University of Southern California Dissertations and Theses

Hemisphere-specific deficits in the control of bimanual movements after stroke

(USC Thesis Other)

Hemisphere-specific deficits in the control of bimanual movements after stroke

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

HEMISPHERE-SPECIFIC DEFICITS
IN THE CONTROL OF BIMANUAL MOVEMENTS AFTER STROKE

BY

RINI VARGHESE

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
in Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOKINESIOLOGY)

August 2021

Copyright 2021 Rini Varghese
ii
DEDICATION

To my husband, Liju: your unconditional love and support made this possible.
And to our soon-to-be-born first child: you were with me when I wrote this in the Spring of 2021.

iii
ACKNOWLEDGEMENTS

In my first Division Seminar in 2017, I shared a quote from Advice for a Young
Investigator (Eng. Translation, L. Swanson, 1999) by renowned neuroanatomist and Nobel prize
winner Santiago Ramón y Cajal: “Tackle small problems first, so that if success smiles and
strength increases one may then undertake the great feats of investigation.” I feel proud that in
my 6 years in the Biokinesiology Doctor of Philosophy Program, I have tackled many small
problems. Sometimes I succeeded, other times I did not; nevertheless, I persevered and learned,
and my strength increased. This journey would have been impossible without the mentorship of
some remarkable BKN Faculty who served on my committee and the unconditional love, care
and support of my family and friends.
I want to thank Dr. Carolee Winstein, my primary advisor, who was my model for good
communication and pushed me to develop the skills for successful collaborations, even within
my own committee. She constantly reminded me of the human aspect of stroke; newly equipped
with my quantitative tools and in my zeal to use them, I would sometimes forget that these were
a means to an end, that at the other end of those investigative tools is a thinking, feeling human.
This is a lesson I will carry with me into the future. She encouraged my creative side and gave
me the intellectual freedom to think and read far outside my area and comfort zone. In service of
that, she let me borrow many books from her for years on end (some of which I only returned
recently!). She trusted me and had confidence in me, choosing me to co-author my first
perspective paper with her (even though I wondered at that time why anyone should care about a
novice’s opinion on any subject). She always valued my input and gave me the confidence so
that I could foster my curiosity and ask questions freely. She reminded me that perfection is the
enemy of good and that the best things result from small but consistent efforts. She was my
iv
constant support and cheerleader and provided me with many opportunities for growth and
reward—cross-country collaborations, funding, and time to attend conferences and workshops,
and recommendations for grants and awards. She was not only supportive of my service
activities, but actively encouraged me to extend my sphere of action and leadership. No matter
where she was, she would find time for our weekly meeting, and, for the last two years, I enjoyed
being her only (PhD) child in the lab…I got all her attention and I think that spoiled me!
I want to thank Dr. James Gordon, who was always generous with his time, despite his
administrative responsibilities, and reiterated regularly that the time he spends with students
brings him joy. Through those meetings, I have gathered countless nuggets of wisdom, such as:
every movement, no matter how simple, complete with its turns and bends, has a story, to never
underestimate how patients will find ways to defy the ‘laws’ of movement control, and how that
is a good thing because if the system were not capable of adapting and changing, studying it and
tailoring physical therapy approaches after it would be moot. He introduced me to the history of
movement science and reminded me that the best way to move the science forward is not simply
to dismiss older theories but to reconcile them with new empirical data. He fostered having a
healthy skepticism of all theories, and, when warranted, to modify them to account for new
observations; and, when doing so, to show the evidence rather than simply state it. I will always
remember that my research will have the greatest impact if I continue to think of movement
problems as an analytical clinician and communicate complex scientific theories intuitively to
my clinical and research peers. These lessons I will cherish and incorporate in the years to come.
I want to thank Dr. Nicolas Schweighofer, who taught me the first and last thing I know
about computational modeling of motor systems, especially my recent venture into Optimal
Control Models. He was patient even when it took me unimaginably long, whether it was re-
v
learning what he called “high school-level” linear algebra or understanding complex concepts of
control engineering. Naturally, it might have taken me longer than his Engineering students to
pick up some concepts, but he never gave up on me. He would give me the space to work out my
problems at my speed but be available to answer my questions (even late into the night!). In that
same spirit, he gave me some of my earliest opportunities to get trained on computational
techniques: to take his Spring class on Modeling the Motor Systems and attend his 2016 Summer
School on Computational Motor Control and Learning in the beautiful countryside of
Montpellier (I was the only Physical Therapist with 18 or 19 other engineers or computer
scientists that year!). I have greatly benefited from his teaching philosophy of “learning by
doing,” and, as a result, I never shied away from a challenge. He was always honest in his
criticisms of my papers and challenged me to aim higher by adopting a better statistical
methodology when there was one, to raise the current standards where possible. He welcomed
me like one of his own lab members and gave me opportunities to attend and present papers at
CNRL meetings. Like many other BKN students, he taught me almost everything I know about
statistical approaches and that every statistical method is just a general linear model in disguise.
In recognition of these lessons, there will be no more t-tests or ANOVAs for me for the rest of
my life!
I want to thank Dr. Sook-Lei Liew, who helped me think critically about neuroimaging
data, especially as it applies to the stroke population. She exposed me to the world of Open
Science and related tools, including GitHub and public neuroimaging databases, while advising
me to be mindful of their drawbacks. She was approachable, warm, and available to me for
advice both in-person and on Slack. I also want to thank her lab group, the NPNL, for letting me
present my Aim 3 DTI findings and giving me feedback at the early stages of my analysis.
vi
I want to thank Dr. Jason Kutch, who helped me think clearly and communicate simply.
This became evident in our collaboration on my first observational study of this Dissertation and
the drafting process of the resulting paper. He also invited me to attend the Summer series of the
AMPL meetings in 2018.
In addition to these wonderful advisors, I want to acknowledge the mentorship of Dr.
Robert Sainburg for his personal investment in my growth as a movement scientist long before
he became a co-sponsor on my F31 and perhaps even before I entered the PhD program. In that
capacity, he hosted me twice in his labs in Hershey and State College. He was instrumental in
helping me understand the complexity of hemisphere-specific control and shaping my thinking in
the early years of my dissertation. He was available by FaceTime as we set up and calibrated the
KineReach device in the Bice Center. He goes out of his way to care for his students. I will never
forget that he bought vegan cheese and made me my very own pizza at the holiday party he
hosted in December 2019 during my time at State College. His passion for science is contagious,
and his understanding of the history and science of cerebral lateralization is deep and inspiring.
As a whole, my mentors found ways to support me individually and together, and I would
like to think that I am the product of their different styles and collaborative thinking. Apart from
their direct contributions to this dissertation, I think the relationships that I have cultivated with
my mentor team is priceless. Together, they showed me by example that, even though the
problems I tackled seemed small and inconsequential, by doing my part honestly and diligently, I
would have contributed to the great feats of investigation that inform our collective
understanding and move the field forward. I will strive to make them proud.
vii
I want to thank my colleagues in BKN who have become dear friends, especially Natalia
Sánchez for her informal mentorship, and past and present members of MBNL for their support
and for being there in big and small ways whenever I needed them.
Thank you to the stroke survivors who volunteered their time for my pilot and final
experiments and teaching me valuable lessons through these precious interactions.
I owe a mountain of gratitude to my husband Liju Varghese. His selfless love and support
have lifted me up at my lowest points, and his devotion to my cause is an active force in my
growth. I am incredibly lucky to have a partner whose values align with my own and thankful for
the generosity of his love. He is my greatest inspiration. Thanks to his as yet unfailing principle
of: “I follow Rini wherever she goes!”, we will continue our journey together, and that’s good
because without him, I would have no place to call home.
My parents-in-law, dad, Peter Varghese, and mom, Lilly Varghese, my amma (mother),
Lata Matthews for being with me every step of the way. They stormed some pretty bad weather
with me and prided themselves in my accomplishments. Their strength and encouragement has
helped me get to this point.
Lastly, I would like to acknowledge my funding sources. My time at USC was supported
by funding from the Division of Biokinesiology (2015-2019) and by an F31 National Research
Service Award from the Eunice Kennedy Shriver NICHD of the National Institutes of Health
(2019-2021: F31HD098796). I am proud to be the first recipient of the prestigious F31 award
from our Division, but, having interacted with the new pool of talented students who came after
me, I am certain that I will not be the last.

viii
TABLE OF CONTENTS
DEDICATION ...................................................................................................................... ii
ACKNOWLEDGEMENTS ................................................................................................... iii
LIST OF TABLES ................................................................................................................ x
LIST OF FIGURES ............................................................................................................. xi
ABSTRACT...................................................................................................................... xviii
CHAPTER 1: General Background ....................................................................................... 1
Clinical Significance .....................................................................................................................1
Motor lateralization: evidence from unimanual and bimanual movements ...................................2
Early theoretical models and paradigms to study bimanual control .............................................5
Limitations in the early theoretical models ...................................................................................7
Current models of bimanual motor control ..................................................................................8
CHAPTER 2: Dissertation Overview .................................................................................... 13
CHAPTER 3: The probability of choosing both hands depends on an interaction between
motor capacity and limb-specific control in chronic stroke ................................................... 15
Abstract ...................................................................................................................................... 15
Introduction................................................................................................................................ 17
Methods ...................................................................................................................................... 20
Results ........................................................................................................................................ 27
Conclusion .................................................................................................................................. 37
Supplements ............................................................................................................................... 39

ix
CHAPTER 4: Relationship between motor capacity of the contralesional and ipsilesional hand
depends on the side of stroke in chronic stroke survivors with mild-to-moderate impairment . 41
Abstract ...................................................................................................................................... 41
Introduction................................................................................................................................ 43
Methods ...................................................................................................................................... 46
Results ........................................................................................................................................ 49
Discussion ................................................................................................................................... 54
Conclusion .................................................................................................................................. 59
Supplements ............................................................................................................................... 60
CHAPTER 5: Corpus callosal microstructure predicts bimanual motor performance in
chronic stroke survivors ....................................................................................................... 61
Abstract ...................................................................................................................................... 61
Introduction................................................................................................................................ 63
Methods ...................................................................................................................................... 67
Results ........................................................................................................................................ 76
Conclusion .................................................................................................................................. 89
Supplements ............................................................................................................................... 90
CHAPTER 6: Flexibility of responsibility assignment for a redundant bimanual task is limited
to some extent by lateralized motor control processes ........................................................... 95
Abstract ...................................................................................................................................... 95
Introduction................................................................................................................................ 97
Discussion ................................................................................................................................. 120
Conclusion ................................................................................................................................ 123
Supplements ............................................................................................................................. 124
APPENDIX ....................................................................................................................... 136
Strong temporal coupling but no interlimb interference observed during discrete asymmetric
bimanual reaching. ................................................................................................................... 136
BIBLIOGRAPHY .............................................................................................................. 139
x
LIST OF TABLES

Table 3.1. Participant characteristics. 27

Table 3.2. Standardized coefficients ± SE from nested mixed-effects logistic
regression.
29

Table 3.3. Average movement times for stroke and able-bodied control
participants. Standardized coefficients from nested mixed-effects
logistic regression in stroke survivors only.
31

Table 4.1. Descriptive Statistics for the full sample (N = 42), and for the two
groups of interest, left hemisphere damage, LHD (n = 21) and right
hemisphere damage, RHD (n = 21).
49

Table 5.1. Subject characteristics. 77

Table 5.2. Robust mixed-effects regression coefficients from Model (1) to
estimate relationship between bimanual movement time and mean
fractional anisotropy (FA) for five segmented regions of the CC
relative to the reference group (CC3, motor).
79

Table 6.1. Participant characteristics. 103

xi
LIST OF FIGURES

Figure 1.1. Line diagram of the Optimal Feedback Controller. The model consists
of an external feedback loop that relies on afferent sensory feedback
(y) and an internal feedback loop that relies on predicted states (x
𝑡 *).
The differentiating features of an optimal controller are: 1) the
intelligent use of feedback in updating of the control policy, L, known
as the minimum intervention principle, and 2) the integration of
external and internal feedback to form the state estimate (x ̂
𝑡 ).
9

Figure 3.1. Photo illustration of the four tasks performed by a chronic stroke
survivor (LHD, UEFM = 51). Left panel shows the two letter-
envelope composite task: folding the letter and inserting in an
envelope. Right panel shows the two photo-album composite task:
receiving the photo album and inserting photo in album sleeve.
22

Figure 3.2. Raw data (1 = bimanual, 0 = unimanual) and model-fitted
probabilities (Pr.) of bimanual choice for left- (LHD, blue) and right-
hemisphere damage (RHD, red) groups. Logistic model fits A. across
all four tasks (n = 200), B. letter-envelope composite task (n = 100),
and C. photo-album composite task (n = 100). Dashed line
corresponds to the mean UEFM score (z = 0) and the intercept of the
model fit.
30
xii

Figure 3.3. Average log-transformed movement times A. across all four tasks (n
= 200), B. letter-envelope composite task (n = 100), and C. photo-
album composite task (n = 100). Non-disabled control participants
(CTRL) were significantly faster compared to chronic stroke
survivors who chose a bimanual strategy. As predicted, there were no
differences in movement time between those stroke survivors who
chose a unimanual compared to a bimanual strategy. *** p < 0.001
32

Figure 4.1. Hypothesized effects represented in schematic figure. (A) The null
hypothesis, wherein the relationship between contralesional (CL)
impairment and ipsilesional (IL) motor capacity is not modified by
the side of stroke lesion, i.e., 𝛽 1
≠ 0 but 𝛽 3
= 0. (B) Alternative
hypothesis 1, wherein ipsilesional deficits are related to contralesional
impairment but only in LHD (blue) and not in RHD (red). (C)
Alternate hypothesis 2, wherein ipsilesional deficits are related to
contralesional impairment but only in LHD and in RHD with severe
impairment (represented in the shaded dark-grey area). For both
alternate hypotheses, 𝛽 1
and 𝛽 3
≠ 0.
45

Figure 4.2. Scatterplots show the relationship between contralesional motor
impairment (CL UEFM) and ipsilesional distal motor performance
(IL dWMFT) for (A) the full sample, (B) LHD, and (C) RHD. Solid
51
xiii
lines represent the linear prediction and shaded areas represent the
95% confidence interval. For ease of interpretation, rounded estimates
of raw values (in seconds for dWMFT and points for UEFM) have
been provided in green in Panel A. Asterisks indicate values
evaluated as outliers.

Figure 4.3. Scatterplots show relationship between contralesional distal motor
performance (CL dWMFT) and ipsilesional distal motor performance
(IL dWMFT) for (A) the full sample, (B) LHD, and (C) RHD. Solid
lines represent the linear prediction and shaded areas represent the
95% confidence interval. For ease of interpretation, rounded estimates
of raw values (in seconds) have been provided in green in Panel A.
Asterisks indicate values evaluated as outliers.
53

Figure 5.1. a) Diffusion pipeline including preprocessing, co-registration and
identifying region of interest (i.e., corpus callosum). (b) Parcellation
of the corpus callosum using the geometric scheme proposed by
Hofer & Frahm (2006). (c) Table showing cortical regions connected
by the fibers running through each of the five CC parcels.
69

Figure 5.2. Overlap of lesions for 41 chronic stroke survivors. Note that images
were not flipped and represent the actual side of unilateral stroke.
71

xiv
Figure 5.3. Callosal FA for each of the five regions plotted against log-
transformed movement times for the two bimanual tasks. Population
regression lines from robust mixed effects regression are shown.
80

Figure 5.4. Model estimated marginal means for CC FA across the five regions
along with individual data points. * p < 0.05, ** p < 0.01, *** p <
0.001.
81

Figure 6.1. Task setup and conditions. 104

Figure 6.2. A. Single canonical example trial showing paths of left (blue) and
right (red) hands during one-cursor and two-cursor conditions.
Starting positions of the hands were shifted by a fixed offset for
display purpose only. Also, note that the two-cursor condition has two
targets and two start circles. B. Scatter plot of mean-centered X
positions for the two hands across all participants for the one- and
two-cursor conditions. Line of equivalence shown in black. Ellipses
contain 95% of the points in the binormal distribution. Notice that due
to the differences in starting position of the two hands, mean of the
distribution of points in each condition is shifted. C. Distribution of
correlation coefficients across all young adult participants (n = 20)
showing that between-hand correlation was significantly more
negative in the one-cursor compared to the two-cursor condition.
110
xv
Collectively, these plots demonstrate that position along the
horizontal (X) axis show task-dependent compensation between
hands only in the one-cursor but not the two-cursor condition.

Figure 6.3. A. Example trial showing measure of directional error. We
determined the initial directional error of the cursor (at peak
acceleration) and the change in direction of each hand in the early
phase of movement, i.e., initial direction of the movement (at peak
acceleration) to later in the movement (at peak velocity). B.
Contribution of each hand to cursor’s initial directional error for one
non-disabled right-handed control participant. As evident here, slope
was significantly more negative for the left hand compared to the
right hand. C. Scatter plot with population-level regression lines
(solid) and individual regression lines (n = 20) for the left and right
hands in young controls. D. Individual slopes derived from random
effects from the full mixed-effects model.
112

Figure 6.4. A. Right hemiparesis (RHP) gains: Scatter plot with population-level
regression lines (solid) and individual regression lines (n = 12) for the
left and right hands in chronic stroke survivors with right
hemiparesis. Adjacent boxplots represent individual slopes derived
from random effects from the full mixed-effects model showing
reversal in correction gains, which were larger (more negative) for the
115
xvi
paretic right hand. B. Left hemiparesis (LHP) gains: Scatter plot with
population-level regression lines (solid) and individual regression
lines (n = 11) for the left and right hands in chronic stroke survivors
with left hemiparesis. Adjacent boxplots represent individual slopes
derived from random effects from the full mixed-effects model.
Paretic left hand in individuals with left hemiparesis did not show the
same increase in correction gains. C. Asymmetry index: A
comparison of gain asymmetry index, gasym (
|𝑔 𝑟𝑖𝑔 ℎ𝑡 |
|𝑔 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑔 𝑙𝑒𝑓𝑡 |
) showing
a strong left-ward bias in controls (green). Compared to controls,
balance between hands shifts to the right hand in individuals with
right hemiparesis (as expected) as well as in those with left
hemiparesis (unexpected finding). D. Scatterplot and second-order
non-linear fit (gasym ~ uefm + uefm
2
) shows a weak relationship
between the degree of motor impairment in the paretic extremity and
correction gain asymmetry index.

Figure 6.5. A. Boxplots comparing directional error at peak velocity between
controls and stroke survivors. B. Boxplots comparing peak speed
between controls and stroke survivors. Paretic hand in stroke groups
is indicated by labels. *** p< 0.001, ** p<0.01
117

xvii

Figure 6.6. A. Average ACF coefficients of the last 16 trials for each hand across
participants in each group. B. Asymmetry index: A comparison of
relative asymmetry in adaptation rates given by, rasym (
|𝑟 𝑟𝑖𝑔 ℎ𝑡 |
|𝑟 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑟 𝑙𝑒𝑓𝑡 |
)
showing a strong left-ward bias in controls (green). Compared to
controls, balance between hands shifts to the right hand in individuals
with right hemiparesis but not in in those with left hemiparesis. C.
Scatterplot and second-order non-linear fit (rasym ~ uefm + uefm
2
)
shows relationship between the degree of motor impairment in the
paretic extremity and trial-by-trial adaptation rate asymmetry index
(not significant).
119

xviii
ABSTRACT

Stroke continues to be the leading cause of adult disability in the US and worldwide, with
as many as two-thirds of survivors experiencing some degree of contralesional arm and hand
weakness. Conventional rehabilitation practices focus on the recovery of the contralesional arm,
but several recent clinical trials investigating the effectiveness of variants of contralesional arm
training have reported negative or neutral outcomes. Naturally, efforts to recover and rehabilitate
the paretic upper extremity in isolation are of little value if it is not engaged in meaningful
functional activities—activities that are predominantly bilateral in nature, requiring the
coordinated engagement of both the paretic and non-paretic limbs. One alternative to traditional
paretic limb training is bilateral upper extremity training; however, its effectiveness has been
shown to vary by the side of hemispheric lesion. Therefore, the current dissertation seeks to
characterize hemisphere-specific deficits in the control of bimanual movements after stroke. To
do this, I first conducted a series of retrospective observational studies (Chapter 3, 4, 5) of an
existing stroke database from a previous phase-IIb clinical trial (Dose Optimization for Stroke
Evaluation, ClinicalTrials.gov ID: NCT01749358). Then, I collected data in a prospective
experimental study (Chapter 6) to uncover the differences in bimanual coordination observed
between individuals with left and right hemisphere stroke.
Chapter 3 begins with a retrospective observational analysis in which we studied the
factors that influence the spontaneous selection of both hands for bimanual tasks—tasks that
would otherwise, in age-similar able-bodied individuals, naturally elicit the use of both hands. To
capture spontaneous, task-specific choices, we covertly observed 50 pre-stroke right-handed
chronic stroke survivors (25 each of left paresis, and right paresis) and 11 age-similar control
adults and recorded their hand use strategies for two pairs of bimanual tasks with distinct
xix
demands: one with greater precision requirements (photo-album tasks), and another with greater
stabilization requirements (letter-envelope tasks). We found that the probability of choosing a
bimanual strategy is greater for those with less severe motor impairment and in those with right
paresis. However, the influence of these factors, i.e., impairment severity and side of lesion, on
bimanual choice varied based on task demands.
In Chapter 4, we further analyze these differences in spontaneous bimanual use by
examining the relationship between unimanual performance of the upper extremities in a subset
of 42 of the sample of 50 chronic stroke survivors examined in Chapter 3. The purpose of this
retrospective analysis was to test the idea that those with right paresis were more likely to use
both hands together more because their less-affected left hand would be slower on its own. To do
this, we looked at the relationship between the degree of impairment in the contralesional hand,
quantified using the Upper Extremity Fugl-Meyer score and the extent of ipsilesional hand
deficits, quantified by the distal component of the Wolf Motor Function Test. We found that in
those with right paresis, the speed of performance with the ipsilesional hand was proportionally
slower to the degree of contralesional impairment. However, this was not the case in those with
left paresis. This interaction between ipsilesional hand and side of lesion was observed not only
with a measure of contralesional impairment but also contralesional hand function.
In Chapter 5, given the well-established role of the CC for bimanual coordination,
especially fibers connecting the larger sensorimotor networks such as prefrontal, premotor, and
supplementary motor regions, we examine the relationship between the microstructural status of
the CC and bimanual performance in chronic stroke survivors (n = 41). We used movement
times for two self-initiated and self-paced bimanual tasks (quantified in Chapter 3) to capture
bimanual performance. Using publicly available control datasets (n = 52), matched closely for
xx
acquisition parameters, including sequence, diffusion gradient strength and number of directions,
we also explored the effect of age and stroke on callosal microstructure. We found that callosal
microstructure was significantly associated with bimanual performance in chronic stroke
survivors such that those with lower callosal FA were slower at completing the bimanual task.
Notably, while the primary sensorimotor regions (CC3) showed the strongest relationship with
bimanual performance, this was closely followed by the premotor/supplementary motor (CC2)
and the prefrontal (CC1) regions. Furthermore, chronic stroke survivors presented with
significantly greater loss of callosal fiber orientation (lower mean FA) compared to
neurologically intact, age-similar controls, who in turn presented with lower callosal FA
compared to younger controls. The effect of age and stroke were observed for all regions of the
CC except the splenium. These findings suggest that in chronic stroke survivors with relatively
localized lesions, callosal microstructure can be expected to change beyond the primary
sensorimotor regions and might impact coordinated performance of self-initiated and cooperative
bimanual tasks.
Lastly, the purpose of Chapter 6 was to understand the principles of responsibility
assignment in a bimanual task and thereby uncover mechanisms underlying the previously
observed hemisphere-specific effects of stroke. To do this, I used a prospective experimental
design and studied a redundant bimanual task wherein we tested the predictions of a leading
theory in motor control, known as the Optimal Feedback Control model (OFC). The OFC model
suggests that responsibility assignment is a flexible process such that errors are assigned to and
corrected for by the limb that is most likely to produce those errors. In right-handed adults, this is
often the less-skilled, non-dominant left limb. The flexibility of this process can be probed by
examining corrections made by each limb after they have acquired alternative use-dependent
xxi
experiences, e.g., if the left limb became more skilled, experiencing fewer errors or if the right
limb became less skilled, experiencing more frequent errors. Such is the case of stroke affecting
the right side of the body, wherein we would predict that the left limb corrects less while the
paretic right limb corrects more for task error.
In this prospective study, we tested this prediction in 20 individuals with an intact
sensorimotor system, as well as 23 chronic stroke survivors (12 right hemiparesis). Consistent
with previous studies, correction gains were asymmetric between the limbs in the non-disabled
young control group such that the left limb corrected more than the right limb. Our data also
supported our predictions in those with right hemiparesis, but not in those with left hemiparesis.
Those with right hemiparesis not only corrected more with their paretic right limb within a trial,
but also corrected less with their now more skilled left limb. They also systematically adapted to
these errors in a feedforward manner over trials. The extent of correction in stroke survivors did
not appear to vary with the degree of motor impairment of the paretic extremity. These findings
lead us to conclude that responsibility assignment is not entirely flexible but may be limited to
some extent by the hemispheric specialization of motor control processes. Prior to this
experiment, I performed an experiment that used an interlimb interference paradigm, motivated
by early models of bimanual control. The main findings of this experiment are presented in the
Appendix.
Collectively, these studies show that mechanisms of deficits in bimanual coordination
after stroke are distinct between individuals with left and right hemisphere damage. For those
with left paresis (right hemisphere stroke), error detection or correction is impaired, and it may
be why these individuals do not spontaneously use both hands. Large inter-individual variability
might suggest sensitivity to the type of error or different mechanisms for correcting it, however
xxii
we did not explicitly test this idea in this dissertation. Potential rehabilitation strategies might
include accuracy training for these individuals. Conversely, in those with right paresis (left
hemisphere stroke), error assignment and adaptation seem to be intact. In fact, extensive
experience helps the once-less-skilled left limb to calibrate its forward model. However, these
individuals were slower than controls. Therefore, potential rehabilitation approaches might
include speed training for these individuals.

1
CHAPTER 1: General Background
Clinical Significance
Stroke continues to be the leading cause of adult disability in the US and worldwide, with
as many as two-thirds of survivors experiencing some degree of contralesional arm and hand
weakness (Gresham et al., 1975; Nakayama et al., 1994). In addition to contralesional paresis,
stroke survivors exhibit significant motor deficits in the ipsilesional arm and hand (Chestnut &
Haaland, 2008; Wetter et al., 2005; C. J. Winstein & Pohl, 1995). Importantly, these motor
deficits are specific to the damaged hemisphere (Mani et al., 2013b; Schaefer et al., 2007;
Tretriluxana et al., 2009; C. J. Winstein & Pohl, 1995). Given that most functional tasks require
the use of both hands (Bailey et al., 2015a; Kilbreath & Heard, 2005a), this form of bilateral
deficit directly limits functional independence in post-stroke patients. In fact, a number of
previous studies have shown that stroke survivors spend significantly less time using their paretic
limb alone and bilaterally (Bailey et al., 2015a; Michielsen et al., 2012a; Vega-González &
Granat, 2005b). One recent study also reported that arm use is influenced by the side of stroke
lesion, such that individuals with non-dominant (left) limb paresis spend less time engaging in
unilateral paretic and bilateral arm use compared to those with dominant (right) limb paresis
(Rinehart et al., 2009a).
Conventional rehabilitation practices focus on the recovery of the contralesional arm, but
a number of recent clinical trials investigating the effectiveness of variants of contralesional arm
training have reported negative or neutral outcomes (C. Winstein, 2018). However, efforts to
recover and rehabilitate the paretic upper extremity in isolation are of little value if it is not
engaged in meaningful functional activities—activities that are predominantly bilateral in nature,
requiring the coordinated engagement of both the paretic and non-paretic limbs. In the wake of
2
the non-superiority of these existing interventions, there is an urgent need to find alternative
approaches. One alternative approach for which there is preliminary evidence (Maenza et al.,
2021) and that is currently being investigated for stroke survivors with severe contralesional
paresis is hemisphere-specific remedial training of the ipsilesional arm and hand
(NCT03634397).
Another alternative is bilateral upper extremity training. Indeed, a form of bilateral
training, known as the Bilateral Arm Training with Rhythmic Auditory Cueing” or BATRAC,
and other modified versions, have been shown to be a promising intervention for the recovery of
arm and hand function post-stroke. BATRAC was developed based on the phenomenon of
frequency-dependent interlimb coupling and consequent paretic limb entrainment by the non-
paretic limb (Whitall et al., 2000). Simply described, this phenomenon refers to the spontaneous
tendency of the limbs to synchronize and symmetrize movements (more details on this are in the
forthcoming sections). One can instinctively appreciate how such a phenomenon might form a
compelling basis for bilateral training in stroke survivors in whom the weaker limb can be
‘forced’ to move in sync with the less-affected limb. However, its effectiveness too has been
shown to vary by the side of hemispheric lesion. Thus, as has been noted previously (Rose &
Winstein, 2004; C. Winstein & Varghese, 2018) it seems that with regards to the application of
bilateral training for post-stroke rehabilitation, there exists, most decidedly, a “set of specific
patient and task characteristics.”

