Bimanual Training After Stroke: Are Two
Hands Better Than One?
Dorian K. Rose and Carolee J. Winstein
Functional recovery of the paretic upper extremity eludes the majority of patients post stroke. Although many tasks
require the coordinated participation of both hands, rehabilitation strategies for the most part have focused on the paretic
limb. This article reviews the behavioral basis of bimanual coordination both in health and after stroke hemiparesis and
reviews clinical research studies that have used a bimanual training protocol for rehabilitation. Our intent is to examine
and evaluate the evidence for the application of such an approach to enhance recovery of upper extremity function.
Based on our review, we suggest a set of prerequisite task features and patient characteristics for consideration in the
application of bimanual training protocols for poststroke rehabilitation. Key words: bimanual, hemiparesis, motor control,
rehabilitation, stroke, upper extremity
F
unctional recovery of the paretic upper
extremity post stroke continues to be one of
the greatest challenges faced by rehabilitation professionals. Although most patients regain
walking ability, 30%–66% of stroke survivors fail
to regain functional use of their arm and hand.1
Upper extremity rehabilitation protocols for stroke
hemiparesis have traditionally focused primarily
on the paretic limb with unilateral strengthening
exercises, neuromuscular reeducation, and/or functional training.2–4 One recent approach, constraintinduced movement therapy, exploits this focus by
physically constraining the nonparetic limb with a
sling or safety mitt.5,6 Many daily tasks, however,
naturally require the coordinated participation of
both hands; this provides a rationale for the incorporation of bimanual movements into upper limb
rehabilitation protocols. A small but growing number of investigations have examined the efficacy of
bilateral training on the recovery of paretic limb
movements post stroke.7–12 This article outlines the
conceptual framework underlying bimanual motor
control, reviews both the bimanual motor control
and therapeutic intervention literature as applied to
persons after stroke, and from this suggests several
guidelines concerning patient characteristics and
task considerations to maximize the potential benefit from bimanual training protocols for rehabilitation post stroke.
The Nature of Bimanual Motor Control
Remarkably, most of the goal-directed move20
ments in which humans engage emerge from the
brilliantly orchestrated and coordinated actions of
both hands. Examples from everyday tasks range
from the seemingly routine tasks of catching a ball,
kneading flour dough, or tying shoelaces to the
artistry and skill of playing the piano or communicating with sign language. This coordinated behavior appears effortless in most cases, suggesting
that the masterful controller, the central nervous
system (CNS), is constrained by a set of rules that
naturally reduce this complex coordinated control
problem.13 However, in other cases, the coordination of the two upper limbs for goal achievement
presents a challenge for the CNS, which in turn,
provides an intriguing experimental paradigm for
motor control studies of these actions. Although
the majority of systematic investigations of this
topic have been conducted during the past 25
years, Woodworth made the first observations
over 100 years ago: “It is common knowledge that
movements with the left and right hand are easy
Dorian K. Rose, PhD Candidate, PT, was a doctoral student
in the Department of Biokinesiology and Physical Therapy,
University of Southern California, Los Angeles, California,
when this manuscript was written.
Carolee J. Winstein, PhD, PT, FAPTA, is Associate
Professor, Department of Biokinesiology and Physical
Therapy, Department of Neurology, Keck School of Medicine,
and is Director of the Laboratory of Motor Behavior and
Neurorehabilitation at the University of Southern California,
Los Angeles, California.
Top Stroke Rehabil 2004;11(4):20–30
© 2004 Thomas Land Publishers, Inc.
www.thomasland.com
Bimanual Training After Stroke
to execute simultaneously. We need hardly to try at
all for them to be nearly the same.”14(p97)
Various taxonomies have been used to characterize motor tasks. One useful taxonomy defines tasks
with discrete events (e.g., tasks that have a definite
beginning and end) or continuous (e.g., tasks
without recognizable events such as a beginning
or end). Tasks can also be described by both the
temporal and spatial features. Experimental paradigms for the study of bimanual coordination have
used both types of tasks (continuous and discrete)
and have examined interlimb coordination in both
temporal and spatial domains. What follows is not
an exhaustive review of this literature, but rather,
a limited but focused synopsis of research that can
provide a reasoned rationale for how, with whom,
and why a bimanual training protocol might be
included in a comprehensive rehabilitation program for those with stroke hemiparesis.
There is strong evidence from systematic investigations of healthy individuals that performance
in unimanual compared to bimanual aiming tasks
is not the same.15–19 Results for unimanual or symmetrical bimanual aiming movements concur with
predictions from Fitts’ Law,20,21 that is, movement
time increases as movement amplitude or precision requirements increase. However, for bimanual asymmetric goal-directed aiming, Fitts’ Law is
not supported. When one limb must move further
or aim toward a smaller target compared to the
other, the behavior of one hand is affected by the
task requirements of the contralateral hand. This
phenomenon is commonly known as the assimilation effect.18 Assimilation effects are usually asymmetric in nature such that the limb performing the
more difficult task impacts the limb performing
the easier task to a greater extent than the converse. For example, using a bimanual Fitts’ aiming
paradigm, movement time of the limb moving to
the easy (i.e., large) target is prolonged (compared
to its unimanual performance) and therefore is
similar to the movement time of the contralateral
limb aimed to the difficult (i.e., small) target.
