5 and comparison with motor
cortex. Exp Brain Res 51:247–260
Kim SG, Ashe J, Hendrich K, Ellermann JM, Merkle H, Ugurbil
K, Georgopoulos AP (1993) Functional magnetic resonance
imaging of motor cortex: hemispheric asymmetry and handedness. Science 261:615–617
Kitamura JI, Shibasaki H, Takagi A, Nabeshima H, Yamaguchi A
(1993) Enhanced negative slope of cortical potentials before
sequential as compared with simultaneous extensions before
sequential as compared with simultaneous extensions of two
fingers. Electroencephalogr Clin Neurophysiol 86:176–182
Kunesch E, Binkofski F, Steinmetz H, Freund HJ (1995) The pattern of motor deficits in relation to the site of stroke lesions.
Eur Neurol 35:20–26
Kurata K (1993) Premotor cortex of monkeys: set- and movementrelated activity reflecting amplitude and direction of wrist
movements. J Neurophysiol 69:187–200
Lee RG, Donkelaar P van (1995) Mechanisms underlying functional recovery following stroke. Can J Neurol Sci 22:257–263
Lemon RN (1993) Stroke recovery. Curr Biol 3:463–465
Megaw ED (1972) Direction and extent uncertainty in step-input
tracking. J Mot Behav 4:171–186
Netz J, Lammers T, Homberg V (1997) Reorganization of motor
output in the non-affected hemisphere after stroke. Brain
120:1579–1586
Palmer E, Ashby P, Hajek VE (1992) Ipsilateral fast corticospinal
pathways do not account for recovery in stroke. Ann Neurol
32:519–525
Pohl PS, Winstein CJ, Onla-or S (1997) Sensory-motor control in
the ipsilesional upper extremity after stroke. Neurorehabilitation 9:67–69 [Corrigendum, Neurorehabilitation 9:245–249]
Rao SM, Binder JR, Bandettini PA, Hammeke TA, Yetkin FZ,
Jesmanowicz A, Lisk LM, Morris GL, Mueller WM, Estkowski
LD (1993) Functional magnetic resonance imaging of complex
human movements. Neurology 43:2311–2318
Riehle A, Requin J (1989) Monkey primary motor and premotor
cortex: single-cell activity related to prior information about
direction and extent of an intended movement. J Neurophysiol
61:534–549
Roland PE, Larsen B, Lassen NA, Skinhoj E (1980) Supplementary motor area and other cortical areas in organization of voluntary movements in man. J Neurophysiol 43:118–136
Shibasaki H, Sadato N, Lyshkow H, Yonekura Y, Honda M,
Nagamine T, Suwazono S, Magata Y, Ikeda A, Miyazaki M
(1993) Both primary motor cortex and supplementary motor
area play an important role in complex finger movement.
Brain 116:1387–1398
Tartaglione A, Inglese ML, Bandini F, Spadavecchia L, Hamsher
K, Favale E (1991) Hemisphere asymmetry in decision making abilities. Brain 114:441–1456
Turton A, Wroe A, Trepte N, Fraser C, Lemon RN (1996) Contralateral and ipsilateral EMG responses to transcranial magnetic
stimulation during recovery of arm and hand function after
stroke. Electroencephalogr Clin Neurophysiol 101:316–
328
Vanman E (1992) The timed response paradigm for interface with
an arm manipulandum [computer program]. USC Movement
Performance and Learning Laboratory Department of Biokinesiology, Los Angeles
Velicki MR, Winstein CJ, Altman K, Pohl PS (1993) Trajectory
parameter specification in subjects post stroke. Soc Neurosci
Abstr 19:546
Weiller C, Chollet F, Friston KJ, Wise RJS, Frackowiak RSJ
(1992) Functional reorganization of the brain in recovery from
striatocapsular infarction in man. Ann Neurol 31:463–472
Weiller C, Ramsay SC, Wise RJS, Friston KJ, Frackowiak RSJ
(1993) Individual patterns of functional reorganization in the
human cerebral cortex after capsular infarction. Ann Neurol
33:181–189
Winstein CJ, Pohl PS (1995) Effects of unilateral brain damage on
the control of goal-directed hand movements. Exp Brain Res
105:63–174
Winstein CJ, Grafton ST, Pohl PS (1997) Motor task difficulty and
brain activity: an investigation of goal-directed reciprocal aiming using positron emission tomography. J Neurophysiol 77:
1581–1594
Winstein CJ, Merians A, Sullivan K (1999) Motor learning after
unilateral brain damage. Neuropsychologia 37:975–987
Wyke M (1967) Effect of brain lesions on the rapidity of arm
movement. Neurology 17:1113–1120
Wyke M (1971) The effects of brain lesions on the performance of
bilateral arm movements. Neuropsychologia 9:33–42