Motor lateralization: evidence from unimanual and bimanual movements
In the preceding section, we presented evidence to support the idea that with regards to
stroke recovery and rehabilitation, many outcomes of clinical interest, including bimanual use,
3
contralesional and ipsilesional deficits as well as response to alternative forms of therapy such as
bilateral training are influenced by the hemispheric side of lesion.
One explanation for these findings comes from a contemporary model of motor
lateralization known as the Dynamic Dominance Hypothesis (DDH). In contrast to the traditional
view of limb dominance which attributes unimanual preferences for the right hand to its general
superiority for motor skills (in right-hand dominant individuals) and the left hand as a weaker
counterpart, the DDH model suggests that the left hand (primarily under control of the right
hemisphere) is more proficient at stabilization of position through impedance control
mechanisms; whereas the right hand (primarily under the control of the left hemisphere) is better
developed for producing precise movement trajectories through predictive control mechanisms
(R. L. Sainburg, 2002).
Early and critical evidence for DDH, comes from the study of interlimb asymmetries in
healthy neurologically intact adults. For example, by studying unimanual single-joint
movements, Sainburg & Schaefer, 2004, showed that when target distance was systematically
varied, the dominant right arm in non-disabled controls scaled peak velocity by adopting a
strategy in which the peak of the acceleration, i.e., magnitude of the force pulse was increased
(pulse height control). This suggested that the right hand was predominantly controlled in a
predictive, feedforward manner. Conversely, the non-dominant left limb scaled peak velocity by
prolonging the time of acceleration (pulse width control), an indication of greater reliance on
feedback loops. That the two hands produce forces in distinct ways by either modifying the size
or the duration of the pulse is interesting and one that we will revisit in a few sections from here
when I discuss limitations of early models of bimanual control. Interlimb asymmetries have also
been observed for two-joint unimanual movements and imagined unimanual movements
4
(Gandrey et al., 2013). Finally, more causal evidence for DDH has been found through the study
of patients with unilateral brain lesions which found that focal lesions to the left or right
hemisphere produce distinct motor and behavioral deficits that reflect damage to the control
processes for which that hemisphere is specialized (Schaefer et al. 2007, 2009a, 2009b; Mutha et
al. 2011)
Indeed, interlimb asymmetries are best exemplified in bimanual tasks, which entail a
natural division of labor between the hands such that each hand assumes a preferred role (Guiard,
1987). In fact, it has been suggested that each hand is preferentially selected by the nervous
system to assume a role consistent with their proficient controller. For example, Woytowicz and
colleagues (E. J. Woytowicz et al., 2018) reported that switching preferred roles of the left and
right hand during a simulated “bread-cutting” task was in fact non-optimal for motor
performance. By examining relevant characteristics of the movement, they demonstrated that the
right hand showed straighter reaching trajectories but was poor at stabilizing, whereas the left
hand exhibited more stable holding performance, but was poor at reaching. Recently, Schaeffer
and Sainburg (2020) demonstrated that sudden and unexpected braking of each limb in its path
(so as to elicit a feedback-mediated long loop reflex response) as the two limbs jointly controlled
a virtual symmetric object, elicited a strong EMG response in the limb contralateral to the
mechanical perturbation only for the left, but not the right limb (Schaffer & Sainburg, 2021).
Whereas a few recent studies have reported on hemisphere-specific control in bimanual
movements in neurologically intact adults, to date, none have studied the role of hemisphere-
specific mechanisms in bimanual control in stroke survivors, which arguably provides more
causal evidence for the role of each hemisphere in bimanual control.

5
Early theoretical models and paradigms to study bimanual control
The prevailing view of bimanual control in the 70’s was adapted from then well-accepted
mechanisms of interlimb coordination of the lower extremities. Seminal evidence reported by
Brown & Sherrington (1911), and later built upon by Grillner and colleagues (1975), showed
that, in decerebrate cat preparations, spinal circuits constituted lower coordinative structures that
were capable of generating automatic behaviors, such as rhythmic hindlimb movements in cats,
even in the absence of descending motor commands (Brown, 1911, 1914; Grillner, 1975). The
idea that the forelimbs, or upper extremities in humans could be controlled in a similar way, even
if perhaps by now vestigial spinal circuits, was therefore a not-too-distant thought from the
understanding of limb coordination at the time.
The earliest empirical support for this hypothesis came from the behavioral observation
made by Kelso and colleagues, (J. A. Kelso et al., 1979) that when individuals were asked to
perform discrete reaches to bilateral targets of two different ‘difficulties’ (defined by ratio of
target amplitude by width), they tended to reach them with similar movement times (MT),
defying Fitts law for the hand that reached the ‘easier’ target. That the limbs showed some, if not
a perfect, degree of temporal synchrony was a repeated observation in the several experiments
that followed. (Franz et al., 1991, 1996a; J. A. S. Kelso et al., 1983; Marteniuk et al., 1984; Rose
& Winstein, 2004) This strong tendency to move synchronously was termed as interlimb
‘coupling.’ However, anomalous observations of interlimb differences were also reported.
Nonetheless, the phenomenon of interlimb coupling was found to be generalizable across a
variety of tasks (discrete, continuous, rhythmic) and so formed the basis for many experimental
paradigms. The defining feature of such an experimental paradigm was that it relied on the
obligatory nature of interlimb coupling, which would presumably give rise to interference
6
between the limbs if they were tasked with different goals. Therefore, experimental tasks used to
test interlimb coupling required the two hands to perform disparate, even competing tasks, e.g.,
cyclic movements of different frequencies, discrete movements to targets set at different
distances or directions etc. It is worth noting that although the first observations of coupling were
made in the context of discrete reaching, experimental paradigms have since largely shifted to
rhythmic and continuous movements.
Kelso’s interpretation of interlimb coupling was one that favored the theoretical view of
Bernstein (and later Turvey) that movements are programmed not in terms of individual muscle
contractions but rather functional groupings of muscles, i.e., synergies. Because such groupings
are under a common command signal, they would be constrained to act as a single unit. He also
argued that once the brain sends such a command signal, it is left to take its course (self-
organize), regulated largely only by low-level automatisms (J. A. S. Kelso & Schöner, 1988)—a
view that had the flavor of earlier theories of locomotor control although with less well-defined
physiology. The exact origins of such a signal continue to remain elusive, but more recent
evidence suggests that supraspinal centers of control, such as the cerebral hemispheres and the
cerebellum may have a role in timekeeping and sequencing but such a signal does not appear to
be impervious to internal and external feedback especially in tasks requiring some degree of
spatial accuracy (Diedrichsen et al., 2010). In fact, as we will discuss in the next section, a
significant drawback of Kelso’s early model of movement coordination, later formulated as the
Haken-Kelso-Bunz model (Haken et al., 1985), is that it does not explain response to
unanticipated errors in the movement.

7
Limitations in the early theoretical models
The idea that the various muscles of the two hands are organized as functional groupings
that act as a single unit under a common command signal encounters certain challenges: First, it
implies that the hands meet the different spatial requirements of movements by varying the
magnitude of muscle activations (a control strategy referred to earlier as pulse height control)
rather than the temporal duration of activations (i.e., pulse width control), which remains largely
invariant. However, there is evidence that a pure pulse height control strategy does not account
for 100% of the variance in spatial accuracy (Gordon & Ghez, 1987), but that there is an
independent influence of errors on the control strategy to meet the accuracy requirements of the
task. Thus, to adjust for errors in the initial impulse, there is a reduction in the height of the
impulse and a prolongation of its duration. Furthermore, if in fact as reported by one study (R. L.
Sainburg & Schaefer, 2004), these control mechanisms differ by limb, then it seems that muscles
of the two hands could not be under a single control signal.
A second issue, along the lines of the previous point, is that a single command signal that
is entirely pre-programmed and allowed to take its course, regulated only by low-level
automatisms, does not sufficiently account for internal (generated by the body) and external
(caused by the environment, such as perturbations) errors. It is now well-established that to
generate accurate movements, there must exist a mechanism for error feedback. Thus, errors
resulting from inaccurate motor planning, execution, or environmental instability and
perturbations, are better accounted for through a system that incorporates feedback control.
Physiologically, long-loop reflexes have been thought to serve this purpose.
The last issue with the aforementioned model is its fundamental assumption that coupling
is obligatory and forms a hard constraint that cannot be overcome. Numerous studies have
8
provided evidence that through cognitive strategies such as conceptually organizing the
‘independent’ goals, providing longer preparation times, direct cueing (rather than requiring
sensory transformation of cues), modifying attentional loads, and clear instructions, any
interference between limbs can be entirely overridden to successfully perform a large repertoire
of everyday bimanual tasks. Therefore, even if some degree of temporal synchronization is
observed between the limbs, it does not necessarily lead to interference.

Current models of bimanual motor control
To address the issues stated above, one considers a framework that incorporates error
feedback to adjust the control policy. The idea of feedback-mediated control, e.g., through a
transcortical servo mechanism has existed for a long time. However, feedback is slow and noisy
and by extension ‘expensive’. A system that is entirely peripherally feedback-depedent will
suffer from instability from oscillations around the desired behavior. Thus, a controller that
utilizes internal feedback of predicted states and integrates it with external feedback, to
judiciously alter the control policy would be ideally suited to deal with issues of delays, cost, and
noise.
Recently, a computational form of such a feedback controller that takes into account a
weighted cost of both the task error and muscle commands, known as the optimal feedback
control (OFC) theory, was proposed by Todorov & Jordan, 2002. A schematic of the model is
presented in Figure 1 below. This model uses a stochastic optimal controller and weighs the
different costs associated with a control policy to allow flexible use of feedback, such that it is
engaged only when task-relevant variance increases—a principle of the OFC model known as
‘minimum intervention.’ Whereas the OFC model has been found to be highly generalizable,
successfully explaining a large multitude of motor behaviors, the underlying neurophysiology of
9
the components of the OFC are still not clear. However, new research that uses focal
deactivations of neural regions by cooling shows promise in uncovering the putative circuits
underlying OFC (Takei et al., 2021). Recent studies have used the OFC model to explain
behavior in redundant bimanual tasks (more specific background and literature review is
presented later). However, notable interlimb asymmetries have been observed in healthy young
adults (Diedrichsen, 2007; Diedrichsen & Dowling, 2009; Mutha & Sainburg, 2009; Schaffer &
Sainburg, 2021; White & Diedrichsen, 2010).

Figure 1.1. Line diagram of the Optimal Feedback Controller. The model consists of an external feedback
loop that relies on afferent sensory feedback (y) and an internal feedback loop that relies on predicted
states (x
𝑡 *). The differentiating features of an optimal controller are: 1) the intelligent use of feedback in
updating of the control policy, L, known as the minimum intervention principle, and 2) the integration of
external and internal feedback to form the state estimate (x ̂
𝑡 ).
10
The role of the corpus callosum in mediating bimanual control
Although critical interactions occur at multiple levels of the nervous system, including
the spinal and supraspinal centers, we will consider the role of the corpus callosum, and examine
its integrity as it relates to bimanual performance in chronic stroke survivors in an attempt to
uncover putative neural structures that mediate bimanual control.
Causal role of the CC in bimanual coordination
Foundational evidence in support of the causal role of CC in interlimb interactions comes
from early investigations in callosotomy patients. In two experiments, Franz and colleagues
(Franz et al., 1996a) demonstrated that unlike controls with an intact corpus callosum, patients
who had undergone callosotomy were able to successfully ‘decouple’ their hands to draw
dissimilar spatial patterns, exhibiting little to no interference. In fact, in previous work, Franz
found that even for spatially distinct patterns of drawing (e.g., circles and lines), the two hands
exhibited marked temporal synchrony of movements.
Years later, Eliassen et al. (2000) provided a further layer of resolution into Franz’s
findings by studying patients who had undergone partial resection of the callosum, which
allowed him to dissociate the functions of anterior and posterior callosum. He found that for self-
initiated tapping movements, patients who had undergone anterior callosotomy were
significantly slower and the synchronicity of movement initiations with both limbs were more
imprecise than in patients who had undergone posterior callosotomy, and concluded that the
anterior CC may have a role in mediating interhemispheric communication of self-referential
cues (such as proprioceptive information acquired from the other limb’s movement) whereas the
posterior CC may mediate the transfer of information related to sensory cues external to the
individual (such as visual stimuli to initiate movements). More support for the role of the anterior
11
callosum for temporal synchrony came from the study of individuals with congenital absence of
the callosum, also known as callosal agenesis.

Structural neuroimaging methods to examine the relationship between CC and bimanual
performance in neurologically intact humans and stroke survivors
Evidence provided in the previous section suggests that an intact callosum acts as a
cortical-level structural constraint when the limbs are tasked with competing spatial goals.
However, when partially intact, the callosum allows necessary transfer of information, and that
different regions of the callosum show sensitivity to the type of information being relayed.
Fortunately, recent advances in structural and functional imaging afford unprecedented
opportunities to approach the study of the callosum non-invasively. In the last two decades, there
have been plenty of investigations of non-disabled younger and older adults that shows that
bilateral prefrontal regions are involved in higher-order planning and response selection (Baxter
et al., 2000; Rowe et al., 2000), and, the premotor and supplementary motor regions, are
involved in temporal sequencing (Hagmann et al., 2006b) and bimanual coordination (Debaere et
al., 2004; Wenderoth et al., 2004; O. Donchin et al., 1998; Opher Donchin, 1999; Van Den Berg
et al., 2010)
Another structural approach that is especially useful for the study of white matter
microstructure is diffusion tensor imaging (DTI). Based on the principles of Brownian motion,
DTI is uniquely suited for imaging structures that restrict the free diffusion of water, such as
organized bundles of myelinated nerve fibers or tracts (see Hagmann et al., 2006 for a
comprehensive review). For example, using DTI metrics, Gooijers et al, showed that in the
absence of augmented visual feedback, bimanual task performance was significantly correlated
with FA of the callosal fibers connecting not only the motor but also the occipital cortices
12
(Gooijers et al., 2013b). DTI methods have also been successfully applied to clinical populations,
including multiple sclerosis and traumatic brain injury (TBI). In young adults with and without
TBI, Caeyenberghs and colleagues showed that the FA of callosal fibers in the prefrontal,
sensory and parietal regions were most predictive of reaction times related to switching between
different coordination modes of a bimanual task (Caeyenberghs et al., 2011b).
Diffusion tensor imaging of the descending motor tracts (i.e., CST) is becoming
increasingly popular for its application as a prognostic biomarker in the subacute and chronic
phase after stroke. In recent years, its utility to study structural effects of transcallosal diaschisis
and reorganization in humans is also emerging. However, most of these studies were limited to
the study of the motor regions of the CC and associations with paretic limb impairment and
function. Yet, to date, there are no investigations that associate callosal microstructure with
bimanual performance in chronic stroke survivors. It is plausible that poor microstructural status
of the callosum, especially fibers that connect constituent regions of the larger sensorimotor
network, might lead to delays in transcallosal transfer of information and slow performance on
tasks that require swift exchange of internal and external sensory cues like bimanual tasks.

13
CHAPTER 2: Dissertation Overview
This dissertation begins by uncovering differences in clinical and behavioral outcomes
such as bimanual use (Chapter 3) and the degree of ipsilesional deficits (Chapter 4) between
chronic stroke survivors with left- and right- hemisphere damage who have mild to moderate
motor impairments. With a focus on bimanual performance quantified in Chapter 3 and in an
attempt to uncover putative neural structures that mediate bimanual control, I then analyze the
role of the corpus callosum, specifically its microstructural status, and examine its integrity as it
relates to bimanual performance (movement time) in chronic stroke survivors (Chapter 5). In this
chapter, I also compare callosal microstructure in stroke survivors to that in controls by
analyzing publicly available diffusion datasets.
After these retrospective, clinico-behavioral and structural neuroimaging investigations, I
go on to examine hemisphere-specific control deficits in bimanual reaching movements using a
prospective experimental design (Chapter 6). As mentioned earlier, only a few recent studies
have reported on hemisphere-specific control in bimanual movements, all of which have
examined movements in neurologically intact adults. To date, none have studied the role of
mechanisms in bimanual control in stroke survivors, which arguably provides more causal
evidence for the role of each hemisphere in bimanual control. I use a redundant bimanual task
(similar to a previous study by White and Diedrichsen) and test the predictions of a leading
theory in motor control, known as the Optimal Feedback Control model, related to flexibility of
feedback control. By examining the influence of unilateral stroke injury on this process, I
uncover hemisphere-specific differences in the control of a redundant bimanual task (Chapter 6).
Lastly, I will summarize our findings and discuss future steps in the final chapter
(Chapter 7). Prior to the experiment described in Chapter 6, I performed an experiment that used
14
an interlimb interference paradigm, motivated by early models of bimanual control. I present the
main findings of this experiment in the Appendix in which I failed to reproduce qualitatively
similar results of movement time interference observed by Kelso.

15
CHAPTER 3: The probability of choosing both hands depends on an interaction between
motor capacity and limb-specific control in chronic stroke

Published as:

Varghese, R, Kutch, JJ, Schweighofer, N, Winstein, CJ. (2020). The probability of choosing
both hands depends on an interaction between motor capacity and limb-specific control in
chronic stroke. Exp Brain Res. 238:2569–2579.

Abstract
A goal of rehabilitation after stroke is to promote pre-stroke levels of arm use for everyday,
frequently bimanual, functional activities. We reasoned that, after a stroke, the choice to use one
or both hands for bimanual tasks might depend not only on residual motor capacity but also the
specialized demands imposed by the task on the paretic hand. To capture spontaneous, task-
specific choices, we covertly observed 50 pre-stroke right-handed chronic stroke survivors (25
each of left, LHD, and right hemisphere damage, RHD) and 11 age-similar control adults and
recorded their hand use strategies for two pairs of bimanual tasks with distinct demands: one
with greater precision requirements (photo-album tasks), and another with greater stabilization
requirements (letter-envelope tasks). The primary outcome was the choice to use one or both
hands. Logistic regression was used to test the two hypotheses that the probability of choosing a
bimanual strategy would be greater in those with less severe motor impairment and also in those
with LHD. When collapsed across the four tasks, we found support for this hypothesis. However,
notably, the influence of these factors on bimanual choice varied based on task demands. For the
16
photo-album pair, the probability of a bimanual strategy was greater for those with LHD
compared to RHD, regardless of the degree of motor impairment. For the letter-envelope pair,
we found a significant interaction between impairment and side of lesion in determining the
likelihood of choosing both hands. Therefore, the manner in which side of lesion moderates the
effect of impairment on hand use depends on the task.

Keywords: bimanual, use, non-use, stroke, lateralization

17
Introduction
Every day, we engage in tasks that are bimanual in nature—buttoning a shirt, lifting a
large object, or folding a piece of paper—tasks that naturally elicit a choice to use both hands
together (Kilbreath & Heard, 2005b). After a stroke, bilateral hand use is critically reduced
(Bailey et al., 2014, 2015b; Michielsen et al., 2012b; Rinehart et al., 2009b; Vega-González &
Granat, 2005a), oftentimes despite adequate sensorimotor capacity. The inability to return to pre-
stroke patterns of bilateral hand use is associated with poor recovery of function (Haaland et al.,
2012). For this reason, there has been growing interest in understanding and potentially
promoting post-stroke arm use, with one of the foremost challenges being how to best assess use
so that it is a close approximation of behavior in the natural environment.
Controlled laboratory environments in which stroke survivors are directed to perform
pre-defined tasks (Bailey et al., 2014; Coelho et al., 2013; Przybyla et al., 2013; Yadav et al.,
2019) under set instructions or time limits, impose a restriction on free, self-selected choices.
Thus, such an approach, although well-controlled, lacks everyday representativeness and
generalizability. In recent years the use of real-world remote monitoring devices, such as
accelerometers and activity monitors, has gained attention (Bailey et al., 2014, 2015b; Franck et
al., 2019; Michielsen et al., 2012b; Rand & Eng, 2015; Rinehart et al., 2009b; Thrane et al.,
2011; Vega-González & Granat, 2005a; Yadav et al., 2019). In this approach, movement counts,
or frequency, often averaged across epochs of time, serve as a metric of use. Similarly,
simultaneous movement counts of both upper limbs serve as a proxy for bilateral use (Bailey et
al., 2014, 2015b; Sterr et al., 2002a; Uswatte et al., 2006; Vega-González & Granat, 2005a).
Although extremely valuable and relevant for providing telehealth services, this approach does
18
not capture the task-specific nature of everyday functional activities in which the two hands do
not simply move together but rather cooperate to accomplish the task goal.
Indeed, the task-specific nature of hand use is perhaps best exemplified in bimanual
tasks, which entail a natural division of labor between the hands such that each hand assumes a
preferred role (Guiard, 1987). Consider, for instance, the skilled action of threading a needle: for
a majority of right-handed individuals, the left hand assumes the role of stabilizing the needle,
while the right hand passes the thread through its eye.
By studying hand selection patterns in able-bodied adults, some studies have suggested
that hand choice results from an interaction of task demands with lateralized motor control
processes (Coelho et al., 2013; Mamolo et al., 2004, 2006; Przybyla et al., 2013; Stone et al.,
2013). According to one theoretical framework of motor lateralization (the Dynamic Dominance
hypothesis), the left hand (primarily under control of the right hemisphere) is more proficient at
stabilization of position through impedance control mechanisms; whereas the right hand
(primarily under the control of the left hemisphere) is better developed for producing precise
movement trajectories through predictive control mechanisms (R. L. Sainburg, 2002). In the
needle-threading example, given that holding the needle in place is an integral part of the task
goal and is well-aligned with the presumed competency of the left hand at position stabilization,
it follows that the left hand is “selected” (likely implicitly) by the nervous system to fill this role.
Conversely, the right hand, adept at precise visuomotor control of the thread’s trajectory, is
selected for the equally important complementary goal of passing it through the needle’s eye.
This example also illustrates that bimanual tasks, by design, lend themselves for evaluation of
the specialized use of the two hands in a functional context. The interaction between task
19
demands and lateralized motor control and their influence on spontaneous choice in chronic
stroke survivors, however, has not been formally explored in the context of bimanual tasks.
Taken together, the choice to use one or both hands in the real-world seems to not only be
self-selected but also task specific. This observational investigation is motivated by one broad
question: In chronic stroke survivors, what factors influence the spontaneous selection of both
hands for bimanual tasks— tasks that would otherwise, in age-similar able-bodied individuals,
naturally elicit the use of both hands? We tested two a-priori hypotheses that those with more
severe motor impairments (effect of impairment) and those with right hemisphere damage (side
of lesion effect) would be less likely to choose both hands together. More importantly, we also
postulated that this effect of impairment would be moderated by the side of lesion (i.e.,
impairment x side of lesion effect) which would be observable in tasks of varying demands. This
hypothesis was motivated by the known lateralization of control discussed above, specifically the
need for specialized control with each limb, i.e., the need for stabilization with the left hand, or
for precise visuomotor mapping and trajectory control with the right hand.
To observe this effect, we selected two pairs of bimanual tasks with distinct
requirements: the first pair involved folding a letter and inserting it into an envelope (letter-
envelope tasks) and the second involved receiving a large and heavy photo album and inserting a
photo into one of the album’s sleeves (photo-album tasks).
We reasoned that the stabilization of lightweight (paper) objects inherent in the pair of
letter-envelope tasks warrants the involvement of the left hand, and so might pose greater
demands in those with right hemisphere damage (RHD), in whom the left hand is weaker. If this
is true, we would expect to see a rise in the selection of both hands in those with less severe
impairment of the left limb. Conversely, owing to the weight of the album and thereafter the
20
need for precise insertion of the photo into the sleeve, the photo album task pair would stipulate
greater involvement of the right hand, thus imposing greater demands on the paretic right hand of
those with left hemisphere damage (LHD). If this is true, we would predict an increased selection
of both hands in those with less severe impairment of the right limb. It is important to note that
the two pairs of tasks are contextually different, separated in time and by the objects being
manipulated. The outcome measures selected here were not intended to infer similarity in task
structure or constraint, but rather to test the primary hypotheses related to motor impairment and
side of stroke lesion, and how these might vary for different bimanual tasks.
We validated the bimanual nature of these four tasks in age-similar able-bodied adults. In
a secondary analysis, we quantified the time taken to complete each of the four tasks. We
reasoned that even though no time limits were imposed on task performance, individuals would
likely choose a motor strategy that would represent the most temporally efficient strategy to
complete the task. If this hypothesis is correct, for the stroke group, we would expect there to be
no difference in movement time between those who choose a unimanual strategy compared to
those who choose a bimanual strategy for any given task.

Methods
Participants
Data for 50 pre-stroke right-handed chronic stroke survivors (25 left hemisphere damage,
LHD) from two previous studies were retrospectively analyzed here. Of these, 42 were enrolled
as part of a larger phase-IIb clinical trial (Dose Optimization for Stroke Evaluation,
ClinicalTrials.gov ID: NCT01749358). Detailed inclusionary and exclusionary criteria for the
DOSE study are available in (C. Winstein et al., 2019a) The remaining 8 were recruited as part
21
of a pilot study in collaboration with the Moss Rehabilitation Research Institute (MRRI, Einstein
Healthcare Network, PA) (Buxbaum et al., 2020).
All participants gave informed consent to participate in accordance with the 1964
Declaration of Helsinki and the guidelines of the Institutional Review Boards for the Health
Sciences Campus of the University of Southern California and the Einstein Healthcare Network.
Only baseline data from the DOSE study, collected between 2012-2015, were included in our
analysis. Data from the collaboration with MRRI were collected in early 2016. Additionally, 11
age-similar, right-handed, able-bodied adults were recruited from the University’s senior clinical
faculty and staff community in early 2017 as control participants specifically for this study. Self-
reported right-handedness was an inclusionary criterion for recruitment of stroke survivors in
both studies as well as for the control participants.

Motor Component of the Upper Extremity Fugl-Meyer (UEFM)
The motor component of the UEFM is a measure of impairment of the contralesional arm
and hand after stroke and involves tests of strength and independent joint control. Item-wise
scoring of the UEFM ranges from 0 (unable to perform) to 2 (able to perform completely) while
total score ranges from 0 to 66, with a higher score indicating lesser impairment (Fugl Meyer et
al., 1975).

Assessment of Choice
Spontaneous choice was assessed using items of the original Actual Amount of Use Test
(AAUT) (Taub et al., 1998). The AAUT is designed to assess spontaneous upper limb use in
stroke survivors for a series of seventeen tasks (3 postural items and 14 task-specific items, see
22
Supplementary Material I). The unique aspect of the AAUT is its covert administration;
participants who have given prior consent to be videotaped, perform the task battery without any
instructions, supervision, or time limits, and performance is captured on video unbeknownst to
the participants, who are later debriefed at the end of the assessment. To effectively
operationalize the covert nature of this test, task performance is captured on video rather than
observed in real-time by the examiner. Informed consent is gathered in a separate private waiting
room prior to testing. The video camera is turned on before the arrival of the participant into the
testing room and the camera’s tally lights are covered to obscure any indications of video
recording. The open-ended nature of the AAUT allows an objective assessment of spontaneous
unobtrusive arm use behavior in a quasi-naturalistic setting with ecologically valid tasks. Video
data were recorded at 30 fps and were available for offline observational analysis.
Figure 3.1. Photo illustration of the four tasks performed by a chronic stroke survivor (LHD, UEFM =
51). Left panel shows the two letter-envelope composite task: folding the letter and inserting in an
envelope. Right panel shows the two photo-album composite task: receiving the photo album and
inserting photo in album sleeve.
Letter-Envelope
Composite
Photo-Album
Composite
23
We selected four of the seventeen items based on an a-priori assumption that these items
would naturally elicit bimanual use in able-bodied adults. We validated this assumption by
collecting data in age-similar able-bodied adults to confirm that these tasks did indeed elicit a
bimanual strategy. The four items were: (1) Fold Letter (standard US Letter, 8.5 x 11”), (2) Insert
Letter into Envelope (standard commercial envelope, size 9, approx. 4 x 9”), (3) Receive Photo
Album (standard 3-ring binder, 2 x 9 x 12”, 2.5 lbs.) and (4) Insert Photo in Album Sleeve (4 x
3.5” polaroid images into a clear sleeve with two 4 x 6” pockets). Fig. 2.1. illustrates the tasks.
The letter envelope composite task consisted of folding the letter and inserting the letter
into the envelope. They were similar in that they involved manipulation of lightweight paper
objects, which required stabilization. The photo-album composite task consisted of receiving the
photo album and inserting the photo in the album sleeve. The photo album was heavy and
somewhat self-stabilizing, especially once received and placed on the table. As might be
intuitive, receiving the album required some degree of proximal strength and control, which
functionally distinguished it from “inserting a photo.” However, because receiving the album had
a relatively short duration (< 2 seconds) and because the two tasks within each composite task
were temporally contiguous with each other, each of the tasks were difficult to isolate.
Conversely, the composite tasks themselves were separated from each other in time. In other
words, for all participants, “inserting the letter into an envelope” closely followed “folding the
letter”; and “inserting a photo in an album sleeve” first required “receiving the album”, placing it
on the table, and opening it to a desired location. But the letter-envelope composite task always
preceded the photo-album composite task by several minutes. It was for this reason that for the
analysis, the two tasks were combined, but the composite tasks were analyzed separately. As will
24
be noted later, the obvious separation between the letter-envelope composite task and the photo-
album composite task reveals itself in the post-hoc analyses of the choice data.

Outcome Measures
The primary outcome was the overall choice of motor strategy, which was quantified as
the selection of one (unimanual, =0) or both hands (bimanual, =1) to accomplish the task goal.
All participants were successful at accomplishing the goal of the four tasks. For the tasks for
which manual roles are somewhat well-defined (insert letter into envelope, insert photo in album
sleeve), a chosen strategy was considered bimanual if both hands were engaged in task-relevant
roles. Unlike these, there was a less clear differentiation of manual roles in the Fold letter and
Receive photo album tasks, thus a chosen strategy was considered to be bimanual as long as the
contralesional arm/hand came in contact with the object and/or provided assistance toward the
successful completion of the goal.
The secondary outcome was the time taken to complete each of the four tasks, or
movement time. As each composite task was performed in a relatively continuous manner, we
implemented a discretization process to mark the start and end of each task (see Supplementary
Material II). In general, start times were defined as the frame when initial contact was made with
the object of interest (letter or envelope or photo album), and end times were defined as the
frame when the goal was accomplished, e.g., when the last fold was completed, or letter was
fully inserted, or album came in contact with the table surface. Movement time was defined as
the time elapsed between the start and end time points for each task.
Two high-school volunteers in the lab served as evaluators and were trained to code the
choice strategies and movement time discretization. Training entailed practice with a small set of
25
training videos under the supervision of the first author (RV). Both evaluators were blinded to
the a-priori hypotheses of this study to ensure independent and unbiased coding of the video
data. Additionally, video for each participant was coded by both evaluators and cross-checked
between them. Any conflicts were resolved by the graduate student supervisor (RV).

Statistical Analysis
All analyses were conducted using the R statistical computing package (version 3.5.1).