A naturalistic modification to the bimanual Fitts’
paradigm occurs in the case of stroke hemiparesis.
After a unilateral stroke, the bimanual constraint
is body-centered with the disparate movement
ability of the two limbs (i.e., paretic, nonparetic),
21
rather than task-centered with disparate target
precision requirements; this offers a unique level
of complexity and clinical relevance to the study
of bimanual coordination.
Bimanual Motor Control After Stroke
Six investigations of bimanual coordination of
participants with adult onset hemiparesis have
appeared in the literature within the past 10
years22–26 with one exception.27 A behavioral analysis of two individuals post stroke appeared over
50 years ago.27 Cohn recorded arm movements
from a patient who moved each arm together or
independently in a cyclic and continuous pronation–supination motion. When the arms moved
together, the movement frequency of the nonparetic limb was slowed dramatically compared
to unilateral performance. In addition, although
not as striking but certainly more intriguing, the
movement trajectory of the paretic limb improved
in frequency and regularity in the bilateral compared to the unilateral condition. Cohn27 did not
quantify these results, but the qualitative images
provided a provocative impression of the potential of bilateral movements for the facilitation of
paretic limb movements after stroke.
More recently, Rice and Newell used a continuous, temporally symmetric22 and temporally
asymmetric23 elbow flexion and extension action
with 18 poststroke individuals. For the temporally symmetric task, participants were asked to
maintain a “comfortable speed” oscillation, initially determined for each limb separately. In the
bilateral condition, the nonparetic limb was not
able to achieve its unilateral frequency but was
constrained to the slower paretic limb frequency.
For the temporally asymmetric task, participants
were to oscillate one limb at twice the frequency
of the other. Those with stroke were unable to
perform this task but instead adopted an in-phase
movement pattern consisting of bimanual flexion
and extension together, at the same rate. Lewis
and Byblow24 examined interlimb temporal and
spatial coordination with a continuous circledrawing task in nine poststroke hemiparetic individuals. Similar to the Rice and Newell study,22,23
the paretic limb influenced the behavior of the
22
TOPICS IN STROKE REHABILITATION/FALL 2004
nonparetic limb and no improvements in the
hemiparetic limb were elicited with the continuous bimanual task.
Dickstein and colleagues25 and Cunningham
and colleagues26 investigated bimanual coordination with a spatially symmetric discrete motor
task. Both of these studies examined movements
about the elbow joint. Dickstein and her collaborators reported a prolonged movement time for
the nonparetic limb during bilateral elbow flexion
compared to the unilateral condition. Although
there was an increase in paretic limb movement
time as well (11%), in the bilateral condition,
movement time for the nonparetic limb was significantly greater (18%). Cunningham and colleagues showed some facilitation of the paretic
limb with a smoother elbow extension velocity
profile during bimanual movements. Their data
also suggested that loading the nonparetic limb in
the bimanual condition may facilitate the paretic
limb’s movements.
A parallel can be drawn between the studies of
healthy adults in which the hand aiming to the
easy target slows down to a speed similar to that of
the hand aiming to the difficult target15–19 and the
results of patients with unilateral paresis in whom
the nonparetic limb slows to a speed similar to
that of the paretic limb.22,25 In both scenarios,
whether the disparate demands on the system
were external (unequal-sized targets) or internal
(central paresis), the limb with the shorter movement time in the unimanual condition adopts a
longer movement time that is similar to that of
the slower contralateral limb. This systematic
regulation of movement time could result from an
implicit constraint for the two hands to achieve
simultaneous target impact (i.e., temporal synchrony constraint).
From a clinical perspective, the results from
Cohn27 and Cunningham and colleagues26 are
particularly interesting: Both reported a smoother
kinematic profile for the paretic limb when it was
moved simultaneously with the nonparetic limb
compared to unimanual movement. (Cohn reported forearm supination and pronation, qualitative
description only; Cunningham and co-workers
reported elbow extension, quantitative description.) When the nonparetic limb is constrained to
slow down, there appears to be some facilitation
of paretic limb movement.
The investigations discussed above are limited
to single joint movements, specifically about the
elbow22,23,25,26 or forearm,27 or use closed-chain,
non-goal-directed tasks.24 In contrast, and outside
the laboratory, humans usually display multiarticular upper limb movements that are purposeful and goal directed. Recently, we developed
and have used two different goal-directed aiming
paradigms in a series of experiments designed to
investigate interlimb coordination after unilateral
stroke.28–30 The results of this work are described
in the next section.