Primary analysis
Fisher’s Exact test was used to compare the proportion of bimanual choices between age-
similar able-bodied controls and chronic stroke survivors.
In the subset of stroke survivors only, to assess the influence of the degree of motor
impairment (UEFM) and the side of stroke lesion on bimanual choice (Strategy), we used nested
mixed-effects multiple logistic regression. Below is the model form (in Wilkinson notation):

logit (Strategy) ~ 1 + UEFM + Side of Lesion + UEFM: Side of Lesion + (1|Subj / Task)
… eq (1)
Random effects were modeled as Task nested within Subject (1| Subj / Task) because
Strategy was repeatedly sampled over the four tasks within each subject. We compared this
model to the null model as well as simpler reduced models using the Likelihood Ratio Tests
(LRT). To systematically explore task-wise differences, we conducted a post-hoc analysis in
which we separated the two letter-envelope tasks from the two photo-album tasks (n = 100, each)
and repeated the nested mixed-effects model.
26
Secondary analysis
Independent two-sample Welch’s t-tests were used to compare average movement times
between age-similar able-bodied controls and chronic stroke survivors across the four tasks.
In the subset of stroke survivors only, we used mixed-effects multiple linear regression to
test the influence of choice on movement time. Based on the previously studied influence of
motor impairment on movement time (Chae et al., 2002; Kamper et al., 2002; Varghese &
Winstein, 2020), we included UEFM in the model as a covariate (form below):
log (Movt time) ~ 1 + Strategy + Side of Lesion + UEFM + (1|Subj) … eq (2)
Consequently, we repeated the nested mixed-effects regression separately for each of the
two composite tasks, i.e., letter-envelope tasks and photo-album tasks.
Other potential confounders (age, chronicity, and sex) were also tested for influence in
the above models using a backward selection approach; those predictors that met a liberal cut-off
of p = 0.2 were preserved in the final reduced model. Based on this selection process, none of the
confounders met the cut-off p-value, except our hypothesized predictors. Continuous variables
(age, chronicity, UEFM scores, and movement time) were assessed for normality. UEFM was
converted to standardized z-scores. Distributions for chronicity and movement time were
positively skewed and so they were log transformed. All necessary assumptions for Generalized
Linear Models, including linearity, equality of variance, independence and normality of errors,
and multicollinearity of independent variables, were tested and found adequate.
One-way ANOVA was used to compare age among the three groups, and Welch’s t-test
was used to compare chronicity and UEFM scores between LHD and RHD. Chi-square test was
used to compare the proportion of females and males among the three groups (i.e., control, LHD,
and RHD). Significance level was set at p = 0.05.
27
Results
Of the total 61 participants, 37 (60.6%) were male. Average age of the full sample was 60 years
and of the stroke survivors was 59.7 years. Average chronicity was 5.4 years post-stroke and
score on the UEFM was 42.7, indicating moderate impairment (M. L. Woodbury et al., 2013; E.
J. Woytowicz et al., 2017). Chronic stroke survivors consisted of equal numbers of individuals
with left- (LHD) and right-hemisphere damage (RHD). RHD were on average slightly more
chronic compared to those with LHD (difference of approximately 10 months); however, median
chronicity was much more comparable between the two groups (difference of 2.73 months).
Overall, there were no significant differences between the two groups with respect to age (p =
0.11), sex (p = 0.66), chronicity (Wilcoxon p = 0.52) or UEFM scores (p = 0.51). Table 3.1
shows group-wise demographic information.

Table 3.1

28
The choice to use both hands is influenced by the degree of motor impairment and the side
of lesion.
Compared to age-similar able-bodied adults who exclusively chose a bimanual strategy,
chronic stroke survivors were significantly less likely to choose both hands for all tasks (Fisher’s
Exact p < 0.0001). In stroke survivors, of the 200 observations, 131 (65.5%) were bimanual.
Furthermore, in the full sample of stroke survivors and controls, 86.3% of all instances, in which
a bimanual strategy was chosen and when a clear preference for each limb could be ascertained
(i.e., of 88 observations), the choice was to use the right hand for the precision component and
the left hand for the stabilization component of the bimanual task.
The final model (eq. 1) was significantly different from a null model (LRT 𝜒 2
(4) = 25.55,
p= 1.18e-5) and from a model with side of lesion alone (LRT 𝜒 2
(2) = 23.64, p= 7.35e-6), and
was only modestly different from a model with UEFM alone (LRT 𝜒 2
(2) = 4.82, p= 0.09).
There was a significant effect of motor impairment on choice, such that those with a
higher UEFM scores (less severe) were more likely to choose both hands together (p= 7.77e-05).
After taking into account the effect of motor impairment, there was also a significant, albeit
small, effect of the side of lesion (p = 0.033) such that those with LHD were more likely to use
both hands together compared to those with RHD. Table 2 (column A) shows standardized
estimates and model performance measures from the mixed-effects logistic regression.

29
Table 3.2
The choice to use both hands is influenced by a task-specific interaction between motor
impairment and side of lesion.
There were systematic differences in the proportion of bimanual choices such that the
two letter-envelope tasks—folding the letter (80%) and inserting the letter into the envelope
(86%)—were nearly twice as likely to elicit a bimanual choice in stroke survivors compared to
the two photo-album tasks—receiving album (44%) and inserting photo in album sleeve (52%).
Results of task-wise mixed models revealed stark differences between the two composite
tasks with respect to the effect of motor impairment and side of lesion on bimanual choice.
Specifically, for the letter-envelope composite task, there was no main effect of UEFM but a
strong interaction effect (β = 42.67, p = 2.06e-04) suggesting that the probability of choosing a
bimanual strategy rises steeply with UEFM for those with RHD but does not significantly vary in
those with LHD. Moreover, the intercept (at the mean UEFM, 42.7) was also higher in RHD
30
compared to LHD (β = 29.22, p = 0.008) suggesting the opposite direction of effect from the
overall model. That is, for the letter-envelope tasks, those with RHD were more likely to choose
both hands together compared to those with LHD. Results for the photo-album tasks were
consistent with the overall effects observed for all four tasks. That is, those with LHD are more
likely to choose a bimanual strategy for the photo-album tasks. Table 2 (columns B & C) show
standardized estimates from the separate task-wise models.
To visualize these results, we used estimates from the logistic regression to compute and
plot predicted probabilities of using a bimanual strategy. Fig. 3.2 shows raw data and model-
fitted probabilities for LHD and RHD across UEFM scores.

Figure 3.2. Raw data (1 = bimanual, 0 = unimanual) and model-fitted probabilities (Pr.) of bimanual
choice for left- (LHD, blue) and right-hemisphere damage (RHD, red) groups. Logistic model fits A.
across all four tasks (n = 200), B. letter-envelope composite task (n = 100), and C. photo-album
composite task (n = 100). Dashed line corresponds to the mean UEFM score (z = 0) and the intercept of
the model fit.

less severe more severe
31
Movement time was significantly longer in chronic stroke survivors but not different
between those who chose a unimanual compared to a bimanual strategy.
Compared to able-bodied adults, chronic stroke survivors took 2.4 times longer on
average across tasks (t = 7.75, p <0.0001); 1.5 times longer for the letter-envelope tasks (t = 7.96,
p <0.0001), and 2.5 times longer for the photo-album tasks (t = 5.54, p <0.0001). Specifically,
for the Receive photo album task, movement times were extremely short (< 2 seconds) for 23
stroke survivors, 3 of whom demonstrated movement times ≤ 1 second (see related limitation in
the Discussion section). After taking into account the degree of motor impairment and the side of
lesion, there were no differences between individuals who chose a unimanual compared to a
bimanual strategy (Table 3).

Table 3.3
32
Across all participants, the letter-envelope tasks took significantly longer (14.4 s)
compared to the photo-album tasks (5.2 s). Figure 3.3 displays boxplots for log-transformed
movement time comparisons between strategy choices for stroke groups and controls.

Figure 3.3. Average log-transformed movement times A. across all four tasks (n = 200), B. letter-
envelope composite task (n = 100), and C. photo-album composite task (n = 100). Non-disabled control
participants (CTRL) were significantly faster compared to chronic stroke survivors who chose a bimanual
strategy. As predicted, there were no differences in movement time between those stroke survivors who
chose a unimanual compared to a bimanual strategy. *** p < 0.001

Discussion
A goal of rehabilitation after stroke is to promote pre-stroke levels of (paretic) arm use
for everyday functional activities—many of which require the use of both hands. The central
question in this study was: when faced with tasks that demand the use of both hands, how do
chronic stroke survivors choose to solve them; what factors determine whether the paretic arm is
engaged as part of the solution? This is the first study to investigate self-selected and task-
specific choices made by chronic stroke survivors in the context of bimanual tasks.
By covertly observing spontaneous use behaviors for two distinct types of bimanual
tasks, we found that, compared to age-similar, able-bodied adults, chronic stroke survivors were
Unimanual
Bimanual
***
***
***
***
***
***
33
significantly less likely to spontaneously choose a bimanual strategy. Generally consistent with
previous reports, (Bailey et al., 2014; Rinehart et al., 2009b; Thrane et al., 2011; Yadav et al.,
2019) the choice to engage both hands together depended on the degree of impairment and the
side of stroke lesion. Notably, we extended prior observations to show that the influence of these
factors on choice varied based on task demands. This is novel because arm use metrics in
previous studies (Bailey et al., 2015b; Michielsen et al., 2012b; Sterr et al., 2002a; Vega-
González & Granat, 2005a) were often collapsed across tasks, so any task-wise effects may have
been lost.

Bimanual use emerges from an interaction between task demands and lateralized motor
control processes.
As alluded to in the Introduction, we expected that the choice of strategy would emerge
from an interaction of task demands with lateralized motor control processes. For the letter-
envelope task, we suspected greater engagement of the left hand due to stabilization
requirements and found that the likelihood of choosing both hands together rose sharply for those
with moderate impairment (UEFM > 30) of the left hand (i.e., RHD), but not the right hand (i.e.,
LHD). Conversely, for the self-stabilizing photo-album, we expected greater engagement of the
right hand due to strength and precision requirements and found that the probability of a
bimanual strategy was greater for those with LHD compared to RHD, regardless of the degree of
motor impairment.
These task-specific effects seem consistent with the predictions of a theoretical
framework known as the dynamic dominance hypothesis (DDH). The traditional view of limb
dominance attributes unimanual preferences for the right hand to its general superiority for motor
34
skills (in right-hand dominant individuals) while the left hand is regarded as a weaker
counterpart—this view is hereafter referred to as global dominance (E. J. Woytowicz et al.,
2018). Conversely, the DDH proposes that each limb is proficient for a different aspect of task
performance, the left hand for position stabilization and the right hand for predictive control of
reach. In the context of a bimanual task, the two hands are preferentially selected by the nervous
system to assume roles consistent with their proficient controller.
In able-bodied adults, Stone and colleagues (Stone et al., 2013) support the idea that these
spontaneously preferred roles are dissociable between limbs. They observed that for a 3D model
building task, even when stabilization requirements were met within the task and the left hand
was free to reach and retrieve, it did not do so with any greater frequency than when stabilization
was required. Instead, the left hand rested or “hovered”, seemingly unable to disengage from its
stabilization role. Recently, Woytowicz and colleagues (E. J. Woytowicz et al., 2018) reported
that switching preferred roles of the left and right hand during a simulated “bread-cutting” task
was in fact non-optimal for motor performance. By examining relevant characteristics of the
movement, they demonstrated that the right hand showed straighter reaching trajectories but was
poor at stabilizing, whereas the left hand exhibited more stable holding performance, but was
poor at reaching.
Taken together, it appears that selecting both hands to solve a bimanual task problem
emerges from an interaction between task demands and specialized control of each limb, rather
than global dominance. Our findings suggest that this task-specific engagement of each limb is
consistent with its preferred role for precision (right hand) or stabilization (left hand), and
persists in the chronic phase after a stroke, especially when there is sufficient capacity in the
paretic hand.
35
A capacity threshold for bimanual use
The preceding discussion raises another question: what amounts to sufficient motor
capacity? The relationship between use and sensorimotor capacity has been previously explained
through the threshold hypothesis (Schweighofer et al., 2009). This hypothesis suggests that when
contralesional arm motor capacity exceeds a certain “functional threshold,” there is a sharp rise
in the likelihood of its use. Although highly variable, the predicted probabilities from our logistic
model render some support for this hypothesis. In the bimanual context, such a threshold likely
varies from task to task and may differ between those with left- versus right-hemisphere damage.
This is visible on Figure 2 where it appears that for the RHD group, the UEFM score at which a
switch from unimanual to bimanual choice (i.e., Pr. > 0.5) occurs is more distinct for the letter-
envelope tasks (UEFM ~ 30) compared to the photo-album tasks. Conversely, for the LHD
group, the threshold is more discernible for the photo-album tasks (UEFM ~ 35-40).
Previous studies report that individuals with moderate to severe paresis (UEFM < 45)
show a disproportionate tendency to exhibit unimanual non-use (Buxbaum et al., 2020; Han et
al., 2013; Thrane et al., 2011). Our data suggest that this capacity threshold may be lower for
bimanual tasks (UEFM ~ 30), compelling the engagement of the paretic limb. Data from one
recent study (Yadav et al., 2019) lends some support to this claim by showing that individuals
with more severe impairment (UEFM < 30) spend relatively more time using both hands together
rather than their paretic hand alone. Nonetheless, the exact quantification of such a capacity
threshold, and whether it is lower for bimanual tasks or simply varies by task demands is open
for further study.

36
The influence of experience, habits, and perception on spontaneous bimanual use
A final but crucial perspective for interpreting these results is that, through a unique
covert observation paradigm, we have been able to capture spontaneous use reflected in self-
selected choices. In able-bodied adults, the spontaneous choice to use both hands appears to be a
well-established behavioral response that reflects a complex implicit decision-making process.
Studies in able-bodied and disabled adults demonstrate that this decision is driven by prior habits
and experiences (Han et al., 2013; S. Kim et al., 2018), instantaneous negotiation of the salient
and non-salient features of the task and environment (e.g., task goals, object affordance)
(Mamolo et al., 2004, 2006; Stone et al., 2013), and actual (estimated) and perceived
consequences of a given action, including its associated costs (e.g., time or energetic costs) and
likelihood of success (Schweighofer et al., 2015; Shadmehr et al., 2010, 2016; Witt et al., 2004).
Stroke survivors (both LHD and RHD) in this study were on average 5 years post-
stroke—a period long enough to acquire new experiences and habits (such as learning new skills
with the ipsilesional left hand in LHD) or even reinforce pre-stroke habits (such as continued
success in using the ipsilesional right hand in RHD after initial failures using the paretic left
hand) (Jones, 2017). Given these considerations, spontaneous choice behaviors observed in our
sample of stroke survivors are likely to be largely stereotypical and successful compensatory
behaviors—behaviors that the individual may have implicitly deemed useful toward attainment
of every day bimanual task goals. With regards to costs associated with selection of a motor
strategy, we did not observe any differences in movement time between stroke survivors who
chose a unimanual compared to a bimanual strategy, lending support to the idea that individuals
would be likely to choose a motor strategy that represents the most efficient strategy in terms of
time to complete the task (but see an important limitation below).
37
Limitations
Interpretations of these findings are limited by several shortcomings. Only a very small
subset of the vast repertoire of bimanual tasks were evaluated here; results may differ for other
bimanual tasks, depending on the duration and complexity of the task (Kantak et al., 2017; E.
Woytowicz et al., 2016). The retrospective design of this study and the relatively small sample
size also limit the generalization of these observations. Nonetheless, the observational nature of
our analysis provides an important and unique perspective of ecologically valid and task-specific
arm choice after stroke. Prospective experiments are needed to assess bimanual use by
systematically varying task demands to test the interaction between task demands and lateralized
motor control. Another limitation was that the severity of motor impairment was restricted in our
cohort of participants with the minimum UEFM score being 19 in LHD and 28 in RHD. To
systematically assess if capacity threshold for bimanual use varies by the type of task and side of
lesion, future studies should consider a wider range of motor impairment scores. Lastly, we are
cautious in interpreting the results of our secondary analysis of movement time due to unreliably
short times recorded for very quick tasks like “receiving the album” as well as the unequal
sample sizes for each strategy. A within-subject design in which each participant, including
controls, performs the task unimanually and bimanually would be more appropriate to accurately
test this hypothesis. The modified AAUT (Sterr et al., 2002b), however, does not require a
unimanual ipsilesional limb strategy.
Conclusion
In conclusion, the present study provides preliminary evidence for self-selected and task-specific
choice for ecologically valid bimanual tasks in chronic stroke survivors. Unlike age-similar able-
bodied adults, chronic stroke survivors do not spontaneously choose both hands to solve routine
38
bimanual tasks. The probability of choosing both hands increases when the contralesional arm is
less impaired. Importantly, the effect of motor impairment is modified both by the side of lesion
and the type of task. We argue that our findings seem inconsistent with the predictions of a
traditional global dominance model. Instead, in chronic stroke survivors, bimanual use emerges
from a task-specific interaction between motor impairment and the side of lesion, such that when
there is sufficient motor capacity, the paretic hand is preferentially selected by the nervous
system to assume a role consistent with its specialized controller.

39
Supplements
I. Actual Amount of Use Test Task Battery
1 Grabbing and pulling out the chair
2 Open file folder
3 Go through material
4 Fold letter
5 Insert letter into envelope
6 Fill out form or hold paper while writing
7 Open box
8 Remove cards from box
9 Receive photo album
10 Turn pages of photo album
11 Insert photo in album sleeve
12 Hand album to experimenter or put away
13 Unfold newspaper
14 Open newspaper to indicated article
Note that the AAUT battery is administered unbeknownst to the participant. To effectively
operationalize the covert nature of this test, task performance is captured on video rather than
observed in real-time by the examiner. Informed consent is gathered in a separate private waiting
room prior to testing. The video camera is turned on before the arrival of the participant into the
testing room and the camera’s tally lights are covered to obscure any indications of video
recording.
40
II. Task-wise discretization. Start (onset) and end (offset) time points for each of the four
bimanual tasks.

Note:
• First full/successful attempt was assessed
• For “Fold letter” and “Receive photo album”, a strategy was considered bimanual even if
the paretic arm provided gross assist
•
III. Data and code availability
The data table and code for analysis are available on the first author’s OSF repository:
https://doi.org/10.17605/osf.io/uh574
T as k O n s e t O ffs e t
Insert letter into
envelope
Initial contact with
envelope or letter
Letter is completely
inserted
Fold letter
Initial contact with the
letter any one hand
Final (2
nd
) crease has been
folded
Insert photo in
album sleeve
Initial contact with photo or
both sleeve & photo
Photo is completely
inserted into sleeve
Receive photo
album
Initial contact with the
photo album
Album placed on the table
surface
41
CHAPTER 4: Relationship between motor capacity of the contralesional and ipsilesional
hand depends on the side of stroke in chronic stroke survivors with mild-to-moderate
impairment

Published as:
Varghese R, Winstein CJ (2020). Relationship between motor capacity of the contralesional
and ipsilesional hand depends on the side of stroke in chronic stroke survivors with mild-to-
moderate impairment. Front Neurol 10:1340.

Abstract

There is growing evidence that after a stroke, sensorimotor deficits in the ipsilesional hand are
related to the degree of impairment in the contralesional upper extremity. Here, we asked if the
relationship between the motor capacities of the two hands differs based on the side of stroke.
Forty-two pre-morbidly right-handed chronic stroke survivors (left hemisphere damage, LHD =
21) with mild-to-moderate paresis performed distal items of the Wolf Motor Function Test
(dWMFT). We found that compared to RHD, the relationship between contralesional arm
impairment (Upper Extremity Fugl-Meyer, UEFM) and ipsilesional hand motor capacity was
stronger (𝑅 𝐿𝐻𝐷 2
= 0.42; 𝑅 𝑅𝐻𝐷 2
< 0.01; 𝑧 = 2.12; p = 0.03) and the slope was steeper (𝑡 = -2.03; p
= 0.04) in LHD. Similarly, the relationship between contralesional dWMFT and ipsilesional
hand motor capacity was stronger (𝑅 𝐿𝐻𝐷 2
= 0.65; 𝑅 𝑅𝐻𝐷 2
= 0.09; 𝑧 = 2.45; p = 0.01) and the slope
was steeper (𝑡 = 2.03; p = 0.04) in LHD compared to RHD. Multiple regression analysis
confirmed the presence of an interaction between contralesional UEFM and side of stroke (𝛽
3
=
42
0.66 ± 0.30; p = 0.024) and between contralesional dWMFT and side of stroke (𝛽
3
= -0.51 ±
0.34; p = 0.05). Our findings suggest that the relationship between contra- and ipsi-lesional
motor capacity depends on the side of stroke in chronic stroke survivors with mild-to-moderate
impairment. When contralesional impairment is more severe, the ipsilesional hand is
correspondingly slower in those with LHD compared to those with RHD.

43
Introduction
It is now well known that unilateral stroke not only results in weakness of the opposite
half of the body, i.e. contralateral to the lesion or contralesional limb, but also significant motor
deficits in the same half of the body, i.e. ipsilateral to the lesion or ipsilesional limb (Chestnut &
Haaland, 2008; Sunderland et al., 1999a; Wetter et al., 2005; C. J. Winstein & Pohl, 1995).
Previous work suggests that deficits in the ipsilesional arm and hand varies with the severity of
contralesional deficits, especially in the sub-acute and chronic phase after stroke (Bustrén et al.,
2017; Metrot et al., 2013; Noskin et al., 2008; Rinne et al., 2017). More interestingly, the
unilateral motor deficits observed for contralesional and ipsilesional limbs seem to be
hemisphere-specific and thus depend on side of stroke lesion (de Paiva Silva et al., 2014, 2018;
Hanna-Pladdy et al., 2002; Harris & Eng, 2006; Mani et al., 2013b; Nowak et al., 2007; Schaefer
et al., 2007). For predominantly right-handed cohorts, contralesional deficits appear to be more
severe in those with right hemisphere damage (RHD), in whom the contralesional limb is non-
dominant. For example, using clinical motor assessments of grip strength and hand dexterity,
Harris and Eng (Harris & Eng, 2006) showed that contralesional motor impairments were less
severe in chronic stroke survivors who suffered damage in the dominant (i.e. left) hemisphere
(LHD) compared to those who suffered damage in the non-dominant (right) hemisphere (Hanna-
Pladdy et al., 2002; Harris & Eng, 2006).
In contrast, considering ipsilesional motor deficits, the evidence is mixed concerning
hemisphere-specific effects. For instance, some studies reported that individuals with LHD
exhibited more severe ipsilesional arm and hand deficits compared to those with RHD (Hanna-
Pladdy et al., 2002; Heap & Wyke, 1972; Sunderland et al., 1999b; Wyke, 1968a) while others
have reported no difference in ipsilesional hand motor capacity between LHD and RHD (Wetter
44
et al., 2005). In acute stroke survivors, Nowak et al demonstrated that deficits in grip force of the
ipsilesional hand were significantly associated with clinical measures of function of the
contralesional hand only in LHD (Nowak et al., 2007). Contrary to this, de Paiva Silva and
colleagues found that compared to controls and LHD, the ipsilesional hand in chronic stroke
survivors was significantly slower and less smooth in RHD especially when contralesional
impairment was relatively more severe (UEFM < 34) (de Paiva Silva et al., 2018).
Taken together, there is converging evidence regarding the relationship between motor
deficits of the contralesional and ipsilesional upper extremity, such that ipsilesional deficits are
worse when contralesional impairment is greater (Figure 1A); however, it is uncertain whether
the relationship between the two limbs depends on which hemisphere is damaged. In particular,
motor deficits of the two limbs are most prominent for tasks that require dexterous motor control
(e. g., grip force, tapping, tracking). For predominantly right-handed cohorts (as is the case in
most studies), contralesional deficits appear to be more severe in those with RHD, in whom the
contralesional limb is non-dominant, whereas ipsilesional deficits are more severe in those with
LHD. An exception to this observation for those with RHD seems to be in the case when
contralesional impairment is most severe (i.e., UEFM < 34) (de Paiva Silva et al., 2018). Thus,
one might predict that as contralesional impairment worsens, individuals with LHD would have
proportionally worse ipsilesional deficits, but individuals with RHD (especially if say UEFM >
34) would not; see (Figure 1 B & C) for two alternative hypotheses. To our knowledge, this
prediction has not before been explicitly tested.
One reason that this prediction remains untested might be methodological in that in these
previous studies participants were categorically classified based on the degree of contralesional
motor impairment (e.g., mild, moderate, severe) (Bustrén et al., 2017; de Paiva Silva et al., 2018;
45
Noskin et al., 2008). Categorization (or worse, dichotomization) of a continuous variable
presents several concerns, including loss of measurement resolution, an assumption of
discontinuity in the underlying construct (in this case motor impairment), unequal subgroup sizes
(or biased sampling), and large unexplained residuals in regression models (Collins et al., 2016;
Naggara et al., 2011; Royston et al., 2006). Overall, if the objective is to understand the nature
and extent of critical response-predictor relationships, then a categorical approach is particularly
problematic.
Our primary objective was to determine if the severity of deficits in the ipsilesional hand
varies directly with that of the contralesional hand and if this relationship differs based on the
side of stroke lesion (i.e., an interaction effect). To accomplish this objective, we conducted a
retrospective regression analysis of an existing dataset taken from mild-to-moderately impaired
chronic stroke survivors. Although the range of contralesional impairment was limited in this
dataset, we preserved its continuity and tested the prediction that the dexterous motor capacity of
the ipsilesional hand varies directly with the severity of the contralesional motor impairment in
individuals with LHD, but not in individuals with RHD; see (Figure 1).

Figure 4.1. Hypothesized effects represented in schematic figure. (A) The null hypothesis, wherein the
relationship between contralesional (CL) impairment and ipsilesional (IL) motor capacity is not modified
by the side of stroke lesion, i.e., 𝛽 1
≠ 0 but 𝛽 3
= 0. (B) Alternative hypothesis 1, wherein ipsilesional
deficits are related to contralesional impairment but only in LHD (blue) and not in RHD (red). (C)
Alternate hypothesis 2, wherein ipsilesional deficits are related to contralesional impairment but only in
LHD and in RHD with severe impairment (represented in the shaded dark-grey area). For both alternate
hypotheses, 𝛽 1
and 𝛽 3
≠ 0.
46

Methods
Participants
A retrospective analysis of data from 42 chronic stroke survivors (n = 21 left-hemisphere
damage, LHD) was conducted. Participants were enrolled as part of a larger phase-IIb clinical
trial (Dose Optimization for Stroke Evaluation, ClinicalTrials.gov ID: NCT01749358) (C.
Winstein et al., 2019c) and provided informed consent in accordance with the 1964 Declaration
of Helsinki and the guidelines of the Institutional Review Board for the Health Science Campus
of the University of Southern California. Participants were at least 150 days post-stroke, pre-
morbidly right-handed with mostly resolved upper extremity paresis. For a complete description
of the inclusionary and exclusionary criteria, please refer to Supplemental Materials.

Outcome Measures
Motor Component of the Upper Extremity Fugl-Meyer (UEFM)
The UEFM (Fugl-Meyer et al., 1975) is an assessment of motor impairment of the contralesional
arm and hand after stroke and includes tests of strength and independent joint control. Item-wise
scoring of the UEFM ranges from 0 (unable to perform) to 2 (able to perform completely) while
47
total score ranges from 0 to 66, with a higher score indicating lesser impairment. The UEFM
score was modeled as a continuous variable for statistical analysis purposes.

Wolf Motor Function Test (WMFT)
The WMFT assesses upper extremity motor capacity through timed functional task performance
(e.g., lifting a can, pencil, or paper clip). Originally designed for patients with moderate to severe
upper extremity motor deficits, the test was later modified by Morris, Crago and Taub to
accommodate individuals with mild motor impairments (Wolf et al., 2001). Hand motor capacity
was assessed for both the contralesional and ipsilesional hands using the distal task battery of the
WMFT (dWMFT) (M. Woodbury et al., 2010) which consisted of the following 8 tasks: lift can,
lift pencil, lift paper clip, stack checkers, flip cards, turn a key in a lock, fold towel, and lift
basket. A Principal Component Analysis of WMFT time scores revealed two clusters: one
consisting of the proximal (#1-8, except 6, i.e., lifting weight to box), the other consisting of the
distal (#9-17, except 14, i.e., grip strength) (Kim et al, unpublished, personal communication).

Statistical Analyses
All analyses were conducted in the R statistical computing package version 3.5.1 (R Core Team,
2012). To test the hypothesis that the inter-limb relationship of motor capacity is modified by the
side of stroke lesion, we used the coefficient of determination (𝑅 2
) and compared the
covariances between LHD and RHD using the Fisher’s Z test. We then performed a simple
linear regression to determine the slope of the relationship between contralesional (CL) UEFM
and ipsilesional (IL) dWMFT (Model 1), and CL dWMFT and IL dWMFT (Model 2). We used
t-tests to compare slope between LHD and RHD.
48
To supplement these primary analyses and as a more robust assessment of the interaction
between the side of lesion and contralesional motor capacity, we used multiple linear regression
of the following form:

In both models, 𝑦 is the average time score on the distal WMFT of the ipsilesional hand. Using
this multiple model, our hypotheses were that 𝛽 1
≠ 0 and 𝛽 3
≠ 0 (see Figure 1). Any statistically
significant interaction was resolved post-hoc using a t-test comparison of estimated marginal
trends between LHD and RHD.
All continuous variables were assessed for normality using Lilliefors test (modified
Kolmogorov-Smirnov test). Distributions for chronicity and average time-score for the distal
WMFT were positively skewed and were therefore log-transformed. Welch’s t-tests were used to
compare age, chronicity, and Upper Extremity Fugl-Meyer scores between LHD and RHD,
whereas Chi-square test was used to compare the proportion of females and males between the
two groups. Each group was standardized to its own unit variance (z-scored) to equalize range
and for subsequent linear regression analysis. Outliers were identified by visual inspection of
scatterplots. Any value of IL dWMFT more extreme than ± 1.5 log-SD was examined carefully
for their influence on interlimb covariance and slope. If removal of these observations did not
change the direction or significance of the effect in the simple model, we included them in the
final model. Residuals of the final model were analyzed to confirm that all necessary
assumptions for multiple regression were met. Significance level (α) was set at p < 0.05.
𝐌𝐨𝐝𝐞𝐥 𝟏 : 𝑦 = 𝛽 0
+ 𝛽 1
(𝐶𝐿 𝑈𝐸𝐹𝑀 ) + 𝛽 2
(𝑆𝑖𝑑𝑒 𝑜𝑓 𝑆𝑡𝑟𝑜𝑘𝑒 ) + 𝛽 3
(CL 𝑈𝐸𝐹𝑀 ∗ 𝑆𝑖𝑑𝑒 𝑜𝑓 𝑆𝑡𝑟𝑜𝑘𝑒 ) + 𝜖
𝐌𝐨𝐝𝐞𝐥 𝟐 : 𝑦 = 𝛽 0
+ 𝛽 1
(𝐶𝐿 𝑑𝑊𝑀𝐹𝑇 ) + 𝛽 2
(𝑆𝑖𝑑𝑒 𝑜𝑓 𝑆𝑡𝑟𝑜𝑘𝑒 ) + 𝛽 3
(CL 𝑑𝑊𝑀𝐹𝑇 ∗ 𝑆𝑖𝑑𝑒 𝑜𝑓 𝑆𝑡𝑟𝑜𝑘𝑒 ) + 𝜖
49
In order to select the predictor variables that best explain the response, we used a
backward selection approach, in which we began by adding all predictor variables in each of the
two above models to explain the response variable 𝑦 . This included our hypothesized predictors,
CL UEFM (or CL dWMFT) and the side of stroke lesion (LHD or RHD), and, potential known
confounders (age, chronicity, and sex). In a combined full model, those predictors that met a
liberal cut-off of p = 0.2 were preserved in the final reduced model. Based on this selection
process, we found sex to be a significant confounder (p = 0.08) in Model 1, and therefore
included it as a predictor in the reduced Model 1. For Model 2, none of the confounders met the
cut-off p-value, except our hypothesized predictors. Additional information on model selection
and model diagnostics is included as supplementary materials. Standard errors and 95% CI of the
estimates of regression coefficients were confirmed by performing 1000 bootstrap replicates.

Results
Descriptive statistics for all participants are provided in Table 1. On average, the 42 stroke
survivors had moderate arm impairment (UEFM = 41.6), were approximately 60 years of age,
5.75 years post-stroke, and were predominantly male (74%). There were no significant
differences between LHD and RHD in the level of impairment, chronicity or the number of
males. Individuals with RHD were younger compared to LHD (median age difference 8.7 years)
but not statistically different.

Table 4.1. Descriptive Statistics for the full sample (N = 42), and for the two groups of interest, left
hemisphere damage, LHD (n = 21) and right hemisphere damage, RHD (n = 21).
50
* Count (percent) for categorical variables

Model 1: Side of lesion modifies the relationship between CL UEFM and IL motor capacity
Contralesional UEFM explained 42% of the variance in ipsilesional hand motor capacity
in LHD (p = 0.001), but less than 1% in RHD (p > 0.05). The slope of this relationship was -0.65
± 0.17 (p = 0.001) in LHD and -0.066 ± 0.23 (p = 0.78) in RHD. Compared to RHD, the
covariance between contralesional UEFM and ipsilesional hand motor capacity was significantly
stronger (Fisher’s 𝑧 = 2.12, p = 0.03) and the slope was steeper in LHD (𝑡 = -2.03, p = 0.04).
Four observations (two each in LHD and RHD) were identified as potential outliers.
After removal of these outliers, contralesional UEFM explained 44.3% of the variance in
ipsilesional hand motor capacity in LHD (p = 0.001), and 2.26% in RHD (p > 0.05). The slope of
this relationship changed to -0.42 ± 0.11 (p = 0.001) in LHD and 0.13 ± 0.20 (p = 0.54) in RHD.
Variable
Overall (N = 42)
Mean * (± SD)
LHD (n = 21)
Mean (± SD)
RHD (n = 21)
Mean (± SD)
two-sided
p-value
Sex (Male)
*
31 (73.8) 15 (71.4) 16 (76.2) ~1
Age (years)
[Min, Max]
59.16 (12.3)
[35.48, 80.54]
60.71 (10.8)
[43.15, 80.54]
57.62 (13.8)
[35.48, 77.28]
0.42
Chronicity
(months)
[Min, Max]
69.37 (36.9)
[26.33, 212.52]
63.32 (24.4)
[30.02, 111.52]
75.41 (46.1)
[26.33, 212.52]
0.29
UEFM
Motor (/66)
[Min, Max]
42 (10)
[19, 59]
41 (13)
[19, 59]
42 (8)
[28, 55]
0.58
51
Again, a comparison of the covariances and slope between the groups revealed that compared to
RHD, the relationship between contralesional UEFM and ipsilesional hand motor capacity was
significantly stronger (Fisher’s 𝑧 = 2.7, p = 0.006) and the slope was steeper in LHD (𝑡 = -2.41, p
= 0.02).
Since these observations did not significantly change the strength of covariance nor the
slope of the relationship, they were preserved in the final multiple model. Analysis of residuals
of the final model did not indicate violations of necessary assumptions in multiple regression in
terms of linearity, equality of variance, independence and normality of errors, and
multicollinearity of independent variables, nor the presence of unduly influential observations.
Nonetheless, estimates below are reported both with and without suspected outliers.
Figure 4.2. Scatterplots show the relationship between contralesional motor impairment (CL UEFM) and
ipsilesional distal motor performance (IL dWMFT) for (A) the full sample, (B) LHD, and (C) RHD. Solid
lines represent the linear prediction and shaded areas represent the 95% confidence interval. For ease of
interpretation, rounded estimates of raw values (in seconds for dWMFT and points for UEFM) have been
provided in green in Panel A. Asterisks indicate values evaluated as outliers.
After adjusting for main effects and significant confounders using multiple regression,
the final reduced form of Model 1 was statistically different from a null model (F = 3.47, p =
52
0.016, adjusted 𝑅 2
= 0.19). Based on estimates from Model 1, CL impairment (UEFM) was
significantly associated with IL hand motor capacity, i.e., dWMFT, (𝛽 1
= -0.72 ± 0.21, p =
0.001; without outliers: -0.44 ± 0.17, p = 0.01) (Figure 2A). There was no significant effect of
the side of lesion (𝛽 2
= 0.026 ± 0.27, p = 0.92; without outliers: 0.22 ± 0.23, p = 0.33). There
was a significant interaction between the side of lesion and CL impairment (𝛽 3
= 0.66 ± 0.30, p
= 0.024; without outliers: 0.56 ± 0.24, p = 0.024). Post-hoc contrasts of estimated marginal
trends indicated that the slope of the relationship between CL UEFM and IL dWMFT was
significantly more negative in LHD compared to RHD (𝑡 = -2.34, p = 0.02; without outliers: -
2.37, p = 0.02). Figure 2 B and C illustrates the interaction.

Model 2: Side of lesion modifies the relationship between CL dWMFT and IL motor capacity
Contralesional dWMFT explained 65% of the variance in ipsilesional hand motor
capacity in LHD (p < 0.001), but only 9% in RHD (p > 0.05). The slope of this relationship was
0.81 ± 0.13 (p < 0.001) in LHD and 0.29 ± 0.22 (p = 0.19) in RHD. A comparison of the
covariances and slope between LHD and RHD revealed that compared to RHD, the relationship
between CL dWMFT and IL motor capacity was significantly stronger (Fisher’s 𝑧 = 2.45, p =
0.01) and the slope was steeper in LHD (𝑡 = 2.03, p = 0.04).
After removing the outlying observations, contralesional dWMFT explained 62% of the
variance in ipsilesional hand motor capacity in LHD (p < 0.001), and < 1% in RHD (p > 0.05).
The slope of this relationship changed to 0.54 ± 0.1 (p < 0.001) in LHD and 0.05 ± 0.21 (p =
0.81) in RHD. Compared to RHD, the relationship between CL dWMFT and IL motor capacity
was significantly stronger (Fisher’s 𝑧 = 2.85, p = 0.004) and the slope was steeper in LHD (𝑡 =
2.11, p = 0.04).
53

Since these observations did not significantly change the strength of covariance or the
slope of the relationship, they were preserved in the final multiple model. Once again, analysis of
residuals did not indicate violations of necessary assumptions in multiple regression nor the
presence of unduly influential observations. Nonetheless, estimates below are reported both with
and without suspected outliers.

Figure 4.3. Scatterplots show relationship between contralesional distal motor performance (CL
dWMFT) and ipsilesional distal motor performance (IL dWMFT) for (A) the full sample, (B) LHD, and
(C) RHD. Solid lines represent the linear prediction and shaded areas represent the 95% confidence
interval. For ease of interpretation, rounded estimates of raw values (in seconds) have been provided in
green in Panel A. Asterisks indicate values evaluated as outliers.

After adjusting for main effects and significant confounders using multiple regression,
the final reduced form of Model 2 was statistically different from a null model (F = 7.48, p <
0.001, adjusted 𝑅 2
= 0.32). Based on estimates from Model 2, CL hand motor capacity was
54
significantly associated with IL hand motor capacity (𝛽 1
= -0.81± 0.16, p < 0.001; without
outliers: 0.54 ± 0.17, p = 0.003) (Figure 3A). There was no significant effect of the side of lesion
(𝛽 2
= 0.12 ± 0.26, p = 0.66; without outliers: 0.19 ± 0.22, p = 0.39). There was an interaction
between the side of lesion and CL hand motor capacity, but it only approached statistical
significance (𝛽 3
= -0.51 ± 0.34, p = 0.05; without outliers: -0.49 ± 0.24, p = 0.046). Figure 3 B
and C illustrates the interaction.
Finally, we note here that we conducted similar analyses for the proximal component of
the WMFT (not reported) but did not observe the same relationships (model adj. 𝑅 2
= approx.
3% for both models, p > 0.25). One possibility is that unlike the distal component, the proximal
WMFT is a much less sensitive metric of motor performance, especially in individuals with
mild-to-moderate impairment.

Discussion
For the first time, we explicitly tested the hypothesis that motor capacity of the ipsilesional hand
is influenced by an interaction between the severity of contralesional deficits and the side of
stroke lesion. Using retrospective analysis of an existing dataset, we found that ipsilesional
motor capacity co-varies with contralesional impairment to a significantly greater degree in
individuals with LHD compared to RHD.

Analysis of the interaction effect
Hints of this interaction were implicit in a few previous studies (Chestnut & Haaland,
2008; de Paiva Silva et al., 2018; Wetter et al., 2005); however, categorical reporting of the
Upper Extremity Fugl-Meyer (UEFM) masked this interesting effect. By preserving the
55
continuity of the UEFM and utilizing continuous standardized z-scores, our statistical approach
allowed a direct comparison of our regression estimates, thus reflecting effect sizes of each of the
candidate predictors. Unlike previous studies, we did not observe a significant effect of the side
of lesion on ipsilesional motor capacity (Heap & Wyke, 1972; Wyke, 1968a) or on contralesional
UEFM or dWMFT (Harris & Eng, 2006). One reason might be that the effect of the side of
lesion observed in those previous studies may have arisen from its interaction with contralesional
impairment. However, because an interaction effect was not explicitly tested and because
contralesional impairment was either collapsed across the groups (Chestnut & Haaland, 2008;
Wetter et al., 2005) or categorical (Bustrén et al., 2017; de Paiva Silva et al., 2018), variance in
the ipsilesional capacity may have been conflated with the effect of the side of lesion, or,
remained unexplained, especially for UEFM scores that fell at the boundaries of the pre-defined
categories. Previously, various cut-off scores for the UEFM have been used to define impairment
categories (Bustrén et al., 2017; de Paiva Silva et al., 2018). Of these, our data suggest that a
UEFM score of 42, which occurs at the intersection of the linear fits for LHD and RHD, would
best reflect the change in the direction of effect—that is, for UEFM scores less than 42, the
ipsilesional limb would be slower in LHD compared to RHD whereas for UEFM scores greater
than 42, those with LHD would be slightly faster compared to RHD. Interestingly, at this score
of 42, there would appear to be no differences in motor capacity of either hand between LHD
and RHD, which might explain why a number of large clinical trials (e.g. EXCITE (Wolf et al.,
2006), ICARE (C. J. Winstein et al., 2016)), designed for mild to moderately impaired stroke
survivors (mean UEFM scores in these studies were 42.5 and 41.6 respectively) may not have
observed, on average, any differences in motor capacity based on the side of stroke.
56
On a related note, in Model 1, we observed that the relationship between contralesional
UEFM and ipsilesional motor performance was (mildly) confounded by sex. Specifically, males
showed slightly faster performance times (mean difference = 0.3 seconds, β = -0.56, p = 0.08)
compared to females. Given that this difference was very small and may have been exaggerated
by the unequal sample sizes, i.e., there were about 3 times more males than females, in LHD,
RHD and overall, we suspect this was an artefact of the unequal sample sizes rather than a true
difference between males and females.

Insights from the type of task
A common link between our study and past reports is that ipsilesional deficits were found
to be most pronounced for distal (dexterous) tasks. These tasks—lift can, lift pencil, lift paper
clip, stack checkers, flip cards, turn a key in a lock, fold towel, and lift basket—nearly always
involve object manipulation and inherently require dexterous motor control of the hands.
Sunderland and colleagues (Sunderland et al., 1999a) demonstrated that early on after a stroke,
spatial accuracy in dexterity tasks performed with the ipsilesional hand correlated with cognitive
deficits, such as apraxia, in individuals with LHD. While individuals included in this study did
not exhibit severe apraxia and were approximately 5 years post-stroke, it is possible that mild
cognitive deficits, including apraxia, may have impacted dexterous task performance in those
with LHD, especially in the more severe ranges of UEFM. Furthermore, we note that our
evaluation of dexterous task performance was through timed tests, and not quality of movement
or accuracy. It has been suggested that the left hemisphere plays an important role in regulating
the timing and speed of movements (de Paiva Silva et al., 2014), and thus, injury to the left
hemisphere, such as to premotor and fronto-parietal networks (Harrington et al., 1998;
57
Harrington & Haaland, 1991) may impair planning and sequencing required for smooth and
rapid performance of dexterous motor tasks.

The role of the left hemisphere in the control of both hands
Our main observation that deficits in ipsilesional hand motor capacity scale with
contralesional impairment only in LHD is qualitatively similar to previous clinico-behavioral

(Heap & Wyke, 1972; Wyke, 1966, 1968a, 1971) and phenomenological evidence.

(Cernacek,
1961; Freitas et al., 2011; Giuliani et al., 1997) These findings are consistent with a rather
simplified organizational model of the nervous system in which certain aspects of motor and/or
cognitive control are lateralized to the left (or dominant) hemisphere, such that damage to the left
hemisphere results in deficits in skilled motor actions of both upper extremities. For example,
using EMG recordings of homologous muscles in the arm, Cernacek (1961) demonstrated that
the frequency of motor irradiations, i.e., unintended motor output in the ipsilateral hand, were
significantly higher from the dominant to the non-dominant extremity. (Cernacek, 1961)
Similarly, Wyke (1968) reported that while individuals with left-sided cerebral lesions exhibited
bilateral motor deficits in speed and limb postural control, deficits in those with right cerebral
lesions were restricted to the contralateral limb. (Wyke, 1968b) Even within the LHD patient
group, the nature and extent of ipsilesional deficits has been shown to be modulated by the
degree of (clinical) paresis. For example, Haaland et al (2009) demonstrated that deficits in
ipsilesional torque amplitude specification were statistically significant in LHD patients with
contralesional upper extremity paresis compared to those with no paresis, despite all other
features of the movement and lesion (e.g., error, speed, lesion volume) being similar. (Haaland et
al., 2009)
58
Lastly, in one of the earliest experiments using functional MRI, Kim and colleagues
(1993) showed that the task-evoked activation of the left hemisphere was substantially greater
for ipsilateral movements compared to the right hemisphere. (S. G. Kim et al., 1993) In later
years, a number of neuroimaging (Bundy et al., 2018; Serrien et al., 2003) and neurophysiologic
(Ganguly et al., 2009; Hess et al., 1986; Sohn et al., 2003) studies have provided confirmatory
evidence for the role of the dominant hemisphere in organizing motor outputs to both hands. Our
results of co-varying deficits between the contralesional and ipsilesional hand in LHD provides
further empirical support for the role of the left hemisphere (in our pre-morbidly right-handed
group) in the control of both hands.
Limitations and future considerations
In interpreting our findings, it is important to consider that this study is a retrospective
analysis of a relatively small dataset. Therefore, sample size, and characteristics such as the
range of impairment, chronicity, age, were limited to what was available. To this point, we
conducted analyses with and without outliers, and found that while exclusion of outliers affected
the strength of the overall model, it did not affect the probability associated with rejecting the
null hypothesis. A prospective study or independent validation in a separate cohort would be
ideal, if larger samples were available. A larger sample would render more robust findings that
are less sensitive to distortions from outlying values.
Along this line, UEFM scores for the RHD group were restricted towards the more severe
range, with the most severely impaired individual’s score being 28. This restriction, however,
was less so in the LHD group (min. UEFM = 19). Although this limitation in range was
circumvented by using group-wise z-scores, we are cautious in generalizing our observations
regarding the interaction effect to more severe ranges of motor impairment in RHD. This is quite
59
apparent in the variability around our estimated linear fits especially towards the extreme ranges
of predictor values for RHD. Indeed, it is possible that for the more severe range in RHD, there
exists a linear relationship between contralesional and ipsilesional motor deficits as illustrated in
Figure 1C. Thus, while we can, with some confidence, reject the null hypothesis, our data are
insufficient to differentiate between the two alternate hypotheses, and warrant a follow-up study.
Finally, it must be emphasized that the absence of a relationship with contralesional
impairment in RHD should not be taken to mean that ipsilesional deficits are absent in this
group. In fact, there is substantial evidence to the contrary. Comparison with an appropriate
control group would be necessary to demonstrate the presence of ipsilesional deficits in RHD
and the functional implications of these deficits. As alluded to earlier, measuring the speed of
performance, as in the case of timed functional tasks assessed here, does not provide specific
information about perceptual errors, spatial accuracy or visuomotor deficits, which, based on
previous evidence, (Boll, 1974; Bracewell et al., 1990) might be a more important component of
motor performance in RHD.

Conclusion
In summary, our results suggest that ipsilesional motor deficits co-vary with the degree of
impairment in LHD, but this relationship is less pronounced in RHD. This observation further
underscores the extensive motor experiences of the pre-morbidly dominant ipsilesional limb and
the importance of the left hemisphere in the control of timed tasks for both hands. For the future,
we propose that a hypothetical model of bilateral deficits in LHD is readily testable through a
prospective study that uses a bimanual experimental paradigm with sensitive kinematic
measures. Such a paradigm could offer important insights into the role and organization of each
hemisphere for the control of uni- and bi-manual movements.
60
Supplements

Inclusionary and Exclusionary Criteria for the DOSE study (Winstein et al, 2019, Stroke)
Participants were included if they were: 1) between 21-75 years of age, 2) pre-morbidly right-
handed, 3) ≥ 150 days post stroke (chronic phase), 4) had mild to moderate residual motor
impairment (Upper Extremity Fugl-Meyer, UEFM score ≥ 19) with mostly resolved upper
extremity paresis. Participants were excluded if they had: 1) severe sensory disturbances (no
response to light touch or complete loss of proprioception as indicated by the UEFM), 2) current
major depressive disorder (score > 3 on PHQ2, depression screening survey) 3) a history of
recent surgeries, significant orthopedic injuries, or pain affecting the upper extremity that would
restrict shoulder and elbow movement, 4) severe cognitive deficits such as aphasia, apraxia or
neglect that would preclude participants from comprehending test instructions or questionnaires.

Data and Code Availability
The complete raw dataset along with a codebook for analysis is available through the first
author’s OSF repository: https://osf.io/pbtk9

61
CHAPTER 5: Corpus callosal microstructure predicts bimanual motor performance in
chronic stroke survivors

Preprinted as:
Varghese R, Chang, B, Kim, B, Liew SL, Schweighofer, N, Winstein CJ (2021). Corpus
callosal microstructure predicts bimanual motor performance in chronic stroke survivors.
bioRxiv 2021.05.14.443663

Abstract
Much of the research using diffusion tensor imaging (DTI) in stroke focuses on characterizing
the microstructural status of corticospinal tracts and its utility as a prognostic biomarker.
However, the ischemic event in the lesioned cortex also triggers structural and functional
alterations in its contralateral homolog through the corpus callosum (CC), known as transcallosal
diaschisis. The few studies that have characterized the microstructural status of the CC using
DTI only examine its relationship with paretic limb performance. Given the well-established role
of the CC for bimanual coordination, especially fibers connecting the larger sensorimotor
networks such as prefrontal, premotor, and supplementary motor regions, we examine the
relationship between the microstructural status of the CC and bimanual performance in chronic
stroke survivors (n = 41). We used movement times for two self-initiated and self-paced
bimanual tasks to capture bimanual performance. Using publicly available control datasets (n =
52), matched closely for acquisition parameters, including sequence, diffusion gradient strength
and number of directions, we also explored the effect of age and stroke on callosal
microstructure. We found that callosal microstructure was significantly associated with bimanual
62
performance in chronic stroke survivors such that those with lower callosal FA were slower at
completing the bimanual task. Notably, while the primary sensorimotor regions (CC3) showed
the strongest relationship with bimanual performance, this was closely followed by the
premotor/supplementary motor (CC2) and the prefrontal (CC1) regions. We used multiple mixed
regression to systematically account for loss of callosal axons (i.e., normalized callosal volume)
as well as differences in lesion size and other metrics of structural damage. Chronic stroke
survivors presented with significantly greater loss of callosal fiber orientation (lower mean FA)
compared to neurologically intact, age-similar controls, who in turn presented with lower callosal
FA compared to younger controls. The effect of age and stroke were observed for all regions of
the CC except the splenium. These preliminary findings suggest that in chronic stroke survivors
with relatively localized lesions, callosal microstructure can be expected to change beyond the
primary sensorimotor regions and might impact coordinated performance of self-initiated and
cooperative bimanual tasks.

63
Introduction
Focal ischemic injury to the central nervous system can result in changes remote from the
site of injury (diaschisis, (von Monakow, 1914). One such case is transcallosal diaschisis in
which the ischemic event in the lesioned cortex triggers structural and functional alterations in its
contralateral homolog through the corpus callosum. Although the exact mechanism of
transhemispheric diaschisis in humans is not well understood, evidence from animal models
suggests a dual process (Jones & Adkins, 2015). First, early after the injury, diaschisis is
evidenced by reductions in cellular metabolism and local cerebral perfusion in the ischemic
penumbra. As a result, axon terminals of transcallosal afferents originating from the lesion site
degenerate and cause higher-order axons to systematically eliminate or prune away from the
intact cortex. In behaviorally adaptive animals, this is simultaneously accompanied by a second
process of reactive synaptogenesis and dendritic arborization in the intact cortex, secondary to
the nonuse of the weak limb and compensatory overuse of the healthy limb. The severity of
asymmetry in limb use is directly related to the size of the arbors, which over time, with the
recovery of symmetric limb use, are pruned (Jones & Schallert, 1994).
The reactive growth and reorganization process in the intact hemisphere stabilizes in the
subacute and chronic phases (Jones & Schallert, 1992) and provides unique insight into post-
stroke recovery. Because reinnervation is accomplished during this time through the sprouting of
collaterals from existing axons rather than the genesis of new long-distance neurons, this process
is most likely to alter the microstructure and connectivity of local axon pools within a given
region. In humans, such microstructural changes can be noninvasively approximated using
diffusion tensor imaging (DTI).
64
There is growing evidence for the utility of DTI-derived metrics in humans as prognostic
biomarkers of recovery after stroke (B. Kim & Winstein, 2017; Puig et al., 2017).
Microstructural status of the corticospinal tract, principally the composite fractional anisotropy
measure (FA), has been shown to predict short and long-term outcomes including stroke
severity, disability, functional improvement (B. Kim, 2017; Lindenberg et al., 2012), and
sensorimotor performance (Findlater et al., 2019) in chronic stroke.
In recent years, DTI has also revealed interesting patterns of transcallosal diaschisis and
reorganization in humans (Egorova et al., 2020; Hawe et al., 2013; Yeh et al., 2013) and its
association with functional motor recovery. For example, compared to controls, lower FA values
in the corpus callosum (CC) have been observed in the subacute (Li et al., 2015; Wang et al.,
2012)and chronic phase after stroke (Stewart, Dewanjee, et al., 2017; Stewart, O’Donnell, et al.,
2017). In the latter study, lower FA in the motor region of the CC was found to correlate with
greater sensorimotor impairment in the paretic hand. In chronic stroke survivors, microstructural
status of the CC also predicted response to therapy. Specifically, lower FA in the motor CC at
baseline was negatively correlated with change in motor function following treatment with a
combined intervention consisting of tDCS with physical/ occupational therapy (Lindenberg et
al., 2012). However, the aforementioned studies were limited by small sample sizes, an almost
exclusive focus on the primary sensorimotor regions of the CC, and traditional clinical measures
of paretic hand motor impairment, which might not directly correspond to changes in CC
microstructure.
To address these limitations, the primary purpose of this study was to determine if CC
microstructure predicts bimanual motor performance in chronic stroke survivors, looking in a
larger sample of 41 chronic stroke survivors and examining all regions of the CC. We utilized a
65
bimanual task because we expected that the chronic persistence of lower FA in the non-
sensorimotor regions of CC might be detrimental to motor performance not captured by
traditional clinical measures of paretic hand motor impairment, which are thought to have
plateaued in the chronic phase (Krakauer & Carmichael, 2017). In particular, motor outcomes
assessed in the previous studies of CC were singularly focused on the paretic limb, e.g.,
impairment (UEFM,(Li et al., 2015; Stewart, Dewanjee, et al., 2017; Wang et al., 2012), self-
reported disability (NDS, SIS hand,(Li et al., 2015; Stewart, Dewanjee, et al., 2017), and
function, i.e., WMFT (Lindenberg et al., 2012), ARAT (Li et al., 2015). However, based on the
long-established evidence for the role of CC in interlimb coordination (Bonzano et al., 2008;
Caeyenberghs et al., 2011a; Fling et al., 2011; Fling & Seidler, 2012; Franz et al., 1996b;
Gooijers et al., 2013a; Sisti et al., 2012), one might expect that the lower FA in non-sensorimotor
callosal regions might be better reflected in the performance of tasks that preferentially engage
bi-hemispheric circuits, such as bimanual tasks. As noted earlier, studies in rodent models of
stroke suggest that the reactive synaptogenesis and reorganization in the intact hemisphere did
not simply emerge in response to repetitive motor activity of the healthy limb, but rather required
a skill learning process. Therefore, successful compensatory behaviors learned over time may
contribute to the transcallosal reorganization process and we expect would be associated with the
microstructural status of the non-sensorimotor regions of CC. To capture performance on such
use-dependent, learned motor behaviors, we unobtrusively observed natural arm use for two
components of a bimanual task (folding and inserting a letter into an envelope) and quantified
the time taken to complete this task.
In addition to studying this task performance in relation to the sensorimotor region of the
CC, we also examined non-sensorimotor regions of CC. Two candidates of special interest for
66
the control of bimanual skills based on previous evidence were the prefrontal region (CC1)
involved in higher-order planning and response selection (Baxter et al., 2000; Rowe et al., 2000),
and, the premotor and supplementary motor regions (CC2), involved in temporal sequencing
(Halsband et al., 1993; Kornysheva & Diedrichsen, 2014; Sadato et al., 1997). We hypothesized
that lower FA in not only the primary sensorimotor but also CC1 and CC2, would correspond
with poor performance on the bimanual task.
Finally, considering that following a stroke, there are widespread alterations in bilateral
brain excitability and brain health (Egorova et al., 2019; Gratton et al., 2012), it is plausible that
the callosal microstructure is also affected in regions directly neighboring the sensorimotor
cortices, such as CC2 and CC5. Some support for this hypothesis comes from a study by
Hayward and colleagues in which they found that in chronic stroke survivors, FA of the anterior
regions of the CC, specifically fibers connecting the prefrontal cortices, was positively correlated
with paretic hand motor impairment (Hayward et al., 2017). Although it does not directly
neighbor the primary motor cortex, the authors argued that the microstructure of the prefrontal
cortex could influence motor impairment via indirect connections through the premotor area.
Recently, similar to the study by Hayward and colleagues, (Pinter et al., 2020) reported that
lower FA in the callosal genu, which corresponds to the fibers connecting the prefrontal areas,
recorded within 72 hours after stroke (acute phase) along with a general disability measure
(modified Rankin scale) explained 53.5% of the variance in stroke recovery at 3 months. In the
same study, the authors also noted that compared to a control group, FA was lower for all
regions of the CC. Although a control group was not examined in the Hayward study, we
hypothesized that FA reductions across the CC, especially in the non-sensorimotor regions, as
reported by Pinter, may persist in the chronic phase. Here, we performed a secondary exploratory
67
analysis to examine this idea. Using both our own data and publicly available data, we compared
FA reductions in chronic stroke survivors that we collected with that of neurologically intact
controls from a publicly available dataset. The control dataset allowed us to more cleanly isolate
the effects of stroke from the more general effects of age, establishing a baseline from which the
stroke effects can be compared. Based on Pinter et al and Hayward et al., we hypothesized that
compared to age-similar controls, FA would be significantly reduced, beyond the general
reductions from aging, in all regions of the CC, including the non-sensorimotor regions.

Methods
Participants
Diffusion tensor imaging data for 41 chronic stroke survivors were available from a
Phase 2B randomized controlled trial (Dose Optimization for Stroke Evaluation,
ClinicalTrials.gov ID: NCT01749358) (C. Winstein et al., 2019b). Only baseline data from the
DOSE study were included in this analysis. These data were collected between 2012 and 2015 on
the Health Sciences Campus of the University of Southern California (USC).
For our exploratory analysis examining differences in the microstructural status of the CC
in chronic stroke survivors versus healthy controls, we used publicly available diffusion datasets
acquired in 24 age-similar older adults and 28 younger adults, matched closely for acquisition
parameters, including sequence, diffusion gradient strength and number of directions
(OpenNeuro.org ID: ds001242). These data were collected between 2016 and 2018 on the
University Park Campus of USC.
68
All individuals gave informed consent to participate in the two studies in accordance with
the 1964 Declaration of Helsinki and the guidelines of the Institutional Review Boards of the
respective campuses of USC where the data were collected.