We were particularly interested in capturing the
temporal and spatial adjustments of both limbs
in spatially symmetric and asymmetric multijoint
aiming movements. In three separate experiments,
we conducted a systematic examination of the
unilateral and bilateral movements of 30 individuals post stroke (18 left cerebral vascular accident,
12 right cerebral vascular accident) who exhibited
mild motor impairment (mean Fugl-Meyer motor
score 60 ± 5) with intact sensation at the shoulder,
elbow, and wrist. We compared their performance
to that of 30 age-matched healthy control participants.
Our initial study examined a spatially symmetric, forward aiming movement. In response to an
LED signal, participants were to reach forward
rapidly and aim with one hand (unimanual) or
both hands (bimanual) to hit a switch(es) mounted on the LED target, as soon and as fast as possible.28 The movement required shoulder flexion
and horizontal adduction with elbow extension
and forearm supination.
Similar to previous reports,22,25 the nonparetic
limb exhibited a prolonged movement time in the
bimanual compared to the unimanual condition.
Kinematic analysis revealed that the primary locus
for this prolongation was the deceleration phase
of the movement. This adaptive response by the
nonparetic limb allowed for a nearly simultaneous
(both limbs) target impact in 81% (417/512) of
the bimanual trials.
Compared to the unimanual condition, the nonparetic limb exhibited a lower peak resultant velocity (PRV) in the bimanual condition. Conversely,
Bimanual Training After Stroke
compared to the unimanual condition, the paretic
limb exhibited a higher PRV in the bimanual condition (Figure 1). This disassociation of limb and
condition was observed for the stroke group but
not the control group. When differences between
movement conditions are described with movement duration alone, it appears that nonparetic
limb movement time is dictated primarily by the
slower paretic limb. However, an examination of
the kinematics reveals facilitation of PRV with the
paretic limb in the bimanual condition that is not
apparent from the movement time data alone.
This facilitation effect is similar to that reported
by Cunningham and colleagues for their elbow
extension paradigm.26
We followed this first investigation with one in
which we combined the internal, body-centered
asymmetry (unilateral paresis) with an external,
target-centered asymmetry (targets located at different distances) in the same experiment.29 As in
the symmetric paradigm, in response to an LED
signal, participants were to reach and aim forward with one hand (unimanual) or both hands
(bimanual) to hit a switch(es) mounted on the
LED target, as soon and as fast as possible.
For the bimanual condition, the limbs moved to
separate targets. One was a “near” target and the
other, a “far” target; the former target was located
50% of the distance in the fore/aft plane from the
participant as the latter. The unilateral paresis and
asymmetric target location afforded two bimanual
asymmetric aiming combinations: (a) nonparetic
limb-far target/paretic limb-near target, and (b)
paretic limb-far target/nonparetic limb-near target. We defined the first combination that pairs
the paretic limb performing an easy task with
the nonparetic limb performing a difficult task as
congruent in that the two task-limb combinations
are more similar in task difficulty level. This is in
contrast to the second combination, which pairs
the paretic limb performing a difficult task with
the nonparetic limb performing an easy task. We
define this aiming condition as incongruent in that
the two task-limb combinations are more dissimilar in level of difficulty.
Our primary finding was that the temporal constraint on bimanual movements was robust even
with the addition of an external task asymmetry
23
to the unilateral stroke-induced paresis. The nonparetic limb prolonged movement execution time
when paired with the paretic limb, compared to
that in the unimanual condition. Furthermore, the
nervous system demonstrated adaptability, with
the magnitude of nonparetic limb prolonged movement time being a function of paretic limb aiming
distance. The increase in nonparetic limb movement time was greater when it was paired with the
paretic limb aiming to the far target than when the
paretic limb was aiming to the near target.
Given the increase in paretic limb PRV in the
bimanual condition observed in our first experiment,28 we were interested to see if this result was
evident in the second experiment that used an
asymmetric paradigm. Enhanced paretic limb PRV
was evident only in the incongruent condition in
Experiment 2. Paretic limb PRV was greater in
the bimanual compared to unimanual condition
only when the paretic limb aimed a far distance
while the nonparetic limb aimed a near distance.
No such difference was found with the congruent
aiming condition (Figure 2).
To determine if aiming distance alone was the
factor that contributed to the enhanced paretic
limb PRV, we compared paretic limb PRV in the
unimanual aiming conditions for the near and far
aiming tasks. Group mean PRV was not different
for the two unimanual aiming conditions; thus,
it appears that both paretic limb aiming distance
and the constraints of bimanual coordination contribute to the enhanced group mean PRV in the
incongruent condition.
Results from this experiment provide further
insight regarding potential requisite task conditions for the facilitation of the paretic limb in
bimanual aiming. Our findings suggest that it
may not be the bimanual task alone that affords
this facilitation; task requirements for the paretic
limb may also be important. Enhanced PRV with
bimanual aiming only occurred when the paretic
limb aimed a far distance. However, there was no
difference in unimanual PRV for near compared
to far aiming, which provides evidence that it is
not only aiming distance that underlies the larger
paretic limb PRV but the bimanual nature of the
task is also requisite for the facilitation effect.
The aforementioned studies focused exclusively
24
TOPICS IN STROKE REHABILITATION/FALL 2004
A. Paretic Upper Extremity
B. Nonparetic Upper Extremity
C. Bimanual Paretic vs. Nonparetic
Figure 1. Ensemble average time series of instantaneous velocity plotted from movement onset to offset. Figures A–C show data from a single stroke participant. (A) Paretic limb profiles for unimanual (UNI;
solid line) and bimanual (BI; dashed line) aiming. Peak resultant velocity (PRV) is greater in bimanual
compared to unimanual aiming. (B) Nonparetic limb profiles for unimanual (solid line) and bimanual
(dashed line) aiming. PRV is greater in unimanual compared to bimanual aiming. (C) Bimanual aiming
profiles of paretic (solid line) and nonparetic (dashed line) limbs plotted together. Paretic limb PRV is
greater than nonparetic limb PRV. Error bars are standard deviation. (Adapted from Rose D, Winstein C.
The coordination of bimanual rapid aiming movements following stroke. Clin Rehabil. In press.)
Bimanual Training After Stroke
A. Congruent Aiming
B. Incongruent Aiming
Figure 2. Line graphs of group means by aiming condition and limb for peak resultant
velocity (PRV). Dashed line indicates paretic;
solid line indicates nonparetic. Error bars are
standard error. (A) Congruent aiming: no difference in paretic limb PRV between unimanual
and bimanual aiming. (B) Incongruent aiming:
paretic limb PRV greater in bimanual compared
to unimanual aiming (p
p < .05).
25
on temporal coordination between the two limbs.
Data from healthy individuals suggest that the
two limbs also exhibit spatial coordination.31,32
Therefore, the third experiment in our series examined both the temporal and spatial coordination of
bimanual aiming and employed a unilateral barrier
paradigm.30
From a seated position and in response to an
LED signal, participants moved both hands away
from midline, laterally and in the frontal plane, to
hit switches mounted on targets, as soon and as
fast as possible. The movement required shoulder
abduction with elbow extension. A vertical barrier
(10-, 15-, and 20-cm high) was placed midway
between the home and target position and only
on one side. The task was adapted from previous
work of Kelso and colleagues17 and Goodman
and co-workers.33 We captured both movement
time and movement displacement of both limbs
and quantified the temporal and spatial coupling,
respectively, between the limbs and groups (stroke
and matched control).
Regression analysis was used to quantify the
interlimb coupling strength. For each participant, nonbarrier limb movement time (60 trials)
or maximum vertical displacement (60 trials)
was regressed on the corresponding barrier limb
dependent measure to determine the strength
of temporal and spatial interlimb coupling.
Nonbarrier limb temporal and spatial behaviors
were predicted significantly by performance of the
contralateral barrier limb. Both temporal (control,
R2 = .61 ± .04; stroke, R2 = .44 ± .03) and spatial
(control, R2 = .62 ± .04; stroke, R2 = .31 ± .04)
coupling were weaker for the stroke group compared to the control group. However, of potential
clinical interest was the result that between-limb
temporal coupling (R2 = .44 ± .03) was more
robust than spatial coupling (R2 = .31 ± .04) to the
central motor paresis.
An analysis of the velocity profile (Figure 3)
corroborates the theory that for bimanual movements the CNS specifies a single temporal structure for both limbs. It is of interest and clinical
relevance to note that the paretic limb, in the role
of nonbarrier limb, mirrors the temporal structure
of the contralateral barrier limb. The normal system’s temporal constraint on the two limbs during
26
TOPICS IN STROKE REHABILITATION/FALL 2004
Stroke Participant
despite a pronounced slowing of the nonparetic
limb’s movements in the bimanual condition.
Future research should attempt to exploit this
temporal constraint by entraining the nonparetic
limb to its unimanual movement time or even
a proportion of that duration during bimanual
tasks.
Bimanual Intervention Paradigms for Upper
Extremity Rehabilitation After Stroke
Figure 3. Single trial time series of instantaneous
resultant velocity plotted from movement onset
to movement offset from a representative stroke
participant. Solid line is barrier (nonparetic)
limb, dashed line is nonbarrier (paretic) limb.
Velocity profile of paretic limb mirrors that of the
nonparetic limb.
bimanual movements persists in those with mild
motor hemiparesis. Could this persistent temporal constraint be harnessed as a rehabilitation
strategy?