Diffusion Imaging
Acquisition
The diffusion MRI scans in stroke survivors was acquired on a GE Signa Excite 3T
scanner using a single shot spin echo EPI pulse sequence with the following parameters: TR =
10,000 ms, TE = 88 ms, FoV = 256 mm, 75 axial slices of thickness = 2.0 mm, gradient strength
of 1000s/mm
2
in 64 diffusion gradient directions. This generated a 2 x 2 x 2 voxel size and a
matrix size of 128 x 128. A high-resolution structural T1-weighted image was acquired prior to
diffusion imaging using the gradient-echo (SPGR) sequence with the following parameters: TR =
24 ms, TE = 3.5 ms, flip angle = 20 ̊, FoV = 240 mm, and slice thickness = 1.2 mm with no gaps,
generating a matrix size of 197 x 233 x 189. Total time was approximately 20 minutes.
The younger and age-similar older control datasets were acquired on a Siemens 3T Trio
Total imaging matrix (Tim) system. A comparison of all scanner acquisition parameters is
provided in the Supplementary Material (I).

Preprocessing
Data pre-processing and analysis followed a standard pipeline using the FMRI Software
Library, FSL (Figure 1). Voxel-wise statistical analysis of the FA data was carried out using
TBSS (Tract-Based Spatial Statistics, (Smith et al., 2006), part of FSL (Smith et al., 2004).
Diffusion images were first preprocessed, including correction for eddy currents and motion-
69
related distortion, followed by brain extraction (BET2, (Jenkinson et al., 2005). Next, FA images
were created by fitting a tensor model to the preprocessed diffusion data using FDT (Smith,
2002). All subjects' FA data were then aligned into a common space (FMRIB58_FA) using the
nonlinear registration tool FNIRT (Andersson et al., 2007a, 2007b), which uses a b-spline
representation of the registration warp field (Rueckert et al., 1999). The FMRIB58_FA is a white
matter template generated from an average of 58 high-resolution, well-aligned FA images from
healthy adults and is in the same coordinate space as the 1mm MNI 152 template. Visual quality
checks (QC) were performed at every step and manual adjustments were made as necessary.
Lastly, the mean FA image was created and thinned to render an FA skeleton which represents
the centers of all tracts common to the group. Each subject's aligned FA data was then projected
onto this skeleton and a voxel-wise threshold FA of 0.2 is applied to remove any edge effects.
Figure 5.1. a) Diffusion pipeline including preprocessing, co-registration and identifying region of
interest (i.e., corpus callosum). (b) Parcellation of the corpus callosum using the geometric scheme
proposed by Hofer & Frahm (2006). (c) Table showing cortical regions connected by the fibers running
through each of the five CC parcels.
70
Assessment of Callosal Microstructure
The corpus callosum (CC) was defined as the primary region of interest. To assess
microstructural status of the CC, we analyzed diffusion images in the standard space, and used
the JHU white matter atlas to mask the CC (JHU ICBM-DTI-81 White-Matter Labels). The CC
was then segmented geometrically on the midsagittal plane into five subregions according to the
Hofer-Frahm parcellation scheme (Hofer & Frahm, 2006). Each of these segments correspond to
fibers connecting homotopic regions of the prefrontal (I), premotor and supplementary motor
(II), primary motor (III), primary somatosensory (IV), and parietal, temporo-occipital (V)
cortices, which in the standard MNI space, consists of 8851, 6871, 5803, 2619, 12729 voxels
respectively. Mask templates are available in the first author’s OSF repository: osf.io/7j9xe
Microstructural status was quantified as the fractional anisotropy (FA) index. The FA
index is a composite measure reflecting the 3-dimensional directional characteristics of diffusion
in each voxel, serving as a proxy for fiber orientation (Hagmann et al., 2006a; Pierpaoli &
Basser, 1996; Soares et al., 2013). It is computed as a normalized fraction of the eigenvalues
derived directly from voxel-wise fitted tensors, and ranges from 0 (isotropic diffusion, spherical
in shape) to 1 (anisotropic diffusion, ellipsoidal in shape). The FA composite measure works
particularly well for directionally homogenous, well-aligned fibers such as those of the CC,
especially after thresholding for edge effects. In a random subset of stroke survivors (n = 20), we
validated the FA index generated in the standard-space CC mask with those in the native FA
maps and found no difference in mean FA between the two spaces. Results from these
comparisons and related bootstrap analyses are provided in Supplementary Materials (II).
In stroke survivors only, we also computed tissue volume as an index of CC
macrostructure. To do this, individual CC masks were drawn in the native space of each
71
participant’s structural T1 image using ITK-SNAP (v. 3.8). We normalized CC volumes to
express them as a percentage of total white matter volume. To compute total white matter
volume, we performed tissue segmentation using FSL’s FAST routine with visual quality
checking to ensure that all of the viable white matter tissue, sparing the lesion, was identified in
the segmentation procedure.

Lesion Reconstruction
Stroke lesions were manually drawn on structural T1 images by trained personnel using
MRIcron, an open-source tool for segmentation and visualization (Rorden & Brett, 2000). A
detailed procedure has been described previously and all T1-weighted images and binarized
lesion masks are available as part of the ATLAS stroke lesion database, vR1.1 (Liew et al.,
2017). Lesion volume was calculated using FSL’s fslstats function. A lesion overlap image
among stroke survivors was generated using the fslmaths -add function and visualized in
FSLeyes (Jenkinson et al., 2012). Figure 2 shows lesion overlap among 41 chronic stroke
survivors.
Figure 5.2. Overlap of lesions for 41 chronic stroke survivors. Note that images were not flipped
and represent the actual side of unilateral stroke.
72
Bimanual Motor Performance
In conjunction with diffusion imaging, behavioral data for 33 of the 41 right-handed
stroke survivors were available for analysis. The behavioral paradigm has been described in
detail previously (Varghese et al., 2020). Briefly, participants were covertly observed as they
performed a natural bimanual task—the letter-envelope task—of the Actual Amount of Use Test.
Data were captured on video and analyzed offline to quantify whether participants chose one or
both hands and the time taken to complete the task at self-selected speed, i.e., movement time.
The letter-envelope task consisted of two components: folding the letter followed closely
by inserting the letter into the envelope. Start times were defined as the frame when initial
contact was made with the letter or envelope, and end times were defined as the frame when the
goal was accomplished, i.e., when the last fold was completed, or letter was fully inserted into
the envelope. Movement time (MT) was defined as the time elapsed between the start and end
time points and was determined for each component of the composite task. For the current
analysis, we analyzed MTs for those participants who chose a bimanual strategy.
We selected the letter-envelope task because a majority of the participants (31 out of 33)
chose a bimanual strategy for at least one of its components (see Supplementary Material II).
However, not every participant performed both components of the task bimanually. Regardless,
MT for every one of the 31 participants who chose a bimanual strategy were included in the
statistical analysis and allowed us to maintain as complete a set of matched brain imaging and
behavioral data as was possible.
Given that the strategy was self-selected, the speed was self-paced, and the testing itself
was conducted unbeknownst to the participants, performance on this task was largely
unconstrained, and served as a proxy for interlimb coordination as if it were in the real world,
73
even if qualitatively variable for each individual. In other words, it served as a reasonable
representation of well-learned stereotypical arm use behaviors that stroke survivors may have
deemed useful for successfully accomplishing every day bimanual tasks.

Statistical Analysis
All analyses were conducted using the R statistical computing package (version 3.5.1).
All continuous variables, age, chronicity, Upper Extremity Fugl-Meyer scores (UEFM), and
movement time, were assessed for normality. Distributions for chronicity and movement time
were positively skewed and so they were log-transformed. Assumptions for generalized linear
models, including linearity, equality of variance, independence and normality of errors were met
and model diagnostics, including leverage and multicollinearity of independent variables, were
tested when appropriate.

Relationship between callosal microstructure and bimanual behavior in chronic stroke
survivors
First, in chronic stroke survivors only, to determine the relationship between mean callosal FA
and bimanual MT, we used robust linear mixed effects regression of the form below:
A robust model was used instead of a simple linear mixed-effects (LMER) approach
because diagnostics revealed 1 participant whose MT had a very low value (2.8 seconds) which
was 1.77 SD away from and Cook’s distance of 0.03. As this point deviated substantially from
the rest of the participants, but its Cook’s distance was not so large as to warrant its removal, we
preserved it in our sample pool and used robust linear mixed effects. We suspected high
74
multicollinearity between the CC regions but found a somewhat small variance inflation factor of
1.14.

𝑙𝑜𝑔 (𝑀𝑇 ) ~ 𝑀𝑒𝑎𝑛 𝐹𝐴 + 𝑀𝑒𝑎𝑛 𝐹𝐴 : 𝐶𝐶 𝑟𝑒𝑔𝑖𝑜𝑛 + 𝑙𝑜𝑔 (𝐶 ℎ𝑟𝑜𝑛𝑖𝑐𝑖𝑡𝑦 ) + 𝐶𝐶 𝑣𝑜𝑙𝑢𝑚𝑒
+ (1|𝐶𝐶 𝑟𝑒𝑔𝑖𝑜𝑛 : 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 ) . . . (1)

In the above model, mean FA was the average FA across all voxels in a given region of
the CC. CC region was a dummy categorical variable with 5 levels, each to code for the 5
segments of the CC, with CC3 (motor) set as the reference level. We hypothesized that mean FA
would be significantly different from 0 for CC3 (primary motor) and CC4 (primary sensory), but
also for CC1 (prefrontal) and CC2 (premotor/supplementary motor). We tested this using
individual post-hoc t-tests of the estimated marginal trends. We also suspected that the slope of
this relationship might be moderated by CC region, so we included an interaction term to test
this. Pairwise comparisons of slopes for each CC region were conducted using Tukey’s HSD.
Finally, because a single value for MT per subject was repeated over five CC regions, there was
no reason to suspect MT to change over the levels of CC region. Instead, to estimate this additive
shift arising from subject- and CC region-wise differences in intercepts, we modeled the random
effects as an interaction between subject and CC region. By so doing, we were able to estimate
variances from the additive shifts for both subject and CC region: for instance, mean FA value
for CC5 could be higher than CC3 in subj#1 but lower in subj#5.
We arrived at the final model (eq. 1) by using a combined forward then backwards
stepwise approach, in which we tested for the confounding effects of age, sex, chronicity, side of
lesion, UEFM score, and normalized total CC volume by adding them to the base model that
75
consisted only of mean FA and CC region. We preserved any covariate that met a cut-off of p =
0.1 in a combined model. Then, from the combined model, we removed predictors that were not
significant (p < 0.05). Based on this selection process, only log-transformed chronicity and
normalized total CC volume were included in the above final model. Notably, by including
normalized total CC volume, we were able to take into account the likely loss in CC tissue
volume.
Lastly, to further corroborate our analysis, we asked whether traditional metrics of
unilateral structural damage provided any additional explanatory power to our final model above.
We did so by adding five measures of corticospinal tract integrity, lesion, and brain atrophy (B.
Kim, 2017; Liew et al., 2017): 1) CST FA asymmetry index, 2) CST lesion load, 3) lesion
volume, 4) lesion FA, and 5) ventricular volume asymmetry. All metrics, except lesion volume,
were computed previously by one of the authors (BK), and available for 29 of the 31 participants
for whom bimanual MT data were also available. As a result of unequal sample sizes however,
models could only be compared using marginal R
2
.
Comparing callosal microstructure between chronic stroke survivors and neurologically intact
adults
Second, to explore the effect of stroke on mean callosal FA, we used linear mixed effects
regression of the following form:

𝑀𝑒𝑎𝑛 𝐹𝐴 ~ 𝐺𝑟𝑜𝑢𝑝 + 𝐶𝐶 𝑟𝑒𝑔𝑖𝑜𝑛 + 𝐺𝑟𝑜𝑢𝑝 ∶ 𝐶𝐶 𝑟𝑒 𝑔𝑖𝑜𝑛 + (1|𝑠𝑢𝑏𝑗 ) + (1|𝑠𝑐𝑎𝑛𝑛𝑒𝑟 ∶ 𝑠𝑢𝑏𝑗 ) . . . (2)
Pairwise comparisons of estimated marginal means for each CC region were conducted
using Tukey’s HSD. Our hypothesis was that mean FA would be lower in stroke survivors
76
compared to age-similar adults. We suspected that while group effects would be largest for CC3
(motor) directly adjacent regions (e.g., CC2, premotor) would also show significant reductions in
FA. Given that our data were obtained from two different scanners, random effects were
estimated as random intercepts for both subject- and scanner-related variances.
Here again, we arrived at the final model (eq. 2) through the same process described
above, testing for the confounding effects of age and sex; neither met a cut-off p = 0.1, so were
removed from the above final model. Note that age was in fact partially embedded within the
grouping factor itself. However, because testing for age-related effects on FA was not the
primary purpose of this study, we only preserved age as a categorical variable. A supplementary
analysis of the relationship between age and FA is provided for the interested reader
(Supplementary Material IV).
A one-way ANOVA was used to compare age among the three groups (i.e., younger
controls, older controls, and stroke survivors) followed by pairwise comparisons using Tukey’s
HSD. Kruskal-Wallis test was used to compare the proportion of females and males among the
three groups. Significance was set at p = 0.05 and adjusted for multiple comparisons when
necessary.

Results
Of the 93 adults whose data were analyzed for this study, 41 were chronic stroke
survivors, and 52 were neurologically intact. The average age was 50.7 years and there were 64
males (68.8%) (see Table 5.1 for full details). Of the controls, 24 were older adults and 28 were
younger adults. There was a significant difference in age between the young controls and the two
older groups (F (2,90) = 156.73, p < 0.001)—older controls (∆ age = 43 years) and stroke
77
survivors (∆ age = 35 years). The stroke group was also younger than the age-similar control
group (∆ age = 8 years). There was a similar representation of sex among the three groups (H (2)
= 0.81, p = 0.66).
Table 5.1. Subject characteristics.

Within chronic stroke survivors, median chronicity was 1.9 years (IQR = 1.2–5.2) post
stroke and median score on the upper extremity Fugl-Meyer (UEFM) was 43 (IQR = 31–50),
indicating moderate impairment. Chronic stroke survivors consisted of 22 individuals with left
hemisphere stroke and 19 with right hemisphere stroke. There was no significant difference in
age (p = 0.196), sex (p = 529), chronicity (p = 0.409), or UEFM (p = 0.633) between the two
stroke groups. Lesion volume was slightly larger in those with right hemisphere strokes but not
significant (∆ mean = 1945.4 cc, p = 0.054). Table 5.1 provides demographic information.

78
On average across all stroke survivors, the lesion constituted < 0.05% (~11 voxels) of the
total CC volume, whereas voxels of the CC constituted < 0.2% (~2 voxels) of the total lesion
volume, confirming a very minor degree of direct injury to the CC (Figure 5.2). Individual
descriptions of lesion locations are provided in the Supplementary Material (III).

Lower callosal FA is associated with slower bimanual performance in chronic stroke
survivors.
After accounting for chronicity and total normalized CC volume, mean FA in all regions
of CC significantly predicted bimanual movement time in chronic stroke survivors, such that
lower (more isotropic) FA was associated with slower performance. To interpret values in Table
2, please note again that CC3 was the reference level (thus Mean FA in Table 5.2 is the slope for
CC3 and estimates for other levels are added to this estimate as described in the post-hoc
marginal trends).
Post-hoc tests of marginal trends revealed that compared to 0, slope was largest for CC3
(motor, b = -4.01 ± 0.87, p < 0.001), followed closely by CC2 (premotor, b = -3.85 ± 0.84, p <
0.001) and CC1 (prefrontal, b = -3.65 ± 0.79, p < 0.001). However, the slope for CC3 did not
significantly differ from CC1 and CC2 as observed in the interaction terms, Mean FA x CC1 and
Mean FA x CC2. The slopes were less steep for CC4 (sensory, b = -3.57 ± 0.78, p < 0.001) and
CC5 (parietal, temporo-occipital, b = -3.22 ± 0.70, p < 0.001) as observed in the interaction
terms, Mean FA x CC4 and Mean FA x CC5. However, post-hoc comparisons of all slopes
revealed that the slope of only CC5 was significantly smaller than CC3 (t = -3.64, p = 0.003) and
CC2 (t = -3.27, p = 0.012). Table 5.2 provides model estimates from robust regression and
Figure 5.3 illustrates the relationship using model fitted predictions.
79
Table 5.2. Robust mixed-effects regression coefficients from Model (1) to estimate relationship between
bimanual movement time and mean fractional anisotropy (FA) for five segmented regions of the CC
relative to the reference group (CC3, motor).

80
Figure 5.3. Callosal FA for each of the five regions plotted against log-transformed movement times for
the two bimanual tasks. Population regression lines from robust mixed effects regression are shown.

Finally, by individually including traditional measures of structural damage in our model,
we found that of the five CST and lesion metrics, three contributed a statistically significant
additive value to CC FA in predicting bimanual movement time: ventricular volume asymmetry,
CST lesion load, and lesion FA. When combined in the same model, marginal R
2
improved by
16.9%, from 20.4% to 37.3%; however, only lesion FA (t = -3.87, p < 0.001) and ventricular
volume asymmetry (t = 3.54, p < 0.001) remained significant. These results suggest that in
addition to callosal FA, lower FA in the white matter directly underlying the lesion and the
relative enlargement of the ipsilesional lateral ventricle predicted slower bimanual performance.
81
Model comparisons and model results with these two factors are provided in the Supplementary
Material (V).
Compared to neurologically intact adults, chronic stroke survivors exhibit lower FA in all
regions of the CC, except the splenium.
There was a significant interaction between group and CC region (F (8, 360) = 21.01, p <
0.001). Compared to neurologically intact older adults, mean FA was lower for all CC regions,
except the splenium (parietal, temporo-occipital region). Greatest decrements were seen for the
primary motor region, CC3 (ΔFA = 0.052, t = 4.84, p < 0.001), but was closely followed by the
premotor and supplementary motor, CC2 (ΔFA = 0.050, t = 4.69, p < 0.001), primary sensory,
CC4 (ΔFA = 0.034, t = 3.15, p = 0.005), and lastly prefrontal regions, CC1 (ΔFA = 0.029, t =
2.73, p = 0.02). Figure 5.4 illustrates this interaction using model estimated marginal means.

Figure 5.4. Model estimated marginal means for CC FA across the five regions along with individual
data points. * p < 0.05, ** p < 0.01, *** p < 0.001.
82
Discussion
There were two main findings in this study: First, callosal microstructure was
significantly associated with bimanual performance in chronic stroke survivors. Notably, a
significant relationship was observed not only with the primary sensorimotor regions (CC3), but
also regions of the premotor/supplementary motor (CC2) and prefrontal (CC1) regions. While
previous studies have reported on the association of callosal microstructure with unimanual
motor outcomes of performance, impairment, function, and general disability, this finding is, to
our knowledge, the first demonstration of a relationship between bimanual performance and
callosal microstructure in stroke survivors. Second, chronic stroke survivors presented with
significantly greater loss of callosal fiber orientation (lower mean FA), compared to
neurologically intact adults. This finding confirms and extends previous reports.
Primary motor, premotor and supplementary motor CC show strongest relationship with
bimanual performance in chronic stroke survivors
After accounting for callosal volume and the time after stroke, lower mean FA across the
CC was associated with longer movement times for a cooperative and sequential bimanual task.
As expected, the largest slope between bimanual MT and FA was seen for the primary motor
region of the CC (CC3), followed closely by the secondary motor regions (CC2). With stroke
lesions largely localized to the motor subcortical areas and descending motor pathways, the
involvement of the primary motor CC was not surprising. Microstructural changes in this region
of the callosum may be closely linked to its evolving role in regulating premovement inhibition
†

especially in the chronic phase.

†
One function of the callosal fibers in CC3 is to carry inhibitory signals between the motor cortices. This mutual
inhibition, exerted especially in the pre-movement phase of unimanual movements, regulates undue excitability of
the unengaged motor cortex and allows minimal interference in the outgoing command signal from the engaged one.
83
A novel demonstration in this study is the relationship between CC2, the rostral portion
of the body of the CC that connects homotopic premotor and supplementary motor regions, and
bimanual performance in stroke. The medial wall of the frontal lobe has been shown to be
causally involved in the control of self-initiated, self-paced cooperative bimanual movements
(Brinkman & Kuypers, 1973; Kazennikov et al., 1998; Preilowski, 1972; Stephan et al., 1999).
Patients who had undergone anterior callosotomy were significantly slower and the timing of
movement initiations with both limbs were more imprecise (Eliassen et al., 2000b) than in
controls. Preliowsky inferred that this may be because the anterior CC serves as a more direct
route for the sharing of motor corollary discharges in the frontal lobe, enabling faster bimanual
performance. The bimanual task we examined was cooperative, and participants were observed
covertly as they self-initiated a movement strategy and completed the task at self-selected speed.
Although not as extreme as callosotomy, it stands to reason that poor microstructural status of
the anterior callosum in stroke survivors, delays transcallosal exchange of corollary discharges
and slows bimanual performance on such a task.
The splenium (CC5) also showed a significant relationship with bimanual motor
performance, but the slope of this relationship was significantly less steep compared to CC2 and
CC3. Given that the task required processing of visual signals before and during performance, it
would make sense why lower FA of the posterior fibers of the CC correlates with slower
movement times. However, because CC5 consists of a heterogeneous fiber population
originating from the parietal, temporal and occipital regions, it is difficult to disambiguate if the
association with slower MT is due to deficits in primary visual processing or higher order

A prevailing hypothesis to explain persistence of paretic limb impairment in the chronic phase after stroke is a
failure to release premovement inhibition.

84
sensory integration. It is important to note that the splenium itself was largely spared from the
observed effects of stroke (more details in the next section), thus, the range of FA values
observed (0.74-0.87) was smaller than that in the body (0.49-0.78) and the genu (0.54-0.80) of
the CC. This may explain why the observed effects for CC5 were smaller than CC2 and CC3.
Another explanation is that our measure of movement time did not capture features of
performance that might better reflect functions of the posterior callosum, such as visuospatial
integration and accuracy (Franz et al., 1996b).
The effects of stroke on CC regions connecting the larger sensorimotor networks
Compared to neurologically intact controls, chronic stroke survivors showed significant
reduction in FA across the genu (CC1) and body (CC 2, 3, 4) of the corpus callosum, but no
difference was found in the splenium (CC5). We extended previous findings of lower callosal
FA in acute stroke (Pinter et al., 2020) survivors to the chronic phase, in those with mild to
moderate motor impairment (Hayward et al., 2017; Stewart, Dewanjee, et al., 2017), and in
pediatric hemiplegia (Hawe et al., 2013). Given that the CC was not directly lesioned (as
demonstrated by < 0.05% of overlap between the lesion and the CC), the reduction in FA across
the CC is evidence for transcallosal diaschisis.
Lower FA specifically in the non-primary sensorimotor regions, the prefrontal (CC1) and
premotor and supplementary motor regions (CC2), supported our a-priori hypothesis regarding
the involvement of non-primary sensorimotor regions. In fact, CC2 showed the second largest
magnitude of the effect of stroke, second only to the motor region of CC—a novel finding in this
study. Changes observed in CC1 and CC2 are especially interesting as they suggest that
transcallosal reorganization after stroke not only impacts the primary motor region, but also
constituent regions of the larger sensorimotor networks, involved in the control of complex
85
motor actions that require anticipatory motor planning and sequencing. Reductions in FA in the
primary motor and primary somatosensory regions of the CC were less surprising and generally
consistent with previous reports, with fairly similar effect sizes (ΔFA ~0.05) to those reported by
(Stewart, Dewanjee, et al., 2017)in mild-to-moderate chronic stroke survivors.
The lack of a group difference in the splenium is also interesting. One straightforward
explanation for this finding is that lesions were highly localized to the primary motor areas and
the descending motor pathways, thus impacting the body of the CC. In about 9 individuals in our
sample, lesions also involved frontal areas, and so could result in microstructural changes along
the genu of the CC. However, only about 5 of the 41 stroke survivors had any involvement of the
temporal or parietal cortex, and none had damage to the occipital cortex, thus sparing the
splenium. A related factor that complicates the interpretation of this finding is that the splenium
comprises of heterogeneous fiber populations connecting the parietal, temporal and occipital
regions. For this reason, unlike other regions of the CC, directional orientation of fibers are not
easily disambiguated and might make potential group differences in any one region less
detectable. Techniques to geometrically segment the splenium to delineate the different fiber
populations have been developed and implemented in adults without brain lesions. However,
because these techniques rely on probabilistic tractography, they might render unreliable results
in those with stroke due to lesion-induced tissue loss.
The effect of age and regional differences in CC
Although not central to this study, we also found a strong and expected effect of age and
an effect of CC region, in a pattern consistent with previously reported topographical
organization of the CC.
86
First, an exploration of the effect of age suggests that aging explained ~19% of the
unadjusted variance in mean FA across the CC (Supplementary Material IV). In our secondary
analysis where we modeled age as a categorical variable, we noted that stroke survivors were
approximately 8 years younger than the older controls, which could hypothetically lend an age-
related advantage to stroke survivors. This was not the case; in fact, stroke seemed to exacerbate
the effect of aging in all regions of the CC where a significant group effect was observed. This is
evidenced by most of the data points for stroke survivors lying below the least-squares line in the
plot (Supplementary Material IV). It also appears that the largest magnitude of the effect of age
in the control group was observed for the prefrontal region, aligning well with previously
reported structural changes in the gray and white matter in the frontal lobe and associations with
age-related decline in executive function, decision making, and movement planning (e.g.,
(Voineskos et al., 2012).
Second, in all three groups, across the 5 regions, we observed a specific pattern in that
FA values were largest for the most posterior regions of the CC (the splenium), followed by the
most anterior regions (the genu) and were smallest in the midbody. This is consistent with the
cytoarchitectonic properties (e.g., fiber density, size, and myelination) observed through
microscopic examination of the various regions of CC and validates the composite FA metric as
bearing good correspondence with such properties (Basser & Pierpaoli, 2011).
Contributions of other metrics of structural damage
We intended to focus all our planned analyses on the corpus callosum. However, the
callosal account of reorganization after stroke is only a partial one. As noted earlier, stroke-
related transcallosal changes occur as a result of a dual process: metabolic and degenerative
changes in the ipsilesional cortex and secondary reorganization in the contralesional cortex.
87
Whatever the changes in either hemisphere, the main assumption in this study is that the
evolving dynamics of interhemispheric relationship would manifest within the microstructure of
the primary commissure.
Taking into consideration the first of these processes, we included metrics of ipsilesional
structural damage in our regression model and found three measures to be of significant additive
value in explaining bimanual performance: the integrity of white matter directly underlying the
lesion (lesion FA), the relative enlargement of the ipsilesional to the contralesional ventricle, and
the direct injury to the corticospinal tract. The latter two have been previously shown to be
important predictors of unimanual motor outcomes.
Limitations and Future Directions
The use of control datasets acquired on a different scanner raises the issue of scanner-
related variability. It is possible that unknown differences in proprietary scanner technology may
have confounded observed group differences. Several attempts were made to address this issue:
first, the control dataset was comparable to our stroke data in terms of critical diffusion imaging
acquisition parameters (e.g., diffusion intensity, 1000s/mm2, directions, 64). Second, to offset
random variances by estimating scanner-related random effects in the mixed model. Finally, for
all 3 groups, we were able to reproduce expected regional differences in FA values across the
CC, and a range of FA values, especially in the sensorimotor CC, similar to those reported in
previous studies. This provides methodological validity to our data.
Known limitations of diffusion tensor fitting include interpretation of isotropic FA values
as poor directional orientation in voxels where multiple fiber populations intersect. This issue is
particularly pronounced with long distance projections, such as the corticospinal tracts, and is at
least in part bypassed by homogenous and homotopic fibers of the CC. Another limitation is that
88
we selected a very small subset of bimanual tasks—folding and inserting the letter—of a known
large repertoire of bimanual skills. We chose this task because we had previously found it to be
most likely to elicit a bimanual strategy (Varghese et al., 2020). The task also presented with the
opportunity to study naturalistic real-world performance in which participants are self-guided
and self-paced. However, to be able to generalize these findings, follow up studies are needed.
Complementary structural changes in the contralesional cortex may also be important but
were not assessed in this study. The extant role of the contralesional cortex is a matter of much
debate. Some suggest that the progressive involvement of compensatory tracts (e.g., cortico-
reticulospinal pathways) in the contralesional cortex through substitutionary use of the less-
affected arm, is maladaptive and may be detrimental toward behavioral restitution and recovery
of ipsilesional pathways. Others propose that such maladaptive changes do not emerge until later
after the stroke (Cirillo et al., 2020; Xu et al., 2019), and that secondary motor areas in the
contralesional cortex play an important role in the performance of complex motor tasks after
stroke (Hoyer & Celnik, 2011; Lotze et al., 2006) and must therefore be engaged. Future studies
could characterize the likely time- and task-dependent changes in contralesional white matter to
clarify its role in post-stroke recovery.
Finally, retrospective design and a relatively modest sample size are also issues. Whereas
correlational analysis is the current standard in research using structural imaging for brain-
behavior analysis, future work that extends structural diffusion imaging to multi-modal imaging
including functional MRI along with larger samples and a prospective design might reveal new
insights into transcallosal diaschisis after stroke in humans.

89
Conclusion
Findings of this study lead us to conclude that in mild-to-moderate chronic stroke
survivors with relatively localized lesions to the motor areas, callosal microstructure can be
expected to change not only in the primary sensorimotor region, but also in the premotor,
supplementary motor and prefrontal regions. These remote widespread changes in the callosal
genu and body are likely to impact performance on cooperative bimanual tasks that require
precise and interdependent coordination of the hands.

90
Supplements

I. Comparison of DTI acquisition parameters across the groups.
Control
(n = 52)
Stroke
(n = 41)
Acquired 2016 – 2018 2012 – 2015
Released 2018 2017
MR scanner Siemens 3T Trio GE Signa Excite 3T
Sequence EPI Spin-echo EPI (single shot)
TR/TE (ms) 10,000/98 10,000/88
FOV (mm) 256 x 256 x 120 256 x 256 x 132
Matrix Dimensions 128 x 128 x 60 128 x 128
Slices 96 59
Voxel Size 2 x 2 x 2 1 x 1 x 2
b-values (s/mm
2
) 1000 1000
# of directions 64 64
DOI 10.18112/openneuro.ds001242.v1.0.0 TBD

91
II. Comparison of mean FA between native and standard space.
There was no difference in mean FA between native and standard space (n = 20). A. Subject-
wise comparisons of mean FA (t = 0.04, p = 0.97). Mean and confidence intervals of difference
in FA between native and standard space after 1000 iterations of bootstrapping is 0.00045 (-
0.021, 0.022). B. QQ plots of bootstrapped t-statistic plotted against theoretical t-distribution.
Mean and confidence intervals of t-statistic after bootstrapping is 0.053 (-1.93, 2.04). C.
Bootstrapped t overlaps with a theoretical t distribution which centers on 0. C. Note that for this
comparison, mean of all voxels in the standard space were used (not just those included after
TBSS thresholding).

92
III. Detailed lesion description for all stroke survivors, and availability of bimanual MT and
other structural metrics.

93
IV. Relationship between age and mean FA.
A simple unadjusted correlation and linear regression was done post-hoc to explore the
effects of age on mean FA across the 93 participants. The regression equation displayed on
the plot below shows that from an average FA value of 0.81 across participants, mean FA
decreases by a factor of 0.00095 for every year. As can be seen in this plot, most of the data
points for stroke survivors lying below the least-squares line.

94
V. Additive value of other metrics of structural damage to predict bimanual performance.
Left-hand column is model without other structural damage metrics and right-hand column is
the model with these metrics.
As seen above, lateral ventricular volume asymmetry (Lat. Ventr. AI) and Lesion FA provided
statistically significant additive value to CC metrics alone in explaining bimanual performance.

95
CHAPTER 6: Flexibility of responsibility assignment for a redundant bimanual task is
limited to some extent by lateralized motor control processes

Abstract
Skilled human action is seldom unimanual and often involves coordination of both hands. A
critical aspect of bimanual coordination is responsibility assignment. How does the nervous
system assign responsibility when an error occurs? The optimal feedback control theory suggests
that responsibility assignment is a flexible process such that errors are assigned to and corrected
for by the limb that is most likely to produce those errors. In right-handed adults, this is often the
less-skilled, non-dominant left limb. Interestingly, the flexibility of this process can be probed by
examining corrections made by each limb after they have acquired alternative use-dependent
experiences, e.g., if the left limb became more skilled, experiencing fewer errors or if the right
limb became less skilled, experiencing more frequent errors. Such is the case of stroke affecting
the right side of the body, wherein we would predict that the left limb corrects less while the
paretic right limb corrects more for task error. In this study, we tested this prediction in 20
individuals with an intact sensorimotor system, as well as 23 chronic stroke survivors (12 right
hemiparesis). Consistent with previous studies, correction gains were asymmetric between the
limbs in the non-disabled young control group such that the left limb corrected more than the
right limb. Our data also supported our predictions in those with right hemiparesis, but not in
those with left hemiparesis. Those with right hemiparesis not only corrected more with their
paretic right limb within a trial, but also corrected less with their now more skilled left limb.
They also systematically adapted to these errors in a feedforward manner over trials. The extent
of correction in stroke survivors did not appear to vary with the degree of motor impairment of
96
the paretic extremity. These findings lead us to conclude that responsibility assignment is not
entirely flexible but may be limited to some extent by the hemispheric specialization of motor
control processes.

Keywords: redundant bimanual, optimal feedback, hemispheric specialization
97
Introduction
Skilled human action is seldom unimanual and often involves the precise coordination of
both hands. As in any task, an important component of successful performance of bimanual tasks
is error correction. To correct for errors in bimanual performance, the nervous system faces a
unique problem of redundancy, in that it must first be able to appropriately assign this error to
either effector. How does the nervous system assign responsibility when an error occurs? Current
theory suggests that responsibility assignment is a flexible process such that errors are assigned
to and corrected for by the limb that is most likely to produce those errors (Todorov & Jordan,
2002).
Neurologically intact right-handed adults consistently showed a greater tendency to
correct with their non-dominant left limb and did so faster compared to the right limb. This
asymmetry between hands has been demonstrated to hold true across different perturbations
including visual rotations (White & Diedrichsen, 2010), target displacements (White &
Diedrichsen, 2010), constant (Mutha & Sainburg, 2009) and velocity-dependent force fields
(Diedrichsen, 2007; Diedrichsen & Dowling, 2009) and other mechanical perturbations (Schaffer
& Sainburg, 2021). Diedrichsen and colleagues argued that this may reflect a high default setting
on feedback-dependent control gains for this limb owing to a greater tendency of the non-
dominant limb to be more error-prone. To test the assumption that control gains are modified
based on recent experience of errors, they examined corrections made by each limb after it had
acquired alternative experiences. In other words, after the left limb became more skilled,
experiencing fewer errors than the right, or after the right limb became less skilled, experiencing
more frequent errors than the left. They found that unimanual exposure of the right limb to
98
greater sensory prediction or target errors
#
led to an increase in the control gain for the right limb
and a concurrent decrease for the left limb, in the respective error testing condition, suggesting
that the error assignment process may be flexible. However, the size of these effects was not
nearly large enough so as to reverse each hand’s default gain setting as would be predicted, and
in fact continued to be higher in the left hand than right for the target error condition—i.e., when
feedback-based corrections were warranted. It could be that the training was not long or intense
enough to produce more robust reversal between the hands. Nonetheless, this somewhat
anomalous observation is intriguing.
An alternative explanation for this observation comes from a prevailing model of motor
lateralization, known as the dynamic dominance hypothesis (Sainburg, 2002). According to this
hypothesis, the sensorimotor control of the two hands are specialized, such that the left hand,
primarily under the control of the right hemisphere, is more finely tuned for feedback-mediated
impedance control which allows it to respond effectively to unstable and unpredictable
environments (R. Sainburg, 2002). It is this specialization that likely manifests itself in the left
hand’s predominant role as the stabilizing hand in a number of bimanual tasks. A recent study
demonstrated that sudden and unexpected braking of each limb in its path as the two limbs
jointly controlled a virtual symmetric object, not only elicited an EMG response from the
perturbed limb, but also a significant response from the contralateral (non-dominant) left limb
when the (dominant) right limb was perturbed (Schaffer & Sainburg, 2021). This finding is
compatible with earlier observations from Diedrichsen’s group supporting the claim that the left

#
Sensory prediction error: an internal error that results from a mismatch between the predicted and actual state
of the limb. This type of error information is available relatively rapidly and does not rely on afferent feedback.
Target error: an external error that results from a mismatch between the desired and actual state of the limb. This
type of error information is available often late into the movement and relies on afferent feedback, such that a
correction cannot be predicted and programmed in advance but may be used to inform the next trial.
99
limb may have a high feedback gain setting. However, this study did not address each limb’s
ability to flexibly modulate its response after training or alternative use-dependent experiences,
hence whether
Taken together, whether the process of error assignment and subsequent high feedback
control gains is entirely flexible and so reversible between the hands or if they are limited to
some extent by lateralized motor processes remains unclear. One way to disambiguate these
processes is by studying them in a clinical model of stroke-induced chronic contralateral
weakness and loss of control (Mani et al., 2013a; Schaefer et al., 2012). Control deficits in the
contralesional limb (even in mildly impaired survivors), e.g., more errors and slower movements,
persist in the chronic phase after stroke and is often accompanied by long durations of functional
use and new skills learning with the ipsilesional limb. The stark difference in performance
between the limbs renders the chronic stroke model uniquely suited for the study of
responsibility assignment. We posited that if the responsibility assignment process is entirely
flexible, theory would predict that regardless of the hemisphere of damage, the paretic limb,
which is now significantly more prone to errors, would rely on sensory feedback to make
corrections and consequently show high control gains, whereas the now well-practiced, more
skilled non-paretic limb would show smaller gains. Conversely, if the feedback-mediated gain
settings are at least to some extent a feature of a specialized right hemisphere-left hand system,
then feedback-mediated gains will only be modified when the putative right hemispheric circuits
are intact, i.e., in those with damage to the left hemisphere and right arm weakness. Conversely,
in those with damage to the right hemisphere, gains should be smaller for both limbs.
We began by reproducing the asymmetric control gains observed in healthy non-disabled
adults as reported by Diedrichsen and colleagues with some notable exceptions in the paradigm.
100
First, rather than introducing an external perturbation, we relied on the mechanical properties of
the movement to induce a deviation in the movement path. This became possible because arm
movements were not restricted along a linear path by a robotic manipulandum but were free to
move in two dimensions on an air sled. Given that inertial resistance to the hand’s acceleration
varies in a direction-dependent manner (Gordon, Ghilardi, Cooper, et al., 1994), we expected
endpoint distribution to be anisotropic with independent variability along both, the direction and
extent, dimensions of movements (Gordon, Ghilardi, & Ghez, 1994). Furthermore, the left hand
is known to produce comparatively larger directional deviations (Gandrey et al., 2013). This
tendency of the left hand to move along a curved path may stem from principles of optimality as
it has been suggested that in certain instances, moving in such a way might minimize the cost
related to the control policy (Izawa et al., 2008). Thus, we expected that in order to maintain the
cursor’s movement along a straight path, deviations in the left hand’s path would induce
equivalent directional compensation in the right hand in the opposite direction. We demonstrate
this by showing that there is significant negative correlation between the x-coordinate of the
position of the two hands, i.e., in the direction orthogonal to that of the movement, and this
relationship is significant only for the one-cursor condition, in which the hands jointly control a
single cursor, but not the two-cursor control condition in which each hand moves its own cursor.
Temporal coupling, on the other hand, is preserved and strong for both conditions.
Next, to demonstrate that the left limb gains are significantly higher than the right limb’s,
we computed the change in direction of each hand in the early phase of movement. We chose the
time of peak acceleration and that of peak velocity as the critical time points to compute change
in direction of the hands as these were the most reliable markers of early phase changes in stroke
survivors (see Methods for details). By correlating the change in hand direction with the cursor’s
101
direction at peak velocity, we found that as expected, correction gains were asymmetric between
the hands with higher gains for the left hand compared to the right. Because participants were
given feedback about the cursor’s position at the end of each movement, we assumed that this
additional information could be used to explicitly modify planning signals for the next
movements (van Beers, 2009). A model of such trial-by-trial learning in the absence of
perturbations but presence of endpoint feedback used to modify planning signals has been
proposed by van Beers, 2009. We complemented our analysis of cursor control gains with
statistical analysis of trial-by-trial learning by fitting an autocorrelation (ACF) function with a
lag of 1 to each individual’s directional error. We found that like the control gains, the learning
rate (ACF coefficient) was also significantly higher for the left compared to the right hand.
Finally, to test our prediction that the flexibility of the responsibility assignment process
is partly limited by specialized control processes, we included two groups of chronic stroke
survivors who suffered unilateral injury to cortico-subcortical circuits and descending motor
pathways and presented with at least some degree of paresis to the contralateral side of the body.
Feedback-mediated correction gains in the paretic hand were found to be higher only in
individuals in whom the left hemisphere was damaged (right hemiparesis), but the right
hemisphere was intact. The paretic right hand also systematically adapted to these errors in a
feedforward manner over trials. The extent of correction in stroke survivors varied to some
extent with the degree of motor impairment of the paretic extremity.
These results raise the possibility that responsibility assignment is not entirely flexible
but may be limited to some extent by the hemispheric specialization of motor control processes.
Lesion studies in humans (Mutha et al., 2011) and recent causal evidence in macaques (Takei et
al., 2021) shows that right hemisphere regions such as PMd and posterior parietal cortices may
102
have an important role in sensory integration, which is necessary for generating feedback-
mediated corrections. These processes may be impaired in those with right hemisphere stroke,
but not in those with left hemisphere stroke.

Methods
Participants
Twenty young adults and 23 chronic stroke survivors (12 right-paresis, 11 left-paresis)
participated in this study. All participants gave informed consent to participate in accordance
with the guidelines of the Institutional Review Boards for the Health Sciences Campus of the
University of Southern California (HS-18-01024). Descriptive information on the three groups
(young adults, right-paresis, and left-paresis) are shown in Table 1.
Of the young adults, 9 were male, whereas among the chronic stroke survivors, 15 were
male; thus, males were over-represented in the stroke group. All participants were right-handed
(mean ± SD EHI = 0.89 ± 0.15). Stroke survivors were significantly older than young adults (t
(1, 42) = 14.18, p < 0.001). Visual bias assessed using a computerized line bisection test was
found to be within normal range (0.07 ± 4.45%) for all participants (norm: 6%, Schenkenberg et
al., 1980). A detailed participant-wise breakdown of the Upper Extremity Fugl-Meyer Motor and
Sensory assessment is provided in the Supplementary Material.
103
Table 6.1. Participant characteristics.

Experimental Setup and Task
Participants performed visually cued target reaching in two conditions: bimanual two-
cursor, and bimanual one-cursor condition. The one-cursor condition was the experimental
condition. For the one-cursor condition, a single start circle (1 cm in diameter) and 3-ring
concentric target circle (3 cm in diameter) were displayed (represented in Figure 1), and a single
cursor (1 cm in diameter) was positioned at the spatial average of the two hands. For the two-
cursor position, two start circles (positioned 12 cm away on each side from the center) and two
targets were displayed. Targets were positioned in the forward direction at two distances: 5 and
15 cm but were combined for analysis.

104
Figure 6.1. Task setup and conditions.

Participants began by positioning their hands in the start circle for 500 ms, after which
they received a visual warning cue (circle edges become bold). Then, after a variable warning
period of 350 to 500 ms, an auditory tone served as a cue to move. Online veridical visual
feedback of the cursor was provided at the beginning of the trial to guide movement of hands
into the start circle but became unavailable during the reaching movement. Participants were
asked to reach to the target and hold their position at the target until they received feedback.
Instructions emphasized speed and accuracy, but not interlimb synchrony. Cursor feedback was
available again at the end of the movement, which was defined as when real-time velocity of
hand <= 0.02 m/s. Participants were also provided summary feedback concerning final positional
accuracy at the end of the trial as a graded numeric and audiovisual score (10 points: cursor
within the innermost ring of the target, i.e., < 1 cm, target turns green with high pitch tone; 3
points: cursor within middle ring of the target, i.e., between 1 to 2 cm, target turns blue with
medium pitch tone; 1 point: cursor within the outermost ring of the target, i.e., 2 to 3 cm, target
105
turns gray with a low pitch tone). Audiovisual feedback lasted for 1 second after which the next
trial began.

Data Processing
Raw position data were sampled at 116Hz and processed offline in MATLAB (2019a).
Processing involved lowpass applying a 4
th
order Butterworth filter with a cutoff of 10 Hz and
differentiated to yield tangential velocity and acceleration. Tangential velocity profiles were bell
shaped with a single clearly defined peak for all trials in the control and stroke groups. Next,
critical time points of the movement were defined on the velocity and acceleration profiles.
Movement onset was defined as the first point in time at which there is a sustained
deflection in the amplitude of tangential velocity above 0. It was determined by searching
backwards in time for a local minimum from the peak of tangential velocity. In young adults,
movement offset was defined as the time point after peak tangential velocity where the amplitude
returned to 0. Similar to onsets, it was determined by searching forwards in time for a local
minimum from the peak of tangential velocity. In stroke survivors, movement offsets are not
easily determined due to multiple peaks, drifts, so a combined approach was used with local
minimum and threshold at least 5% below peak speed amplitude.
As a proxy for the planned direction of movement that is as yet unaltered by
proprioceptive or visual feedback, we chose the time of positive peak acceleration, instead of a
fixed point in time point or a fixed distance. We did so 1) to avoid a bias from known properties
of scaling of amplitude and timing when moving to two different distances, as was the case in
this study, and 2) because the arms were on the air sled, they were not perfectly still, moving
slightly within the start circle, along the axis perpendicular to the target direction, rendering
106
movement direction at onsets to be unreliable. As a proxy for a later point in the movement
which may or may not be influenced by proprioceptive feedback, we chose the time of peak
velocity. The time of peak velocity also naturally proved to be a far more reliable timepoint in
chronic stroke survivors compared to movement offsets and hence was chosen as the timepoint
when sensory feedback is likely available. For each trial, the difference between movement
direction at peak acceleration and that at peak velocity for each hand served as a measure of
directional adjustment (Δθ
𝐻𝑎𝑛𝑑 ).
All analyses, described in detail in the following sections were done in R (3.5.1). We
followed an outlier removal procedure in which for each participant condition, if values for peak
velocity, Δθ
𝐻𝑎𝑛𝑑 , or directional errors at peak acceleration or peak velocity, were 3SD away (±)
from the mean these were removed from the analyses. Based on this procedure, 56 out of 5160
trials (1.08%) across all subjects were removed.

Analysis 1: The contribution of each hand to cursor position as a proxy for control gain
and responsibility assignment
The goal of this analysis was to determine the contribution of each hand’s change in
position towards the cursor’s deviation from the center. For each trial we determined the initial
directional error of the cursor (at peak acceleration) and the change in direction of each hand in
the early phase of movement, i.e., initial direction of the movement (at peak acceleration) to later
in the movement (at peak velocity). Trials in which directional error or directional change
exceeded ± 3 SD were defined as outliers and removed from subsequent analysis. Of the pool of
5160 movement trials in 43 adults, 1.163%, or 80 trials constituted outliers and were removed.
To quantify the contribution of each hand to the cursor’s position, we estimated the slope of the
107
regression line between the cursor’s initial direction and change in each hand’s direction:
θ
𝐶𝑢𝑟𝑠𝑜𝑟 ~ 𝑎 + 𝑔 𝐻𝑎𝑛𝑑 ∗ Δθ
𝐻𝑎𝑛𝑑 . Similar to White and Diedrichsen, the slope of the regression
line (𝑔 𝐻𝑎𝑛𝑑 ) was the proxy for the adjustment gain for each hand. As an index of relative
contribution of each hand, we computed the gain asymmetry index as follows: 𝑔 𝑎𝑠𝑦𝑚 =
|𝑔 𝑟𝑖𝑔 ℎ𝑡 |
|𝑔 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑔 𝑙𝑒𝑓𝑡 |
. Thus, values closer to 1 indicate greater right-hand contribution to the cursor’s
position whereas values closer to 0 indicate greater left-hand contribution, and values closer to
0.5 indicate equal contribution of the hands.
To compare the stroke groups with the controls while accounting for within-subject and
within-group variability, we used a mixed effects model of the following form with random
slopes and intercepts:
θ
𝐶𝑢𝑟𝑠𝑜𝑟 = 1 + 𝐺𝑟𝑝 + (𝐺𝑟𝑝 𝑥 𝐻𝑎𝑛𝑑 𝑥 Δθ) + (𝐺𝑟𝑝 𝑥 𝐻𝑎𝑛𝑑 𝑥 Δθ | 𝑆𝑢𝑏𝑗 )

Analysis 2: Trial-by-Trial Learning
The central premise of this analysis is that if a copy of the (directional) error is stored, we
might expect the limb to which the error is assigned to also learn over trials (van Beers, 2009).
Therefore, the goal of this analysis was to determine if each limb adapts to the error on a given
trial, by modifying their response on the next. However, given that errors are large on the first
few trials, the first four trials of the 20-trial block were eliminated to avoid bias from a known
early fast learning process (a procedure for determining the number of trials to be excluded is
described later in this section).
Then, to estimate the trial-by-trial slow adaptive response, we fit an autocorrelation
function to the directional error at peak velocity of the last 16 trials for each hand for each
participant. The idea in this type of an analysis is that in the absence of external perturbations
108
and without online visual feedback, but with only endpoint feedback available, the first (entirely
planned) aim point is generally far away from the target. But as the trials proceed, movements
are aimed more accurately, both in extent and direction, so that they become more closely
clustered together. This close clustering is not random, but rather driven from the error on the
previous trial and its endpoint feedback. Thus, an ACF (lag=1) is positive when the direction of
each hand at peak velocity of consecutive movements tend to be close together, whereas it is
negative when consecutive hand directions tend to be far apart, on opposite sides of the mean
direction as if oscillating between two extremes.
In order to determine how many trials needed to be excluded, we determined the time
constant of the learning curve given by the following exponential decay function: (𝑎 −
𝑏 ) 𝑒 𝑥𝑝 (−𝑡 /𝑡 𝑐 ) + 𝑏 , where 𝑡 is the trial number, 𝑎 and 𝑏 are constants and 𝑡 𝑐 is the time
constant. Due to the anisotropy of movement endpoints, variance in the extent dimension of the
movement is greater than that in the directional dimension (Gordon, Ghilardi, Cooper, et al.,
1994; van Beers, 2009). Fitting learning curves without accounting for these differences in
variance in the two dimensions would therefore be biased by the extent dimension. Therefore, to
account for these differences, learning curves were fit to a weighted, normalized distance
measure, the Mahalanobis distance.

Results
We performed an experiment in which adult humans with and without unilateral neurological
injury from stroke performed bimanual reaches such that the two hands moved independently to
control two cursors or jointly to control a single cursor. Our multi-fold analyses were aimed at
understanding whether the process of error assignment and consequent high feedback correction
109
gains is entirely flexible and therefore reversible between the hands by virtue of extensive use-
dependent alternative experiences evident in chronic stroke survivors, or if they are limited to
some extent by lateralized motor processes.

Task-dependent compensation is observed for one- but not two-cursor condition.
In order to ensure that our experimental one-cursor condition worked in the way we
expected, we first tested our assumption that deviations in the left hand’s movement path would
induce equivalent directional compensation in the right hand in the opposite direction. If this
assumption holds up, we should find a significant negative correlation between the x-coordinate
of the position of the two hands at peak velocity only in the one-cursor condition in which the
hands would need to co-vary their positions to compensate for each other, but not in the two-
cursor condition wherein such compensation is not required by the task.
In our sample of young controls (n = 20), our data supported this assumption. We found
that there was a significant negative correlation (R
2
= 0.79, across participants) only for the one-
cursor condition but not for the two-cursor condition (R
2
= 0.08, across participants) (Figure 2B).
We further conducted individual-level correlations and compared the distributions between the
one-cursor and two-cursor condition using a t-test. Here again, we found that the distribution of
correlation coefficients for the one-cursor condition was significantly more negative than the
two-cursor condition (t = 5.63, p < 0.001) (Figure 2C).
These findings serve as a critical check of our experimental manipulation and suggsests
that directional compensation is in fact task-dependent and observed only for the experimental
one-cursor condition. The focus of our next analysis therefore shifts from between-hand
110
correlations to hand-cursor correlations in the one-cursor condition to examine contributions of
each hand to the cursor’s position.

Figure 6.2. A. Single canonical example trial showing paths of left (blue) and right (red) hands during
one-cursor and two-cursor conditions. Starting positions of the hands were shifted by a fixed offset for
display purpose only. Also, note that the two-cursor condition has two targets and two start circles. B.
Scatter plot of mean-centered X positions for the two hands across all participants for the one- and two-
cursor conditions. Line of equivalence shown in black. Ellipses contain 95% of the points in the binormal
distribution. Notice that due to the differences in starting position of the two hands, mean of the
distribution of points in each condition is shifted. C. Distribution of correlation coefficients across all
young adult participants (n = 20) showing that between-hand correlation was significantly more negative
in the one-cursor compared to the two-cursor condition. Collectively, these plots demonstrate that
position along the horizontal (X) axis show task-dependent compensation between hands only in the one-
cursor but not the two-cursor condition.

One
Cursor
Two
Cursors
Left
Right
Cursor
target
targets
A B C
5 cm
111
Within-trial correction gains are asymmetric between the hands.
As can be observed in the scatter of the points from the one-cursor condition in Figure 2B
(dark brown), the left hand’s position only accounted for 79% of the variance in the right hand’s
position. Thus, even though close to the line of equivalence, the slope of this regression was ≠ 1,
which suggests that the contributions of the hands were not 50-50 as prescribed by the one-
cursor task. Given that the cursor’s position is the spatial average of the two hands, by definition
this would lead to an error in the cursor’s estimated position. Each hand would therefore need to
change its path to correct for this cursor error.
To examine how much each hand needed to change its position to correct for the cursor’s
error, we regressed the change in the direction of each hand in the early phase of movement
(Figure 3A) against the cursor’s initial directional error and used the slope of this regression line
as an index of the correction gain.
We found that consistent with previous work, the correction gains were asymmetric
between the hands. Compared to the right hand, the left showed a significantly more negative
slope (gleft = -0.53 ± 0.06, p < 0.001; gright = -0.13 ± 0.08, p = 0.1). Figure 3C shows the
difference in slopes between the population-level regression lines for the two hands using mixed
effects models (for model output see Supplementary Material). Individual slopes from the
random effects confirms the difference in slopes between the hands (Figure 3D). The relative
gain asymmetry index (
|𝑔 𝑟𝑖𝑔 ℎ𝑡 |
|𝑔 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑔 𝑙𝑒𝑓𝑡 |
) was also significantly smaller than 0.5 (0.23 ± 0.11, t(19)
= –10.84, p < 0.001) indicating a bias toward the left hand.
This finding suggests that in neurologically intact adults, correction gains are asymmetric
between the hands with the left hand changing its direction significantly more to account for the
cursor’s directional error.
112
Figure 6.3. A. Example trial showing measure of directional error. We determined the initial directional
error of the cursor (at peak acceleration) and the change in direction of each hand in the early phase of
movement, i.e., initial direction of the movement (at peak acceleration) to later in the movement (at peak
velocity). B. Contribution of each hand to cursor’s initial directional error for one non-disabled right-
handed control participant. As evident here, slope was significantly more negative for the left hand
compared to the right hand. C. Scatter plot with population-level regression lines (solid) and individual
regression lines (n = 20) for the left and right hands in young controls. D. Individual slopes derived from
random effects from the full mixed-effects model.

113
Responsibility assignment in stroke survivors is flexibly reversed only in those with right
hemiparesis but not in those with left hemiparesis.
The previous results confirmed that correction gains are asymmetric between the hands
with the left hand showing larger correction gains compared to the right hand. To test our
hypothesis related to the flexibility of responsibility assignment, we examined the same
correction gains in individuals with left and right hemiparesis. We predicted that if this process is
entirely flexible, we would see larger correction gains in the paretic hand regardless of the side
of paresis. Conversely, if the flexibility of this process is somewhat limited by its reliance on the
specialized feedback-mediated processes localized in the right hemisphere, then we would
observe correction gains to increase for the paretic hand only in individuals in whom the right
hemisphere and therefore its functions are preserved.
We found that only in those with right hemiparesis (left hemisphere damage) correction
gains were reversed between the hands. Compared to the left (non-paretic hand), the right
(paretic) hand showed a significantly more negative slope (gright = -0.3 ± 0.05, p < 0.001; gleft = -
0.13 ± 0.09, p = 0.22). The non-paretic hand gain was more variable than the paretic hand.
Figure 4A shows the difference in slopes between the population-level regression lines for the
two hands using mixed effects models (for model output see Supplementary Material).
Individual slopes from the random effects confirms the difference in slopes between the hands
(Figure 4A boxplots). 3 of the 12 individuals showed a pattern similar to controls and appeared
to have very mild motor impairments indicated by a relatively high score on UEFM (≥ 60). The
relative gain asymmetry index (
|𝑔 𝑟𝑖𝑔 ℎ𝑡 |
|𝑔 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑔 𝑙𝑒𝑓𝑡 |
) was also significantly greater than 0.5 (0.65 ±
0.21, z = 2.42) indicating a bias toward the right hand.
114
Conversely, in those with left hemiparesis, correction gains did not differ between the
hands but was slightly more negative for the right hand (gright = -0.24 ± 0.11, p = 0.034; gleft = -
0.18 ± 0.06, p = 0.22). Once again, the non-paretic hand gain was more variable than the paretic
hand. Given that gains were negative, albeit reduced, for both hands the relative gain asymmetry
was not significantly different from 0.5 (0.57 ± 0.20, z = 1.21) suggesting somewhat equal
contributions from both hands. Individual responses were highly variable.
The relative asymmetry index for both stroke groups was significantly higher than the
control group (F (2,40) = 27.9, p< 0.001, Figure 4C) and appeared to vary to some extent with
the degree of motor impairment of the paretic extremity (2
nd
order fit shown in Figure D) but did
not reach significance.
Together, these findings suggest that correction gains are increased for the limb more
prone to error, i.e., the paretic limb, only in those with right hemiparesis but not in those with left
hemiparesis. The absence of such an increase in gains in those with left hemiparesis might
suggest that the process of error assignment is not entirely flexible but limited by hemisphere-
specific control processes. In this case, deficits in feedback-mediated control in those with left
hemiparesis may explain why both hands show deficient correction for the cursor’s error.

Figure 6.4. A. Right hemiparesis (RHP) gains: Scatter plot with population-level regression lines (solid)
and individual regression lines (n = 12) for the left and right hands in chronic stroke survivors with right
hemiparesis. Adjacent boxplots represent individual slopes derived from random effects from the full
mixed-effects model showing reversal in correction gains, which were larger (more negative) for the
paretic right hand. B. Left hemiparesis (LHP) gains: Scatter plot with population-level regression lines
(solid) and individual regression lines (n = 11) for the left and right hands in chronic stroke survivors with
left hemiparesis. Adjacent boxplots represent individual slopes derived from random effects from the full
115
mixed-effects model. Paretic left hand in individuals with left hemiparesis did not show the same increase
in correction gains. C. Asymmetry index: A comparison of gain asymmetry index, g asym (
|𝑔 𝑟𝑖𝑔 ℎ𝑡 |
|𝑔 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑔 𝑙𝑒 𝑓𝑡
|
)
showing a strong left-ward bias in controls (green). Compared to controls, balance between hands shifts
to the right hand in individuals with right hemiparesis (as expected) as well as in those with left
hemiparesis (unexpected finding). D. Scatterplot and second-order non-linear fit (g asym ~ uefm + uefm
2
)
shows a weak relationship between the degree of motor impairment in the paretic extremity and
correction gain asymmetry index.

116
Performance in the non-paretic hand of individuals with right hemiparesis was better than
controls.
The previous results suggest that those with right hemiparesis not only corrected more
with their paretic right limb within a trial, but also corrected less with their now more skilled left
limb. However, we have not yet established that the non-paretic limb has in fact become more
skilled as a result of long-standing practice with this limb after the stroke. So, in this follow up
analysis, we asked whether the non-paretic left limb was better in those with right hemiparesis
compared to controls, i.e., if it demonstrated smaller directional errors.
We found that compared to controls, directional errors were in fact smaller in the now
more skilled non-paretic left limb in individuals with right hemiparesis (RHP) (t = 9.96, p<
0.001) (Figure 5A, left panel). As expected, the paretic limb in both stroke groups showed larger
directional errors compared to controls (F = 59.86, p< 0.001), however variability was
substantially higher in those with left hemiparesis (LHP) compared to those with RHP.
Surprisingly, the paretic right hand in RHP did not have larger directional errors than the non-
paretic right limb in LHP (Figure 5A, right panel).
It is important to note however, that improvements in directional accuracy were not
without a tradeoff in movement speed. We found that as expected stroke survivors were
significantly slower compared to controls (F = 55.91, p< 0.001). While the non-paretic left hand
in RHP was predictably faster than the paretic left hand in LHP (Figure 5B, left panel), once
again, it was rather surprising that the paretic right hand in RHP was not slower than the non-
paretic right limb in LHP (Figure 5B, right panel).