The studies reviewed above are instructive about
what features of a bimanual task may be used to
facilitate paretic limb movements. Cunningham
and colleagues suggest that loading of the nonparetic limb may be beneficial.26 Our data provide
evidence that peak velocity of the paretic limb
can be facilitated through interlimb effects from
the nonparetic limb for rapid aiming tasks. It is of
interest and, to some extent, counterintuitive that
in these cases a paretic limb facilitation occurs
Within the last 4 years, six published reports
have appeared that employed bimanual intervention protocols for upper extremity rehabilitation
post stroke. Two of these involved an intervention that used a custom-designed arm-training
machine and required repetitive rhythmical shoulder flexion with elbow extension and shoulder
extension with elbow flexion timed to an auditory cue.7,8 A third also used a custom-designed
apparatus to enable bilateral passive and active
motion of forearm pronation and supination and
wrist flexion and extension.9 Two other studies
involved functional reaching, grasping, and placing of various objects.10,11 The sixth study was an
EMG-triggered neuromuscular stimulation program.12 These studies are briefly reviewed in the
next section.
Chronic stroke participants (mean Fugl-Meyer
Upper Extremity Motor Score = 15 ± 2 out of
66) in Whitall and colleagues’ study7 performed
rhythmic bilateral shoulder flexion with elbow
extension on a custom-designed arm-training
machine for four 5-minute bouts, three times a
week for 6 weeks. Pre- versus posttest comparisons revealed decreased arm impairment (FuglMeyer Upper Extremity Motor Performance Test),
increased arm function (Wolf Motor Function
Test), increased daily use (University of Maryland
Arm Questionnaire for Stroke), and increased
isometric strength of the paretic arm elbow and
wrist flexors. These improvements were sustained
8 weeks after cessation of the training.
This same group of investigators report interesting evidence that bilateral proximal exercise may
positively influence distal fine motor control.8
Improvements in paretic finger function in two
of four participants occurred after the same type
Bimanual Training After Stroke
of training protocol described above. The lack of
a control group in both of these studies and the
small number of participants in the latter, however, limit the interpretation of the results.
Hesse and colleagues9 enrolled 12 severely
involved chronic stroke participants with moderate flexor spasticity to perform passive or active
assistive forearm pronation and supination or
wrist flexion and extension with a computerassisted arm trainer for fifteen, 15-minute sessions
over 3 weeks, in addition to participation in an
ongoing comprehensive rehabilitation program.
Wrist and finger spasticity decreased in 8 participants and volitional motor ability improved
for 5 of the 12 participants after the intervention.
Although the intervention did not lead to functional motor changes in this severely involved
cohort, participants reported that the intervention
eased hand hygiene and relieved spasticity-related
pain. As with the previous two studies discussed,
these investigators did not use a control group.
Mudie and Matyas10 report observational kinematic improvements in paretic limb performance
of three functional tasks requiring proximal and
distal arm musculature—placing a block on a
shelf at arm’s length and shoulder height, drinking
from a glass, and transporting a peg from the table
to place the peg on the undersurface of a shelf set
at eye height—after bilateral isokinematic training
(spatiotemporally identical movements performed
bilaterally but with each limb independently). This
type of training was superior to unilateral training
of the paretic limb, bilateral training where the
stronger limb guided the paretic limb through the
movement, and bilateral training where the two
limbs performed different movements.
In contrast, Byblow and Lewis11 did not find
superiority for a bimanual compared to a unimanual training intervention in a small crossover design study of six individuals post stroke.
Although some participants did improve on some
tasks with bimanual training, these results were
not consistent across tasks and participants, leading to no overall differences between the two
interventions.
Finally, Cauraugh and Kim12 examined the efficacy of a unilateral versus bilateral EMG-triggered
neuromuscular stimulation program that targeted
27
the wrist and finger extensors. EMG-triggered
stimulation was applied to the impaired limb to
assist with wrist and finger extension while either
the unimpaired limb remained at rest (unilateral training) or while the unimpaired wrist and
fingers actively extended simultaneously with
the paretic limb (bilateral training). Participants
completed three sets of 30 trials on four separate
days over 2 weeks. Outcome measures at the
post test of paretic limb reaction time, sustained
muscle contraction capability, and number of
blocks moved in a functional task all favored the
bilateral movement training group. There was no
follow-up testing, so the long-term effects of this
intervention are not known.
Overall, there appears to be at least a short-term
beneficial effect for bilateral training protocols,
although the results are not overwhelming. As
with any intervention, a bimanual approach may
not be the most appropriate for all individuals
post stroke, nor may it be appropriate throughout
the course of rehabilitation. As one can tell by a
review of this small number of published studies,
bimanual intervention is a relatively generic term
and can encompass a wide range of therapeutic
protocols. One may simply not be able to conclude that a bimanual intervention is or is not
beneficial. It will be important to qualify such a
statement by describing the specific nature of that
intervention, what is meant by an improvement,
and the population to whom it is directed.
Implications for Rehabilitation
Although these studies report paretic limb
improvement with a bimanual intervention
approach, in each instance the critical feature(s) of
the intervention underlying these improvements
is not clear. The two literature sets reviewed—one
focused on the nature of bimanual motor control,
the other focused on investigations of efficacy for
bimanual training protocols to enhance recovery
after stroke—exist somewhat in isolation of one
another. There has yet to be a hypothesis-driven
test of the critical features inherent to bimanual
coordination that are thought to enhance paretic
limb movements in the application of bimanual
training protocols.
28
TOPICS IN STROKE REHABILITATION/FALL 2004
As with any therapeutic intervention, it would
help rehabilitation professionals to know if individuals of a certain level of stroke severity, age,
time post stroke, side of hemispheric lesion, or
lesion location would most benefit from bimanual
training protocols. This list represents only a sample of the myriad patient characteristics that could
influence rehabilitation outcomes. Again, because
the numbers of studies conducted are few, it is not
possible to generalize the results. However, we can
begin to draw some speculative conclusions from
the evidence that does exist.
Although neither the motor control nor intervention studies reviewed point to the role of
sensation in successful coordination of bimanual
actions, it should be highlighted. Proprioceptive
information is vital for controlling naturalistic
movements involving multiple joints34 and is
considered to play an even larger role in bimanual coordination. Jackson and colleagues35,36 have
argued that a sensorimotor mechanism, based
upon proprioceptive coding of limb position and
motion, exists to maintain interlimb coordination
during movement execution. Although untested
to date, this would suggest that intact proprioception is a critical prerequisite for beneficial effects
with bimanual training protocols.
In the first two experiments from our laboratory,
we describe a group average increase in paretic
limb PRV in the bimanual condition. A closer look
revealed that 63% of participants demonstrated
this increase in the symmetrical aiming experiment
and 59% in the asymmetrical aiming experiment.
A subgroup analysis revealed a trend (although
not statistically significant) for those who demonstrated this enhanced paretic limb PRV to have
slightly lower Fugl-Meyer motor scores (59 ± 5),
indicative of a more impaired upper limb, relative
to those who did not exhibit this enhancement
(62 ± 5). This result will require corroboration
before a definitive conclusion can be made, but
it is suggestive that impairment level may be an
important factor to consider in establishing inclusion criteria for appropriate application.
Related to our findings, Lewis and Byblow11
concluded from their intervention study of 6
individuals post stroke that in most of the tasks
performed by the more well-recovered patients
there was no real additional beneficial effect of
bilateral practice. When a positive influence of the
bilateral intervention was suggested, it tended to
be in tasks with lower performance scores for participants with moderate motor deficits. They also
note that the task that had the largest involvement
of proximal musculature also had the more reliable facilitatory effects. Given the contribution of
bilateral descending pathways to proximal musculature,37 movements requiring activation of these
proximal muscles may profit most from bilateral
training protocols.
Cunningham and colleagues describe a facilitation of the paretic limb during bilateral elbow
extension, but this group effect was only observed
when the oldest participant was removed from the
data set. Five of the six study participants demonstrated paretic limb facilitation in the bilateral
condition, only the eldest participant did not.26
There may be age-related deficits in the ability to
engage in dual-task performance, but again this
merits further investigation.
McCombe-Waller and Whitall8 report that two
of four participants with preservation of distal
hand function demonstrated improved paretic
finger function after bilateral arm-based training. These two had lesions affecting their motor
dominant hemispheres, and the two who had
more equivocal results had lesions to their nondominant hemispheres. Although no definitive
conclusions can be made given this small sample
size, there is the suggestion that a lesion to the
dominant hemisphere may provide an advantage
in recovery from stroke with bilateral training.
Exchange of information from one to the other
side of the brain and spinal cord is possible via
several commissural fiber systems. Of particular
interest, the brain structure most implicated in
the coordination of bimanual tasks is the rich
callosal interconnection between the supplementary motor hand areas38 that is significantly larger
than the sparse callosal interconnection between
the hand areas of the primary motor cortices.39
Although earlier studies implicated the supplementary motor area in bimanual coordination,40–42
it is now widely accepted that the locus of bimanual control is more distributed than relegated to a
single brain structure or even brain region.
Bimanual Training After Stroke
The control of bimanual movements is not allocated to one neuroanatomical structure, but rather
appears to be the result of a distributed network
involving both cortical and subcortical areas.43–46
Wiesendanger and Serrien have concluded, “It
begs credulity to imagine that one particular brain
structure could control the diversity of bimanual
activities.”47(p501) Lesion location alone, therefore,
may not be very useful in predicting who will or
will not benefit from bimanual training protocols.
Given the distributed nature of bimanual coordination, we can also conclude that the majority
of our patients will manifest deficits in bimanual
coordination, and, therefore, bimanual training
activities should be at least a part of any comprehensive rehabilitation program.
These studies provide insight as to who may
benefit from bilateral training protocols, but these
observations were made post hoc. None of the
studies were designed to specifically answer this
question. We can conclude with confidence that
inclusion of a bimanual training protocol did not
result in any decline in the participants’ movement abilities or any reported adverse events.