117
Figure 6.5. A. Boxplots comparing directional error at peak velocity between controls and stroke
survivors. B. Boxplots comparing peak speed between controls and stroke survivors. Paretic hand in
stroke groups is indicated by labels. *** p< 0.001, ** p<0.01

Trial-by-trial adaptive response is asymmetric between the hands. After a stroke, adaptive
response is increased in the paretic limb only in those with right hemiparesis but not in
those with left hemiparesis.
By definition, correction gains examined in the previous sections index adjustments made
within a trial. We reasoned that the same error signal (i.e., between the desired and predicted
states of the cursor and the hands) that warrants an adjustment within a trial, might also improve
motor planning on the next trial. If this is the case, then we expect trial-by-trial error to not be
118
random but exhibit a statistical relationship such that error on a given trial is significantly
correlated with the error on the next trial (van Beers, 2009).
We found that in neurologically intact controls, similar to within-trial correction gains,
trial-by-trial adaptation rates, i.e., ACF coefficients, were significantly higher for the left
compared to the right hand (t(19) = 3.75, p = 0.003). Figure 6B shows difference between hands
in participant-wise ACF coefficients for each of the three groups. Compared to controls, there
was a reversal between hands in both stroke groups (group x hand interaction, F(2,80) = 7.29, p
= 0.0012). In individuals with right hemiparesis, trial-by-trial adaptation rates were similarly
higher for the paretic right hand compared to the non-paretic left hand (t(11) = 3.004, p =
0.0056). No significant difference was also found in trial-by-trial adaptation rates between the
hands in individuals with left hemiparesis (t(10) = 1.35, p = 0.22).
To examine if correction gains were related to adaptation rates, we correlated the gain
asymmetry index with the adaptation asymmetry index (Supplementary Figure F). We did not
find a significant correlation between the asymmetries observed in correction gains and
adaptation rates in any of the groups (r = 0.35, p = 0.15 in young adults; r = -0.36, p = 0.26 in
right paresis; r = -0.15, p = 0.66 in left paresis), which might suggest that the two processes are
likely independently controlled.

Figure 6.6. A. Average ACF coefficients of the last 16 trials for each hand across participants in each
group. B. Asymmetry index: A comparison of relative asymmetry in adaptation rates given by, r asym
(
|𝑟 𝑟𝑖𝑔 ℎ𝑡 |
|𝑟 𝑟𝑖𝑔 ℎ𝑡 | +| 𝑟 𝑙𝑒𝑓𝑡 |
) showing a strong left-ward bias in controls (green). Compared to controls, balance
between hands shifts to the right hand in individuals with right hemiparesis but not in in those with left
hemiparesis. C. Scatterplot and second-order non-linear fit (r asym ~ uefm + uefm
2
) shows relationship
119
between the degree of motor impairment in the paretic extremity and trial-by-trial adaptation rate
asymmetry index (not significant).

120
Discussion
The purpose of this experiment was to understand whether the process of error
assignment and consequent correction gains is entirely flexible and therefore reversible between
the hands by virtue of extensive use-dependent alternative experiences seen in chronic stroke
survivors, or if they are limited to some extent by hemispheric specialization of motor control
processes. To test our hypothesis that the flexibility of the responsibility assignment process is
partly limited by specialized control processes, we included two groups of chronic stroke
survivors who suffered unilateral injury to cortico-subcortical circuits and descending motor
pathways and presented with at least some degree of paresis to the contralateral side of the body.
Consistent with previous studies, we found that in neurologically intact adults, correction
gains were asymmetric between the limbs such that the left limb corrected more than the right
limb. Partially consistent with the predictions of optimal feedback control (OFC) theory,
correction gains were reversed between the hands in chronic stroke survivors with right
hemiparesis (RHP), such that the paretic right hand that is now more error prone showed larger
correction gains, whereas correction gains for the left non-paretic hand decreased. To confirm
that these lower gains were accompanied by better performance with the non-paretic left hand,
we compared directional errors between the left hand in RHP with that of controls and found that
the left hand in fact showed smaller errors compared to the controls. The extent of correction
seemed to vary to some extent with the degree of paretic limb motor impairment but was not
statistically significant. Further, our analysis of trial-by-trial adaptation showed that the paretic
right hand also systematically adapted to these errors over trials. However, unlike previous
findings (White & Diedrichsen, 2010), gain and adaptation asymmetry were not positively
correlated suggesting that the two processes may be independent of each other. Surprisingly,
121
individuals with left hemiparesis did not show greater correction gains with their paretic hand as
would have been predicted by OFC. Rather, correction gains did not differ between the hands
and was slightly more negative for the non-paretic right hand. This lack of interlimb asymmetry
in this group might suggest that right hemisphere circuits may have a role in feedback-mediated
adjustments.
Our findings seem consistent with the Dynamic Dominance Hypothesis, which suggests
that the sensorimotor control of the two hands are specialized, such that the left hand, primarily
under the control of the right hemisphere, is more finely tuned for feedback-mediated impedance
control which allows it to respond effectively to unstable and unpredictable environments.
Lesion analysis of right hemisphere stroke in previous work has also similarly shown that right
parietal cortex strokes may impair feedback-mediated accuracy of movements (Mutha et al.,
2011). A recent study in which transient focal cooling was applied to the right hemisphere in
macaques suggests that the dorsal premotor cortex (PMd) and parietal area 5 may be involved in
sensory integration, necessary for feedback mediated adjustments via long loop reflexes. (Takei
et al., 2021). In this study, they found that transient cooling of the right PMd impaired both
spatial accuracy and speed of corrective responses to a rapid disturbance in hand position. These
responses were qualitatively similar to those evoked by reductions in feedback gains, L, in the
OFC model. Conversely, similar inactivation by cooling of the parietal A5 interfered only with
spatial accuracy but not with speed of correction, a behavior that was predicted by a reduction in
estimation gains, K, to anywhere up to 60% of its optimal values, in the OFC model. By
dissociating the function of these two regions of the right hemisphere and by qualitatively
reproducing these experimental data with OFC simulations, this study provides causal evidence
for right PMd and A5 in feedback-mediated control of such movements.
122
In those with right hemiparesis, why did control gains not increase for both limbs? If the
left limb in non-disabled adults exhibits high feedback gains, after an injury, the motor system
could simply choose to tune up feedback gains for both limbs. However, a simple argument
against such a strategy might be that bilaterally increased gains may render task performance
unstable. After all, the goal in the one cursor task is to transport the cursor to the target,
regardless of the individual positions of the limbs. Thus, even though the states of the limbs are
estimated, these state estimations are used to infer the state of the cursor. Thus, control gains are
set so as to meet the demands of the task as it pertains to the cursor’s position and increasing
gains for one limb. Furthermore, increasing gains for both hands would also mean that noise in
these signals would be amplified, which would also result in greater task error. Thus, a unilateral
increase in gains would be a reasonable strategy in a redundant bimanual task like the one
examined here.

Limitations
There were several limitations in this study. First, the sample size of stroke patients was
small, but variability especially in the left hemiparesis group was high. It is possible the lack of
high correction gains observed for this group may simply be due to a small sample size. Second,
an important assumption in our study was that because stroke survivors were in the chronic
phase, they would have gained sufficient new learning with their ipsilesional limb, however, we
did not quantify use or experience in any meaningful way. Furthermore, there is evidence that
new skill learning may be greater for those with right hemiparesis in whom the ipsilesional left
limb was previously non-dominant. However, such de-novo learning is less conspicuous for the
right ipsilesional limb. Next, assessments of sensory deficits were limited to those provided by
123
the Upper Extremity Fugl Meyer score, which does not provide a sensitive measure. It is possible
that stroke survivors not only suffer from noisy command signals but also noisy feedback
signals. Whether the lack of feedback-mediated gain corrections is due to a problem in sensory
integration or simply highly noisy feedback cannot be dissociated in this study. Better sensory
assessments might help disambiguate the two. Finally, participants in the left and right
hemisphere stroke groups were comparable in terms of the degree of motor impairment,
however, despite common clinical features, stroke recovery and behavior is very heterogeneous.

Conclusion
The results of this study lead us to conclude that responsibility assignment is not entirely
flexible but may be limited to some extent by the hemispheric specialization of motor control
processes.

124
Supplements

ID Age(Years) DOS Chronicity Sex SOL UEFM EHI
L001 69 7/31/07 141 M LHD 59 100
L002 51 6/12/2011 95 M LHD 47 90
L003 60 2/14/2014 63 F LHD 51 100
L004 59 12/9/2013 67 F LHD 36 100
L005 68 8/5/2017 23 M LHD 26 100
L006 67 12/21/2015 43 M LHD 20 87.5
L007 70 12/2008 127 M LHD 62 100
L008 64 9/28/2009 117 F LHD 60 80
L009 83 7/2009 121 M LHD 60 88.89
L010 50 11/10/2014 57 M LHD 63 60
L011 62 4/2008 137 M LHD 55 90
L012 61 3/2018 19 M LHD 54 100
R001 75 3/5/2005 171 M RHD 41 100
R002 52 9/2011 93 F RHD 58 100
R003 73 1/4/2015 54 M RHD 58 100
R004 62 8/26/2017 23 F RHD 41 100
R005 63 4/2014 64 F RHD 52 75
R006 30 4/22/2015 51 F RHD 64 75
R007 52 11/5/2012 81 M RHD 27 100
R008 75 8/16/2010 107 M RHD 53 50
R009 70 9/4/2006 155 M RHD 36 100
R011 65 1/1/2009 128 M RHD 40 75
R012 66 1/5/2005 176 F RHD 49 100
A. (i) Stroke participant description
125

A. (ii) Detailed Upper extremity Fugl-Meyer motor and sensory score:

126
A. Synchrony of the time at which peak velocity occurs is preserved for both bimanual
task conditions.

127
B. Full Mixed Model Regression Estimates:

128

C. Fast learning demonstrated by steep decrease in Mahalanobis distance.

129
D. General description and comparisons of one- and two-cursor movements in young
adults. Plots show data from n = 20, mean ± standard error. As shown in these plots,
compared to two-cursor movements, one-cursor movements of both hands started
earlier (A), reached peak velocity faster (B), had shorter overall movement times (C),
overshot less (D). The left, but not the right hand, also showed significantly smaller
directional errors in the one-cursor condition compared to the two-cursor condition (E).
Consistent with previous studies, there were also significant differences between the
limbs; the left limb consistently started movements earlier (by ~50 ms) but took longer
to reach peak velocity and had slower overall movement times, often showed less
overshooting and significantly larger directional errors compared to the right hand.
Left
Right Left
Right
Left
Right
“Go ”
Left
Right
Left
Right
90 ˚ straight
0.02
ns
A B C
D E

130

E. A. Relationship between gain asymmetry index and adaptation asymmetry index. B.
Gains and adaptation rates separated by hand with colors representing groups.

A.
B.

131

F. Adapting an optimal feedback controller to simulate bimanual control in the
independent two-cursor and coupled one-cursor condition. Here, we only describe the
structure of the controller. However, sensitivity analysis for parameters is underway and
not presented in this dissertation.

Line diagram of the OFC controller:

132
CHAPTER 7: Summary and conclusions
The overall purpose of this dissertation was to characterize hemisphere-specific deficits
in the control of bimanual movements post stroke. In Chapter 3, I examined spontaneous use of
both hands in individuals with left- and right-hemisphere stroke and neurologically intact
controls. We inferred regarding the role of hemisphere-specific control by examining tasks with
different demands related to stabilization in the two stroke groups. I found that in chronic stroke
survivors, the probability of choosing both hands depends on an interaction between motor
capacity and limb‐specific control. Unlike age-similar able-bodied adults, chronic stroke
survivors do not spontaneously choose both hands to solve routine bimanual tasks. The
probability of choosing both hands increases when the contralesional arm is less impaired.
Importantly, the effect of motor impairment is modified both by the side of lesion and the type of
task. We argued that our findings seem inconsistent with the predictions of a traditional global
dominance model. Instead, in chronic stroke survivors, bimanual use emerges from a task-
specific interaction between motor impairment and the side of lesion, such that when there is
sufficient motor capacity, the paretic hand is preferentially selected by the nervous system to
assume a role consistent with its specialized controller.
Next, in Chapter 4, I reasoned that because bimanual tasks necessitated capacity not only
in the contralesional limb, but also appropriate and sufficient control of the ipsilesional limb,
examining the relationship between deficits in distal function between ipsilesional and
contralesional limbs might reveal insights into hemisphere-specific differences observed in
Chapter 3. I found that ipsilesional deficits covaried with contralesional impairments in those
whose stroke affects the left hemisphere, but this relationship is less pronounced in those whose
stroke affects the right hemisphere. We concluded that this finding likely underscores the

133
extensive motor experiences of the pre-morbidly dominant ipsilesional limb and the importance
of the left hemisphere in the control of timed tasks for both hands.
To examine the influence of stroke on callosal white matter, which is a putative neural
substrate for interhemispheric communication necessary for quick and accurate bimanual
performance, Chapter 5 utilized retrospective analysis of diffusion data in the same chronic
stroke survivors from Chapter 3. I performed diffusion analysis with the corpus callosum as the
region of interest and supplemented these data with a publicly available diffusion dataset in
controls. I correlated behavioral data (movement time) from Chapter 3 with fractional anisotropy
and found that FA of the CC significantly predicted bimanual performance. Chronic stroke
survivors who exhibited lower FA in the CC were slower at performing the letter-envelope tasks
of the AAUT. By comparing stroke data with controls, I found that chronic stroke survivors
presented with significantly greater loss of callosal fiber orientation (lower mean FA), compared
to age-similar neurologically intact adults who in turn exhibited lower FA compared to younger
controls. I concluded that in mild-to-moderate chronic stroke survivors with relatively localized
lesions to the motor areas, callosal microstructure can be expected to change not only in the
primary sensorimotor region, but also in the premotor, supplementary motor and prefrontal
regions. These remote widespread changes in the callosal genu and body are likely to impact
performance on cooperative bimanual tasks that require precise and interdependent coordination
of the hands. Thus, using retrospective analysis of clinico-behavioral data, we confirmed in
Chapters 3 and 4 that several outcomes of potential clinical interest are influenced by the
hemispheric side of lesion. Through preliminary analysis of diffusion imaging, we also found
that bimanual performance is correlated with callosal microstructure.

134
Features of our behavioral observations seemed to be consistent with the Dynamic
Dominance Hypothesis, a model of motor lateralization. However, both of these behavioral
studies were retrospective and so limited by the measures that were available for analysis. Thus,
to more directly test hemisphere-specific deficits in bimanual control I designed a set of
prospective experiments. First, I conducted a study of interlimb interference using a modified
interference paradigm described by Kelso (1979) (not described in the main body of the
Dissertation but findings added as an Appendix). I found that despite strong temporal synchrony,
especially of the time of peak velocity, there was no interference between the limbs. It is possible
that a simple distance manipulation may not have elicited the type of interference observed by
Kelso in his study. However, another explanation might be that for discrete tasks such as the one
I used, participants could use a cognitive strategy to overcome any interference. Brief description
of this study is provided in the Appendix.
Next, to test a more contemporary model of motor control, I adapted a paradigm from
White and Diedrichsen to study the control of cooperative redundant bimanual reaching. In this
study, participants performed visually cued reaching such that the position of the cursor was the
spatial average of the two hands. Online feedback was withheld, and participants were asked to
be quick and accurate with their movements. Previously, White and Diedrichsen, showed that
neurologically intact right-handed adults consistently showed a greater tendency to correct with
their left non-dominant limb and did so faster than with the right limb. They concluded that the
left hand’s tendency to correct for errors may be because it is more error prone.
An alternative explanation is that the increase in the left hand’s gain is due to its
proficiency with feedback-mediated control under the primary control of the right hemisphere.
To dissociate these two explanations, we reproduced the findings of White and Diedrichsen in

135
young neurotypical controls. Then, we tested the specific prediction related to the flexibility of
the error assignment process in chronic stroke survivors with left and right hemisphere damage.
If this process is entirely flexible, theory would predict that regardless of the hemisphere of
damage, the paretic limb, which is now significantly more prone to errors, would rely on sensory
feedback to make corrections and consequently show high control gains, whereas the now well-
practiced, more skilled non-paretic limb would show smaller gains. Conversely, if the feedback-
mediated gain settings are at least to some extent a feature of a specialized right hemisphere-left
hand system, then correction gains will only be modified when the putative right hemispheric
circuits are intact, i.e., in those with damage to the left hemisphere and right arm paresis. Further,
in those with damage to the right hemisphere, gains should be smaller for both limbs. I found
support for the alternative explanation that the error assignment process is not entirely flexible
but limited by specialized processes in the right hemisphere/left limb system. However, wide
variability was observed in the response of those with right hemisphere stroke (i.e., the right
hemisphere-left hand system). In addition, the sample size for this group was small (i.e., n = 11).
Lastly, I have recently developed a computer simulation for a bimanual optimal feedback
controller with common cursor and goal, similar task requirements of the redundant task
described in Chapter 6. I plan to conduct parameter estimation and sensitivity analysis so as to
qualitatively reproduce features of bimanual movements that I observed in chronic stroke
survivors. Larger samples may help reduce variability in stroke groups, however, recruitment
was halted in March 2020 on account of the COVID19 pandemic.

136
APPENDIX
Strong temporal coupling but no interlimb interference observed during discrete
asymmetric bimanual reaching.
Prior to the experiment described in Chapter 6, we performed an experiment based on the
original paradigm described by Kelso (1979) with important modifications described in Table 1
below. The goal of this experiment was to elicit interlimb interference in young non-disabled
adults and then characterize the nature of such interference in stroke survivors with left and right
hemisphere damage. Participants were asked to reach quickly and accurately to asymmetric
target distances (far target set at 3 times the distance of short target; 5 and 15 cm). Data
acquisition and processing were the same as described in Chapter 6. Temporal variables were
also determined in the same way as described in Chapter 6. Critical time points in the movement
(onset, time of peak speed, and offset) were correlated between the hands using Pearson’s
correlation. To rule out that correlations observed in bimanual movements arose simply from
stereotypy in movement profiles, we used a procedure for trials in the unimanual condition
whereby we searched for the order of trials that would lend maximum correlation in time of peak
velocity between hands. We then used this trial order to correlate onsets and offsets.
Distributions of participant-wise correlation coefficients were compared between unimanual and
bimanual conditions. We found that despite strong temporal synchrony, especially at the time of
peak velocity, there was no interference between the limbs. Thus, I was unable to qualitatively
reproduce the findings of Kelso (1979).

137
Table A1. Differences in interference paradigm between Kelso (1979) and one used in this
study.

Figure A1. A. Task conditions: unimanual conditions for the left and right hand reaching to two
distances, 5 and 15 cm, and corresponding bimanual conditions with left-far (asym-L) and right-
far (asym-R) conditions. B. Predictions for movement time based on interlimb interference
hypothesis.

138
Figure A2. A. Distribution of correlation coefficients in the bimanual (solid fill) and
corresponding unimanual (clear fill) conditions. All comparisons between UM and BM
significant at p< 0.001, except that for offsets, which were either weak (asymL) or not significant
(asymR). B. Boxplot showing correlation coefficients from the Bimanual asymmetric condition
only. As seen in this plot, strongest correlations were observed for the time of peak speed,
followed by onsets and least strong for offsets. C. MT for unimanual and the two bimanual
conditions as predicted to be evidence of interference (dashed lines) and as were found (solid
lines). We did not see support for our predictions (Figure A1B). There was no interference in MT
between the limbs.

139
BIBLIOGRAPHY

Andersson, J. L. R., Jenkinson, M., & Smith, S. (2007a). Non-linear optimisation. FMRIB
technical report TR07JA1. Practice.
Andersson, J. L. R., Jenkinson, M., & Smith, S. (2007b). Non-linear registration, aka Spatial
normalisation FMRIB technical report TR07JA2. FMRIB Analysis Group of the
University of Oxford, 2(1), e21.
Bailey, R. R., Klaesner, J. W., & Lang, C. E. (2014). An accelerometry-based methodology for
assessment of real-world bilateral upper extremity activity. PLoS ONE, 9(7).
https://doi.org/10.1371/journal.pone.0103135
Bailey, R. R., Klaesner, J. W., & Lang, C. E. (2015a). Quantifying Real-World Upper-Limb
Activity in Nondisabled Adults and Adults with Chronic Stroke. Neurorehabilitation and
Neural Repair, 29(10), 969–978. https://doi.org/10.1177/1545968315583720
Bailey, R. R., Klaesner, J. W., & Lang, C. E. (2015b). Quantifying Real-World Upper-Limb
Activity in Nondisabled Adults and Adults with Chronic Stroke. Neurorehabilitation and
Neural Repair, 29(10), 969–978. https://doi.org/10.1177/1545968315583720
Basser, P. J., & Pierpaoli, C. (2011). Microstructural and physiological features of tissues
elucidated by quantitative-diffusion-tensor MRI. Journal of Magnetic Resonance, 213(2),
560–570. https://doi.org/10.1016/j.jmr.2011.09.022
Baxter, M. G., Parker, A., Lindner, C. C. C., Izquierdo, A. D., & Murray, E. A. (2000). Control
of response selection by reinforcer value requires interaction of amygdala and orbital
prefrontal cortex. Journal of Neuroscience, 20(11), 4311–4319.
https://doi.org/10.1523/jneurosci.20-11-04311.2000

140
Boll, T. J. (1974). Right and left cerebral hemisphere damage and tactile perception:
Performance of the ipsilateral and contralateral sides of the body. Neuropsychologia,
12(2), 235–238. https://doi.org/10.1016/0028-3932(74)90008-6
Bonzano, L., Tacchino, A., Roccatagliata, L., Abbruzzese, G., Mancardi, G. L., & Bove, M.
(2008). Callosal contributions to simultaneous bimanual finger movements. Journal of
Neuroscience, 28(12), 3227–3233. https://doi.org/10.1523/JNEUROSCI.4076-07.2008
Bracewell, R. M., Husain, M., & Stein, J. F. (1990). Specialization of the right hemisphere for
visuomotor control. Neuropsychologia. https://doi.org/10.1016/0028-3932(90)90001-5
Brinkman, J., & Kuypers, H. G. J. M. (1973). Cerebral control of contralateral and ipsilateral
arm, hand and finger movements in the split-brain rhesus monkey. Brain, 96(4), 653–
674. https://doi.org/10.1093/brain/96.4.653
Brown, T. G. (1911). The intrinsic factors in the act of progression in the mammal. Proceedings
of the Royal Society of London. Series B, Containing Papers of a Biological Character,
84(572), 308–319. https://doi.org/10.1098/rspb.1911.0077
Brown, T. G. (1914). On the nature of the fundamental activity of the nervous centres; together
with an analysis of the conditioning of rhythmic activity in progression, and a theory of
the evolution of function in the nervous system. The Journal of Physiology, 48(1), 18–46.
https://doi.org/10.1113/jphysiol.1914.sp001646
Bundy, D. T., Szrama, N., Pahwa, M., & Leuthardt, E. C. (2018). Unilateral, 3D Arm Movement
Kinematics Are Encoded in Ipsilateral Human Cortex. The Journal of Neuroscience.
https://doi.org/10.1523/jneurosci.0015-18.2018

141
Bustrén, E. L., Sunnerhagen, K. S., & Alt Murphy, M. (2017). Movement kinematics of the
ipsilesional upper extremity in persons with moderate or mild stroke. Neurorehabilitation
and Neural Repair, 31(4), 376–386. https://doi.org/10.1177/1545968316688798
Buxbaum, L. J., Varghese, R., Stoll, H., & Winstein, C. J. (2020). Predictors of Arm Nonuse in
Chronic Stroke: A Preliminary Investigation. Neurorehabilitation and Neural Repair,
34(6), 512–522. https://doi.org/10.1177/1545968320913554
Caeyenberghs, K., Leemans, A., Coxon, J., Leunissen, I., Drijkoningen, D., Geurts, M., Gooijers,
J., Michiels, K., Sunaert, S., & Swinnen, S. P. (2011a). Bimanual coordination and
corpus callosum microstructure in young adults with traumatic brain injury: A diffusion
tensor imaging study. Journal of Neurotrauma, 28(6), 897–913.
https://doi.org/10.1089/neu.2010.1721
Caeyenberghs, K., Leemans, A., Coxon, J., Leunissen, I., Drijkoningen, D., Geurts, M., Gooijers,
J., Michiels, K., Sunaert, S., & Swinnen, S. P. (2011b). Bimanual Coordination and
Corpus Callosum Microstructure in Young Adults with Traumatic Brain Injury: A
Diffusion Tensor Imaging Study. Journal of Neurotrauma, 28(6), 897–913.
https://doi.org/10.1089/neu.2010.1721
Cernacek, J. (1961). Contralateral motor irradiation-cerebral dominance: Its changes in
hemiparesis. Archives of Neurology.
https://doi.org/10.1001/archneur.1961.00450080047005
Chae, J., Yang, G., Park, B. K., & Labatia, I. (2002). Delay in initiation and termination of
muscle contraction, motor impairment, and physical disability in upper limb hemiparesis.
Muscle and Nerve, 25(4), 568–575. https://doi.org/10.1002/mus.10061

142
Chestnut, C., & Haaland, K. Y. (2008). Functional Significance of Ipsilesional Motor Deficits
After Unilateral Stroke. Archives of Physical Medicine and Rehabilitation, 89(1), 62–68.
https://doi.org/10.1016/J.APMR.2007.08.125
Cirillo, C., Brihmat, N., Castel-Lacanal, E., Le Friec, A., Barbieux-Guillot, M., Raposo, N.,
Pariente, J., Viguier, A., Simonetta-Moreau, M., Albucher, J. F., Olivot, J. M.,
Desmoulin, F., Marque, P., Chollet, F., & Loubinoux, I. (2020). Post-stroke remodeling
processes in animal models and humans. Journal of Cerebral Blood Flow and
Metabolism, 40(1), 3–22. https://doi.org/10.1177/0271678X19882788
Coelho, C. J., Przybyla, A., Yadav, V., & Sainburg, R. L. (2013). Hemispheric differences in the
control of limb dynamics: A link between arm performance asymmetries and arm
selection patterns. Journal of Neurophysiology, 109(3), 825–838.
https://doi.org/10.1152/jn.00885.2012
Collins, G. S., Ogundimu, E. O., Cook, J. A., Manach, Y. Le, & Altman, D. G. (2016).
Quantifying the impact of different approaches for handling continuous predictors on the
performance of a prognostic model. Statistics in Medicine.
https://doi.org/10.1002/sim.6986
de Paiva Silva, F. P., De Freitas, S. M. S. F., Silva, P. V., Banjai, R. M., & Alouche, S. R.
(2014). Ipsilesional arm motor sequence performance after right and left hemisphere
damage. Journal of Motor Behavior, 46(6), 407–414.
https://doi.org/10.1080/00222895.2014.924473
de Paiva Silva, F. P., Freitas, S. M. S. F., Banjai, R. M., & Alouche, S. R. (2018). Ipsilesional
arm aiming movements after stroke: Influence of the degree of contralesional
impairment. Journal of Motor Behavior. https://doi.org/10.1080/00222895.2017.1306479

143
Debaere, F., Wenderoth, N., Sunaert, S., Van Hecke, P., & Swinnen, S. P. (2004). Cerebellar and
premotor function in bimanual coordination: Parametric neural responses to
spatiotemporal complexity and cycling frequency. NeuroImage, 21(4), 1416–1427.
https://doi.org/10.1016/j.neuroimage.2003.12.011
Diedrichsen, J. (2007). Optimal task-dependent changes of bimanual feedback control and
adaptation. Current Biology: CB, 17(19), 1675–1679.
https://doi.org/10.1016/j.cub.2007.08.051
Diedrichsen, J., & Dowling, N. (2009). Bimanual coordination as task-dependent linear control
policies. Human Movement Science, 28(3), 334–347.
https://doi.org/10.1016/j.humov.2008.10.003
Diedrichsen, J., Shadmehr, R., & Ivry, R. B. (2010). The coordination of movement: Optimal
feedback control and beyond. Trends in Cognitive Sciences, 14(1), 31–39.
https://doi.org/10.1016/j.tics.2009.11.004
Donchin, O. (1999). Primary and supplementary cortex in bimanual movements: A study of
cortical physiology (Issue December).
Donchin, O., Gribova, A., Steinberg, O., Bergman, H., & Vaadia, E. (1998). Primary motor
cortex is involved in bimanual coordination. Nature, 395(6699), 274–278.
https://doi.org/10.1038/26220
Egorova, N., Dhollander, T., Khlif, M. S., Khan, W., Werden, E., & Brodtmann, A. (2020).
Pervasive White Matter Fiber Degeneration in Ischemic Stroke. Stroke, 1507–1513.
https://doi.org/10.1161/STROKEAHA.119.028143

144
Egorova, N., Liem, F., Hachinski, V., & Brodtmann, A. (2019). Predicted Brain Age After
Stroke. Frontiers in Aging Neuroscience, 11(December), 4–11.
https://doi.org/10.3389/fnagi.2019.00348
Eliassen, J. C., Baynes, K., & Gazzaniga, M. S. (2000a). Anterior and posterior callosal
contributions to simultaneous bimanual movements of the hands and fingers. Brain: A
Journal of Neurology, 123 Pt 12, 2501–2511. https://doi.org/10.1093/brain/123.12.2501
Eliassen, J. C., Baynes, K., & Gazzaniga, M. S. (2000b). Anterior and posterior callosal
contributions to simultaneous bimanual movements of the hands and fingers. Brain,
123(12), 2501–2511. https://doi.org/10.1093/brain/123.12.2501
Findlater, S. E., Mazerolle, E. L., Pike, G. B., & Dukelow, S. P. (2019). Proprioception and
motor performance after stroke: An examination of diffusion properties in sensory and
motor pathways. Human Brain Mapping, 40(10), 2995–3009.
Fling, B. W., & Seidler, R. D. (2012). Fundamental differences in callosal structure,
neurophysiologic function, and bimanual control in young and older adults. Cerebral
Cortex, 22(11), 2643–2652. https://doi.org/10.1093/cercor/bhr349
Fling, B. W., Walsh, C. M., Bangert, A. S., Reuter-Lorenz, P. A., Welsh, R. C., & Seidler, R. D.
(2011). Differential callosal contributions to bimanual control in young and older adults.
Journal of Cognitive Neuroscience, 23(9), 2171–2185.
https://doi.org/10.1162/jocn.2010.21600
Franck, J. A., Smeets, R. J. E. M., & Seelen, H. A. M. (2019). Changes in actual arm-hand use in
stroke patients during and after clinical rehabilitation involving a well-defined arm-hand
rehabilitation program: A prospective cohort study. PLoS ONE, 14(4).
https://doi.org/10.1371/journal.pone.0214651