Numerous previous investigations of bimanual
coordination for neurologically intact adults have
reported extreme variability in the degree to which
individuals coordinate their two limbs; in some
cases, the two limbs are influenced by one another
quite readily, and for other individuals, the two
29
limbs operate quite independently of one another.15–19,30,48 For example, in our asymmetric barrier
experiment, strength of bimanual coupling for our
control participants, as quantified by regression
analysis, ranged from 0.0, indicative that none
of the variance in nonbarrier limb behavior was
explained by the barrier limb, to 0.96, indicative
that 96% of the variance in the nonbarrier limb
was explained by the barrier limb (group mean =
0.61 ± .03).30 This relatively high between-subject
variability for neurologically normal individuals
should not be forgotten when the effectiveness of
bilateral training protocols are evaluated for individuals after unilateral stroke.
Bimanual coordination for functional tasks is
ubiquitous in everyday life, yet has not received a
great deal of attention as a modality for poststroke
rehabilitation. The focus has been either on teaching the individual how to use their nonparetic
arm to compensate for the weakened arm or on
strengthening and functional tasks for the paretic
arm alone. We have provided evidence that, even
after a hemispheric stroke, bimanual movements
retain a similar underlying structure to that seen
in the healthy participant. There is an entire
class of actions that has been virtually ignored by
rehabilitation specialists, a class of actions that,
if practiced properly, may indeed have beneficial
effects on upper limb functional recovery after a
hemispheric stroke.
REFERENCES
1. Kwakkel G, Lollen GB, Wagenaar RC. Therapy
impact on functional recovery in stroke rehabilitation: a critical review of the literature. Physiotherapy.
1999;13:457–470.
2. Butefisch CM, Hummelsheim H, Denzler P, Mauritz
KH. Repetitive training of isolated movements
improves the outcome of motor rehabilitation
of the centrally paretic hand. J Neurolog Sci.
1995;130:59–68.
3. Sunderland A, Fletcher D, Bradley L, Tinson D,
Hewer RL, Wade DT. Enhanced physical therapy
for arm function after stroke: a one-year follow-up
study. J Neurol Neurosurg Psych. 1994;57:856–858.
4. Davies PM. Steps to Follow. New York: SpringerVerlag; 1985.
5. Morris DM, Crago JE, DeLuca SC, Pidikiti RD,
Taub E. Constraint-induced movement therapy for
motor recovery after stroke. NeuroRehabilitation.
1997;9:29–43.
6. Taub E, Uswatte G, Pidikiti R. Constraint-induced
movement therapy: a new family of techniques with
broad application to physical rehabilitation – a clinical review. J Rehabil Res Dev. 1999;36:237–251.
7. Whitall J, Waller SM, Silver KHC, Macko RF. Repetitive
bilateral arm training with rhythmic auditory cueing improves motor function in chronic hemiparetic stroke. Stroke. 2000;31:2390–2395.
8. McCombe-Waller S, Whitall J. Fine motor control in
adults with and without chronic hemiparesis: baseline comparison to nondisabled adults and effects
of bilateral arm training. Arch Phys Med Rehabil.
2004;85:1076–1083.
9. Hesse S, Schulte-Tigges G, Konrad M, Bardeleben
A, Werner C. Robot-assisted arm trainer for the
passive and active practice of bilateral forearm and
wrist movements in hemiparetic subjects. Arch Phys
Med Rehabil. 2003;84:915–920.
10. Mudie HM, Matyas TA. Can simultaneous bilateral
movement involve the undamaged hemisphere
in reconstruction of neural networks damage by
30
TOPICS IN STROKE REHABILITATION/FALL 2004
stroke? Disabil Rehabil. 2000;22:23–37.
11. Lewis G, Byblow W. Neurophysiological and behavioral adaptations to a bilateral training intervention in individuals following stroke. Clin Rehabil.
2004;18:48–59.
12. Cauraugh J, Kim S. Two coupled motor recovery protocols are better than one. Stroke. 2002;33:1589–
1594.
13. Bernstein N. The Coordination and Regulation of
Movements. Oxford, England: Pergamon Press; 1967.
14. Woodworth RS. Le mouvement. Paris: Doin; 1903.
15. Kelso JAS, Southard DL, Goodman D. On the coordination of two-handed movements. J Exp Psych.
1979;5:229–238.
16. Kelso JAS, Southard DL, Goodman D. On the
nature of human interlimb coordination. Science.
1979;203:1029–1031.
17. Kelso JAS, Putnam CA, Goodman D. On the spacetime structure of human interlimb coordination. Q
J Exp Psych. 1983;35A:347–375.
18. Marteniuk RG, MacKenzie CL, Baba DM. Bimanual
movement control: information processing and interaction effects. Q J Exp Psych. 1984;36A:335–365.
19. Fowler B, Duck T, Mosher M, Mathieson B. The
coordination of bimanual aiming movements: evidence for progressive desynchronization. Q J Exp
Psych. 1991;43A:205–221.
20. Fitts PM. The information capacity of the human
motor system in controlling the amplitude of
movement. J Exp Psych. 1954;47:381–391.
21. Fitts PM, Peterson JR. Information capacity of discrete
motor responses. J Exp Psych. 1964;67:103–112.
22. Rice MS, Newell KM. Interlimb coupling and left
hemiplegia because of right cerebral vascular accident. Occup Ther J Res. 2001;21:12–28.
23. Rice MS, Newell KM. Upper-extremity interlimb
coupling in persons with left hemiplegia due to
stroke. Arch Phys Med Rehabil. 2004;85:629–634.
24. Lewis GN, Byblow WD. Bimanual coordination
dynamics in poststroke hemiparetics. J Motor
Behavior. 2004;36:174–186.
25. Dickstein R, Hocherman S, Amdor G, Pillar T.
Reaction and movement times for patients with
hemiparesis for unilateral and bilateral elbow flexion. Phys Ther. 1993;73:374–385.
26. Cunningham CL, Phillips-Stoykov ME, Walter CB.
Bilateral facilitation of motor control in chronic
hemiplegia. Acta Psychologica. 2002;110:321–337.
27. Cohn R. Interaction in bilaterally simultaneous
voluntary motor function. Arch Neurol Psychiatry.
1951;65:472–476.
28. Rose DK, Winstein CJ. The coordination of bimanual symmetrical rapid aiming movements following
stroke. Clin Rehabil. In press.
29. Rose DK, Winstein CJ. The coordination of bimanual
asymmetric rapid aiming movements: the effect of
diminished central drive. Manuscript in preparation.
30. Rose DK, Winstein CJ. Temporal and spatial constraints in a spatially asymmetric task: the effect of
diminished central drive. Manuscript in preparation.
31. Franz EA, Zelaznik HN, McCabe G. Spatial topolog-
ical constraints in a bimanual task. Acta Psychologia.
1991;77:137–151.
32. Franz EA. Spatial coupling in the coordination
of complex actions. Q J Exp Psychol. 1997;50A:
684–704.
33. Goodman DA, Kobayashi RB, Kelso JAS. Maintenance
of symmetry as a constraint in motor control. Can
J Appl Sport Sci. 1983;8:238.
34. Ghez C, Sainburg R. Proprioceptive control of
interjoint coordination. Can J Physiol Pharmacol.
1995;73:273–284.
35. Jackson GM, Jackson GM, Kritikos A. Attention
for action: coordinating bimanual reach-to-grasp
movements. Br J Psychol. 1999;90:247–270.
36. Jackson GM, Jackson SR, Husain M, Harvey M,
Kramer T, Dow L. The coordination of bimanual
prehension movements in a centrally deafferented
patient. Brain. 2000;123:380–393.
37. Martin JH. Neuroanatomy, 2nd ed. Stamford, CT:
Appleton and Lange; 1996.
38. Rouiller EM, Babalian A, Kazennikov O, Moret V, Yu
XH, Wiesendanger M. Transcollosal connections of
the distal forelimb representations of the primary
and supplementary motor cortical areas in macaque
monkeys. Exp Brain Res. 1994;102:227–243.
39. Jenny AB. Commissural projections of the cortical motor hand area in monkeys. J Comp Neurol.
1979;188:137–146.
40. Tanji J, Okano K, Sato KC. Relation of neurons in
the nonprimary motor cortex to bilateral hand
movement. Nature. 1987;327:618–620.
41. Tanji J, Okano K, Sato KC. Neuronal activity in cortical motor areas related to ipsilateral, contralateral,
and bimanual digit movements of the monkey. J
Neurophysiol. 1988;60:325–343.
42. Brinkman C. Supplementary motor area of the
monkey’s cerebral cortex: short-and long-term deficits after unilateral ablation and the effects of subsequent callosal section. J Neurosci. 1984;4:918–929.
43. Donchin O, Gribova A, Steinberg H, Bergman H,
Vaadia E. Primary motor cortex is involved in bimanual coordination. Nature. 1998;395:274–278.
44. Kermadi I, Liu Y, Rouiller EM. Do bimanual motor
actions involve the dorsal premotor, cingulated
and posterior parietal cortices? Comparison with
primary and supplementary motor cortical areas.
Somatosens Motor Res. 2000;17:255–271.
45. Debaere F, Swinnen SP, Beatse E, Sunaert S, Van
Hecke P, Duysens J. Brain areas involved in interlimb
coordination: a distributed network. NeuroImage.
2001;14:947–958.
46. Swinnen SP. Intermanual coordination: from behavioural principles to neural-network interactions.
Nature Rev Neurosci. 2002;3:350–361.
47. Wiesendanger M, Serrien DJ. The quest to understand bimanual coordination. In: Progress in Brain
Research. North-Holland: Elsevier; 2004:491–505.
48. Sherwood DE. The coordination of simultaneous
actions. In: Wallace SA, ed. Perspectives on the
Coordination of Movement. North-Holland: Elsevier;
1989:303–327.