145
Franz, E. A., Eliassen, J. C., Ivry, R. B., & Gazzaniga, M. S. (1996a). Dissociation of Spatial and
Temporal Coupling in the Bimanual Movements of Callosotomy Patients. Psychological
Science, 7(5), 306–310. https://doi.org/10.1111/j.1467-9280.1996.tb00379.x
Franz, E. A., Eliassen, J. C., Ivry, R. B., & Gazzaniga, M. S. (1996b). Dissociation of Spatial and
Temporal Coupling in the Bimanual Movements of Callosotomy Patients. Psychological
Science, 7(5), 306–310. https://doi.org/10.1111/j.1467-9280.1996.tb00379.x
Franz, E. A., Zelaznik, H. N., & McCabe, G. (1991). Spatial topological constraints in a
bimanual task. Acta Psychologica, 77(2), 137–151. https://doi.org/10.1016/0001-
6918(91)90028-X
Freitas, S. M. S. F., Gera, G., & Scholz, J. P. (2011). Timing variability of reach trajectories in
left versus right hemisphere stroke. Brain Research, 1419, 19–33.
https://doi.org/10.1016/j.brainres.2011.08.039
Fugl Meyer, A. R., Jaasko, L., & Leyman, I. (1975). The post stroke hemiplegic patient. I. A
method for evaluation of physical performance. Scandinavian Journal of Rehabilitation
Medicine.
Fugl-Meyer, A. R., Jääskö, L., Leyman, I., Olsson, S., & Steglind, S. (1975). The post-stroke
hemiplegic patient. 1. A method for evaluation of physical performance. Scandinavian
Journal of Rehabilitation Medicine. https://doi.org/10.1038/35081184
Gandrey, P., Paizis, C., Karathanasis, V., Gueugneau, N., & Papaxanthis, C. (2013). Dominant
vs. Nondominant arm advantage in mentally simulated actions in right handers. Journal
of Neurophysiology, 110(12), 2887–2894. https://doi.org/10.1152/jn.00123.2013
Ganguly, K., Secundo, L., Ranade, G., Orsborn, A., Chang, E. F., Dimitrov, D. F., Wallis, J. D.,
Barbaro, N. M., Knight, R. T., & Carmena, J. M. (2009). Cortical Representation of

146
Ipsilateral Arm Movements in Monkey and Man. Journal of Neuroscience.
https://doi.org/10.1523/JNEUROSCI.2471-09.2009
Giuliani, C. A., Purser, J. L., Light, K. E., & Genova, P. A. (1997). Impairments in arm control
in subjects with left and right hemisphere stroke. NeuroRehabilitation, 9(1), 71–87.
https://doi.org/10.1016/S1053-8135(97)00015-2
Gooijers, J., Caeyenberghs, K., Sisti, H. M., Geurts, M., Heitger, M. H., Leemans, A., &
Swinnen, S. P. (2013a). Diffusion tensor imaging metrics of the corpus callosum in
relation to bimanual coordination: Effect of task complexity and sensory feedback.
Human Brain Mapping, 34(1), 241–252. https://doi.org/10.1002/hbm.21429
Gooijers, J., Caeyenberghs, K., Sisti, H. M., Geurts, M., Heitger, M. H., Leemans, A., &
Swinnen, S. P. (2013b). Diffusion tensor imaging metrics of the corpus callosum in
relation to bimanual coordination: Effect of task complexity and sensory feedback.
Human Brain Mapping, 34(1), 241–252. https://doi.org/10.1002/hbm.21429
Gordon, J., & Ghez, C. (1987). Trajectory control in targeted force impulses. III. Compensatory
adjustments for initial errors. Experimental Brain Research, 67(2), 253–269.
https://doi.org/10.1007/BF00248547
Gordon, J., Ghilardi, M. F., Cooper, S. E., & Ghez, C. (1994). Accuracy of planar reaching
movements. II. Systematic extent errors resulting from inertial anisotropy. Experimental
Brain Research, 99(1), 112–130. https://doi.org/10.1007/BF00241416
Gordon, J., Ghilardi, M. F., & Ghez, C. (1994). Accuracy of planar reaching movements. I.
Independence of direction and extent variability. Experimental Brain Research, 99(1),
97–111. https://doi.org/10.1007/BF00241415

147
Gratton, C., Nomura, E. M., Pérez, F., & D’Esposito, M. (2012). Focal brain lesions to critical
locations cause widespread disruption of the modular organization of the brain. Journal
of Cognitive Neuroscience, 24(6), 1275–1285.
Gresham, G. E., Fitzpatrick, T. E., Wolf, P. A., McNamara, P. M., Kannel, W. B., & Dawber, T.
R. (1975). Residual Disability in Survivors of Stroke—The Framingham Study. New
England Journal of Medicine, 293(19), 954–956.
https://doi.org/10.1056/NEJM197511062931903
Grillner, S. (1975). Locomotion in vertebrates: Central mechanisms and reflex interaction.
Physiological Reviews, 55(2), 247–304. https://doi.org/10.1152/physrev.1975.55.2.247
Guiard, Y. (1987). Asymmetric division of labor in human skilled bimanual action: The
kinematic chain as a model. Journal of Motor Behavior, 19(4), 486–517.
https://doi.org/10.1080/00222895.1987.10735426
Haaland, K. Y., Mutha, P. K., Rinehart, J. K., Daniels, M., Cushnyr, B., & Adair, J. C. (2012).
Relationship between arm usage and instrumental activities of daily living after unilateral
stroke. Archives of Physical Medicine and Rehabilitation, 93(11), 1957–1962.
https://doi.org/10.1016/j.apmr.2012.05.011
Haaland, K. Y., Schaefer, S. Y., Knight, R. T., Adair, J., Magalhaes, A., Sadek, J., & Sainburg,
R. L. (2009). Ipsilesional trajectory control is related to contralesional arm paralysis after
left hemisphere damage. Experimental Brain Research, 196(2), 195–204.
https://doi.org/10.1007/s00221-009-1836-z
Hagmann, P., Jonasson, L., Maeder, P., Thiran, J.-P., Wedeen, V. J., & Meuli, R. (2006a).
Understanding Diffusion MR Imaging Techniques: From Scalar Imaging to Diffusion.
Radiographics, 26, 205–224.

148
Hagmann, P., Jonasson, L., Maeder, P., Thiran, J.-P., Wedeen, V. J., & Meuli, R. (2006b).
Understanding Diffusion MR Imaging Techniques: From Scalar Diffusion-weighted
Imaging to Diffusion Tensor Imaging and Beyond. RadioGraphics, 26(suppl_1), S205–
S223. https://doi.org/10.1148/rg.26si065510
Haken, H., Kelso, J. A., & Bunz, H. (1985). A theoretical model of phase transitions in human
hand movements. Biological Cybernetics, 51(5), 347–356.
https://doi.org/10.1007/BF00336922
Halsband, U., Ito, N., Tanji, J., & Freund, H.-J. (1993). The role of premotor cortex and the
supplementary motor area in the temporal control of movement in man. Brain, 116(1),
243–266.
Han, C. E., Kim, S., Chen, S., Lai, Y. H., Lee, J. Y., Osu, R., Winstein, C. J., & Schweighofer,
N. (2013). Quantifying arm nonuse in individuals poststroke. Neurorehabilitation and
Neural Repair, 27(5), 439–447. https://doi.org/10.1177/1545968312471904
Hanna-Pladdy, B., Mendoza, J. E., Apostolos, G. T., & Heilman, K. M. (2002). Lateralised
motor control: Hemispheric damage and the loss of deftness. Journal of Neurology
Neurosurgery and Psychiatry, 73(5), 574–577. https://doi.org/10.1136/jnnp.73.5.574
Harrington, D. L., & Haaland, K. Y. (1991). Hemispheric specialization for motor sequencing:
Abnormalities in levels of programming. Neuropsychologia, 29(2), 147–163.
https://doi.org/10.1016/0028-3932(91)90017-3
Harrington, D. L., Haaland, K. Y., & Knight, R. T. (1998). Cortical networks underlying
mechanisms of time perception. The Journal of Neuroscience : The Official Journal of
the Society for Neuroscience.

149
Harris, J. E., & Eng, J. J. (2006). Individuals with the dominant hand affected following stroke
demonstrate less impairment than those with the nondominant hand affected.
Neurorehabilitation and Neural Repair, 20(3), 380–389.
https://doi.org/10.1177/1545968305284528
Hawe, R. L., Sukal-Moulton, T., & Dewald, J. P. A. (2013). The effect of injury timing on white
matter changes in the corpus callosum following unilateral brain injury. NeuroImage:
Clinical, 3, 115–122. https://doi.org/10.1016/j.nicl.2013.08.002
Hayward, K. S., Neva, J. L., Mang, C. S., Peters, S., Wadden, K. P., Ferris, J. K., & Boyd, L. A.
(2017). Interhemispheric Pathways Are Important for Motor Outcome in Individuals with
Chronic and Severe Upper Limb Impairment Post Stroke. Neural Plasticity, 2017.
https://doi.org/10.1155/2017/4281532
Heap, M., & Wyke, M. (1972). Learning of a unimanual motor skill by patients with brain
lesions: An experimental study. Cortex, 8(1), 1–18. https://doi.org/10.1016/S0010-
9452(72)80023-6
Hess, C. W., Mills, K. R., & Murray, N. M. F. (1986). Magnetic stimulation of the human brain:
Facilitation of motor responses by voluntary contraction of ipsilateral and contralateral
muscles with additional observations on an amputee. Neuroscience Letters.
https://doi.org/10.1016/0304-3940(86)90565-3
Hofer, S., & Frahm, J. (2006). Topography of the human corpus callosum revisited-
Comprehensive fiber tractography using diffusion tensor magnetic resonance imaging.
NeuroImage, 32(3), 989–994. https://doi.org/10.1016/j.neuroimage.2006.05.044

150
Hoyer, E. H., & Celnik, P. A. (2011). Understanding and enhancing motor recovery after stroke
using transcranial magnetic stimulation. Restorative Neurology and Neuroscience, 29(6),
395–409.
Izawa, J., Rane, T., Donchin, O., & Shadmehr, R. (2008). Motor Adaptation as a Process of
Reoptimization. Journal of Neuroscience, 28(11), 2883–2891.
https://doi.org/10.1523/JNEUROSCI.5359-07.2008
Jenkinson, M., Beckmann, C. F., Behrens, T. E. J., Woolrich, M. W., & Smith, S. M. (2012). Fsl.
Neuroimage, 62(2), 782–790.
Jenkinson, M., Pechaud, M., & Smith, S. (2005). BET2: MR-based estimation of brain, skull and
scalp surfaces. Eleventh Annual Meeting of the Organization for Human Brain Mapping,
17, 167.
Jones, T. A. (2017). Motor compensation and its effects on neural reorganization after stroke. In
Nature Reviews Neuroscience (Vol. 18, Issue 5, pp. 267–280).
https://doi.org/10.1038/nrn.2017.26
Jones, T. A., & Adkins, D. L. (2015). Motor system reorganization after stroke: Stimulating and
training toward perfection. Physiology, 30(5), 358–370.
https://doi.org/10.1152/physiol.00014.2015
Jones, T. A., & Schallert, T. (1992). Overgrowth and pruning of dendrites in adult rats
recovering from neocortical damage. Brain Research, 581(1), 156–160.
https://doi.org/10.1016/0006-8993(92)90356-E
Jones, T. A., & Schallert, T. (1994). Use-dependent growth of pyramidal neurons after
neocortical damage. Journal of Neuroscience, 14(4), 2140–2152.

151
Kamper, D. G., McKenna-Cole, A. N., Kahn, L. E., & Reinkensmeyer, D. J. (2002). Alterations
in reaching after stroke and their relation to movement direction and impairment severity.
Archives of Physical Medicine and Rehabilitation, 83(5), 702–707.
https://doi.org/10.1053/apmr.2002.32446
Kantak, S., Jax, S., & Wittenberg, G. (2017). Bimanual coordination: A missing piece of arm
rehabilitation after stroke. Restorative Neurology and Neuroscience, 35(4), 347–364.
https://doi.org/10.3233/RNN-170737
Kazennikov, O., Hyland, B., Wicki, U., Perrig, S., Rouiller, E. M., & Wiesendanger, M. (1998).
Effects of lesions in the mesial frontal cortex on bimanual co- ordination in monkeys.
Neuroscience, 85(3), 703–716. https://doi.org/10.1016/S0306-4522(97)00693-3
Kelso, J. A. S., Putnam, C. A., & Goodman, D. (1983). On the space-time structure of human
interlimb co-ordination. The Quarterly Journal of Experimental Psychology Section A,
35(2), 347–375. https://doi.org/10.1080/14640748308402139
Kelso, J. A. S., & Schöner, G. (1988). Self-organization of coordinative movement patterns.
Human Movement Science, 7(1), 27–46. https://doi.org/10.1016/0167-9457(88)90003-6
Kelso, J. A., Southard, D. L., & Goodman, D. (1979). On the nature of human interlimb
coordination. Science (New York, N.Y.), 203(4384), 1029–1031.
https://doi.org/10.1126/SCIENCE.424729
Kilbreath, S. L., & Heard, R. C. (2005a). Frequency of hand use in healthy older persons.
Australian Journal of Physiotherapy, 51(2), 119–122. https://doi.org/10.1016/S0004-
9514(05)70040-4

152
Kilbreath, S. L., & Heard, R. C. (2005b). Frequency of hand use in healthy older persons.
Australian Journal of Physiotherapy, 51(2), 119–122. https://doi.org/10.1016/S0004-
9514(05)70040-4
Kim, B. (2017). Relationship Between Brain Structure and Motor Behavior in Chronic Stroke
Survivors. In ProQuest Dissertations and Theses. University of Southern California.
Kim, B., & Winstein, C. (2017). Can Neurological Biomarkers of Brain Impairment Be Used to
Predict Poststroke Motor Recovery? A Systematic Review. Neurorehabilitation and
Neural Repair, 31(1), 3–24. https://doi.org/10.1177/1545968316662708
Kim, S. G., Ashe, J., Hendrich, K., Ellermann, J. M., Merkle, H., Uǧurbil, K., & Georgopoulos,
A. P. (1993). Functional magnetic resonance imaging of motor cortex: Hemispheric
asymmetry and handedness. Science, 261(July), 615–616.
https://doi.org/10.1126/science.8342027
Kim, S., Park, H., Han, C. E., Winstein, C. J., & Schweighofer, N. (2018). Measuring habitual
arm use post-stroke with a bilateral time-constrained reaching task. Frontiers in
Neurology, 9(OCT), 1–5. https://doi.org/10.3389/fneur.2018.00883
Kornysheva, K., & Diedrichsen, J. (2014). Human premotor areas parse sequences into their
spatial and temporal features. Elife, 3, e03043.
Krakauer, J. W., & Carmichael, S. T. (2017). Broken movement: The neurobiology of motor
recovery after stroke. MIT Press.
Li, Y., Wu, P., Liang, F., & Huang, W. (2015). The microstructural status of the corpus callosum
is associated with the degree of motor function and neurological deficit in stroke patients.
PLoS ONE, 10(4), 1–17. https://doi.org/10.1371/journal.pone.0122615

153
Liew, S. L., Anglin, J. M., Banks, N. W., Sondag, M., Ito, K. L., Kim, H., Chan, J., Ito, J., Jung,
C., Khoshab, N., Lefebvre, S., Nakamura, W., Saldana, D., Schmiesing, A., Tran, C., Vo,
D., Ard, T., Heydari, P., Kim, B., … Stroud, A. (2017). A large, open source dataset of
stroke anatomical brain images and manual lesion segmentations. BioRxiv, 1–11.
https://doi.org/10.1101/179614
Lindenberg, R., Zhu, L. L., Rüber, T., & Schlaug, G. (2012). Predicting functional motor
potential in chronic stroke patients using diffusion tensor imaging. Human Brain
Mapping, 33(5), 1040–1051. https://doi.org/10.1002/hbm.21266
Lotze, M., Markert, J., Sauseng, P., Hoppe, J., Plewnia, C., & Gerloff, C. (2006). The role of
multiple contralesional motor areas for complex hand movements after internal capsular
lesion. Journal of Neuroscience, 26(22), 6096–6102.
https://doi.org/10.1523/JNEUROSCI.4564-05.2006
Maenza, C., Wagstaff, D. A., Varghese, R., Winstein, C., Good, D. C., & Sainburg, R. L. (2021).
Remedial Training of the Less-Impaired Arm in Chronic Stroke Survivors With
Moderate to Severe Upper-Extremity Paresis Improves Functional Independence: A Pilot
Study. Frontiers in Human Neuroscience, 15, 645714.
https://doi.org/10.3389/fnhum.2021.645714
Mamolo, C. M., Roy, E. A., Bryden, P. J., & Rohr, L. E. (2004). The effects of skill demands
and object position on the distribution of preferred hand reaches. Brain and Cognition,
55(2), 349–351. https://doi.org/10.1016/j.bandc.2004.02.041
Mamolo, C. M., Roy, E. a, Rohr, L. E., & Bryden, P. J. (2006). Reaching patterns across working
space: The effects of handedness, task demands, and comfort levels. Laterality, 11(5),
465–492. https://doi.org/10.1080/13576500600775692

154
Mani, S., Mutha, P. K., Przybyla, A., Haaland, K. Y., Good, D. C., & Sainburg, R. L. (2013a).
Contralesional motor deficits after unilateral stroke reflect hemisphere-specific control
mechanisms. Brain, 136(4), 1288–1303. https://doi.org/10.1093/brain/aws283
Mani, S., Mutha, P. K., Przybyla, A., Haaland, K. Y., Good, D. C., & Sainburg, R. L. (2013b).
Contralesional motor deficits after unilateral stroke reflect hemisphere-specific control
mechanisms. Brain, 136(4), 1288–1303. https://doi.org/10.1093/brain/aws283
Marteniuk, R. G., MacKenzie, C. L., & Baba, D. M. (1984). Bimanual Movement Control:
Information processing and Interaction Effects. The Quarterly Journal of Experimental
Psychology Section A, 36(2), 335–365. https://doi.org/10.1080/14640748408402163
Metrot, J., Froger, J., Hauret, I., Mottet, D., Van Dokkum, L., & Laffont, I. (2013). Motor
recovery of the ipsilesional upper limb in subacute stroke. Archives of Physical Medicine
and Rehabilitation. https://doi.org/10.1016/j.apmr.2013.05.024
Michielsen, M. E., Selles, R. W., Stam, H. J., Ribbers, G. M., & Bussmann, J. B. (2012a).
Quantifying nonuse in chronic stroke patients: A study into paretic, nonparetic, and
bimanual upper-limb use in daily life. Archives of Physical Medicine and Rehabilitation,
93(11), 1975–1981. https://doi.org/10.1016/j.apmr.2012.03.016
Michielsen, M. E., Selles, R. W., Stam, H. J., Ribbers, G. M., & Bussmann, J. B. (2012b).
Quantifying nonuse in chronic stroke patients: A study into paretic, nonparetic, and
bimanual upper-limb use in daily life. Archives of Physical Medicine and Rehabilitation,
93(11), 1975–1981. https://doi.org/10.1016/j.apmr.2012.03.016
Mutha, P. K., & Sainburg, R. L. (2009). Shared Bimanual Tasks Elicit Bimanual Reflexes
During Movement. Journal of Neurophysiology, 102(6), 3142–3155.
https://doi.org/10.1152/jn.91335.2008

155
Mutha, P. K., Sainburg, R. L., & Haaland, K. Y. (2011). Critical neural substrates for correcting
unexpected trajectory errors and learning from them. Brain, 134(12), 3647–3661.
https://doi.org/10.1093/brain/awr275
Naggara, O., Raymond, J., Guilbert, F., Roy, D., Weill, A., & Altman, D. G. (2011). Analysis by
categorizing or dichotomizing continuous variables is inadvisable: An example from the
natural history of unruptured aneurysms. In American Journal of Neuroradiology.
https://doi.org/10.3174/ajnr.A2425
Nakayama, H., Jorgensen, H. S., Raaschou, H. O., & Olsen, T. S. (1994). Recovery of upper
extremity function in stroke patients: The Copenhagen Stroke Study. Arch Phys Med
Rehabil, 75(4), 394–398. https://doi.org/10.1016/0003-9993(94)90161-9
Noskin, O., Krakauer, J. W., Lazar, R. M., Festa, J. R., Handy, C., Brien, K. A. O., & Marshall,
R. S. (2008). Ipsilateral motor dysfunction from unilateral stroke: Implications for the
functional neuroanatomy of hemiparesis. 401–406.
https://doi.org/10.1136/jnnp.2007.118463
Nowak, D. A., Grefkes, C., Dafotakis, M., Küst, J., Karbe, H., & Fink, G. R. (2007). Dexterity is
impaired at both hands following unilateral subcortical middle cerebral artery stroke.
European Journal of Neuroscience, 25(10), 3173–3184. https://doi.org/10.1111/j.1460-
9568.2007.05551.x
Pierpaoli, C., & Basser, P. J. (1996). Toward a quantitative assessment of diffusion anisotropy.
Magnetic Resonance in Medicine, 36(6), 893–906.
https://doi.org/10.1002/mrm.1910360612
Pinter, D., Gattringer, T., Fandler-Höfler, S., Kneihsl, M., Eppinger, S., Deutschmann, H.,
Pichler, A., Poltrum, B., Reishofer, G., Ropele, S., Schmidt, R., & Enzinger, C. (2020).

156
Early Progressive Changes in White Matter Integrity Are Associated with Stroke
Recovery. Translational Stroke Research, 11(6), 1264–1272.
https://doi.org/10.1007/s12975-020-00797-x
Preilowski, B. F. B. (1972). Possible contribution of the anterior forebrain commissures to
bilateral motor coordination. Neuropsychologia, 10(3), 267–277.
https://doi.org/10.1016/0028-3932(72)90018-8
Przybyla, A., Coelho, C. J., Akpinar, S., Kirazci, S., & Sainburg, R. L. (2013). Sensorimotor
performance asymmetries predict hand selection. Neuroscience.
https://doi.org/10.1016/j.neuroscience.2012.10.046
Puig, J., Blasco, G., Schlaug, G., Stinear, C. M., Daunis-i-Estadella, P., Biarnes, C., Figueras, J.,
Serena, J., Hernández-Pérez, M., Alberich-Bayarri, A., Castellanos, M., Liebeskind, D.
S., Demchuk, A. M., Menon, B. K., Thomalla, G., Nael, K., Wintermark, M., & Pedraza,
S. (2017). Diffusion tensor imaging as a prognostic biomarker for motor recovery and
rehabilitation after stroke. Neuroradiology, 59(4), 343–351.
https://doi.org/10.1007/s00234-017-1816-0
R Core Team. (2012). R: A language and environment for statistical computing. R Foundation
for Statistical Computing. https://doi.org/ISBN 3-900051-07-0
Rand, D., & Eng, J. J. (2015). Predicting daily use of the affected upper extremity 1 year after
stroke. Journal of Stroke and Cerebrovascular Diseases, 24(2), 274–283.
https://doi.org/10.1016/j.jstrokecerebrovasdis.2014.07.039
Rinehart, J. K., Singleton, R. D., Adair, J. C., Sadek, J. R., & Haaland, K. Y. (2009a). Arm use
after left or right hemiparesis is influenced by hand preference. Stroke, 40(2), 545–550.
https://doi.org/10.1161/STROKEAHA.108.528497

157
Rinehart, J. K., Singleton, R. D., Adair, J. C., Sadek, J. R., & Haaland, K. Y. (2009b). Arm use
after left or right hemiparesis is influenced by hand preference. Stroke, 40(2), 545–550.
https://doi.org/10.1161/STROKEAHA.108.528497
Rinne, P., Hassan, M., Fernandes, C., Han, E., Hennessy, E., Waldman, A., Sharma, P., Soto, D.,
Leech, R., Malhotra, P. A., & Bentley, P. (2017). Motor dexterity and strength depend
upon integrity of the attention-control system. Proceedings of the National Academy of
Sciences, 115(3), E536–E545. https://doi.org/10.1073/pnas.1715617115
Rorden, C., & Brett, M. (2000). Lesion analysis. Behavioural Neurology, 12(0953–4180), 191–
200.
Rose, D. K., & Winstein, C. J. (2004). Bimanual Training After Stroke: Are Two Hands Better
Than One? Topics in Stroke Rehabilitation, 11(4), 20–30. https://doi.org/10.1310/NCB1-
JWAA-09QE-7TXB
Rowe, J. B., Toni, I., Josephs, O., Frackowiak, R. S. J., & Passingham, R. E. (2000). The
prefrontal cortex: Response selection or maintenance within working memory? Science,
288(5471), 1656–1660. https://doi.org/10.1126/science.288.5471.1656
Royston, P., Altman, D. G., & Sauerbrei, W. (2006). Dichotomizing continuous predictors in
multiple regression: A bad idea. Statistics in Medicine. https://doi.org/10.1002/sim.2331
Rueckert, D., Sonoda, L. I., Hayes, C., Hill, D. L. G., Leach, M. O., & Hawkes, D. J. (1999).
Nonrigid registration using free-form deformations: Application to breast MR images.
IEEE Transactions on Medical Imaging, 18(8), 712–721.
Sadato, N., Yonekura, Y., Waki, A., Yamada, H., & Ishii, Y. (1997). Role of the supplementary
motor area and the right premotor cortex in the coordination of bimanual finger
movements. Journal of Neuroscience, 17(24), 9667–9674.

158
Sainburg, R. (2002). Evidence for a dynamic-dominance hypothesis of handedness.
Experimental Brain Research, 142(2), 241–258. https://doi.org/10.1007/s00221-001-
0913-8
Sainburg, R. L. (2002). Evidence for a dynamic-dominance hypothesis of handedness.
Experimental Brain Research, 142(2), 241–258. https://doi.org/10.1007/s00221-001-
0913-8
Sainburg, R. L., & Schaefer, S. Y. (2004). Interlimb differences in control of movement extent.
Journal of Neurophysiology, 92(3), 1374–1383. https://doi.org/10.1152/jn.00181.2004
Schaefer, S. Y., Haaland, K. Y., & Sainburg, R. L. (2007). Ipsilesional motor deficits following
stroke reflect hemispheric specializations for movement control. Brain, 130(8), 2146–
2158. https://doi.org/10.1093/brain/awm145
Schaefer, S. Y., Mutha, P. K., Haaland, K. Y., & Sainburg, R. L. (2012). Hemispheric
Specialization for Movement Control Produces Dissociable Differences in Online
Corrections after Stroke. Cerebral Cortex, 22(6), 1407–1419.
https://doi.org/10.1093/cercor/bhr237
Schaffer, J. E., & Sainburg, R. L. (2021). Interlimb Responses to Perturbations of Bilateral
Movements are Asymmetric. Journal of Motor Behavior, 53(2), 217–233.
https://doi.org/10.1080/00222895.2020.1760196
Schenkenberg, T., Bradford, D. C., & Ajax, E. T. (1980). Line bisection and unilateral visual
neglect in patients with neurologic impairment. Neurology, 30(5), 509–517.
https://doi.org/10.1212/wnl.30.5.509
Schweighofer, N., Han, C. E., Wolf, S. L., Arbib, M. A., & Winstein, C. J. (2009). A Functional
Threshold for Long-Term Use of Hand and Arm Function Can Be Determined:

159
Predictions From a Computational Model and Supporting Data From the Extremity
Constraint-Induced Therapy Evaluation (EXCITE) Trial. Physical Therapy.
https://doi.org/10.2522/ptj.20080402
Schweighofer, N., Xiao, Y., Kim, S., Yoshioka, T., Gordon, J., & Osu, R. (2015). Effort,
success, and nonuse determine arm choice. Journal of Neurophysiology, 114(1), 551–
559. https://doi.org/10.1152/jn.00593.2014
Serrien, D. J., Cassidy, M. J., & Brown, P. (2003). The importance of the dominant hemisphere
in the organization of bimanual movements. Human Brain Mapping, 18(4), 296–305.
https://doi.org/10.1002/hbm.10086
Shadmehr, R., De Xivry, J. J. O., Xu-Wilson, M., & Shih, T. Y. (2010). Temporal discounting of
reward and the cost of time in motor control. Journal of Neuroscience, 30(31), 10507–
10516. https://doi.org/10.1523/JNEUROSCI.1343-10.2010
Shadmehr, R., Huang, H. J., & Ahmed, A. A. (2016). A Representation of Effort in Decision-
Making and Motor Control. Current Biology, 26(14), 1929–1934.
https://doi.org/10.1016/j.cub.2016.05.065
Sisti, H. M., Geurts, M., Gooijers, J., Heitger, M. H., Caeyenberghs, K., Beets, I. A. M.,
Serbruyns, L., Leemans, A., & Swinnen, S. P. (2012). Microstructural organization of
corpus callosum projections to prefrontal cortex predicts bimanual motor learning.
Learning and Memory, 19(8), 351–357. https://doi.org/10.1101/lm.026534.112
Smith, S. M. (2002). Fast robust automated brain extraction. Human Brain Mapping, 17(3), 143–
155. https://doi.org/10.1002/hbm.10062
Smith, S. M., Jenkinson, M., Johansen-Berg, H., Rueckert, D., Nichols, T. E., Mackay, C. E.,
Watkins, K. E., Ciccarelli, O., Cader, M. Z., & Matthews, P. M. (2006). Tract-based

160
spatial statistics: Voxelwise analysis of multi-subject diffusion data. Neuroimage, 31(4),
1487–1505.
Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Behrens, T. E. J., Johansen-
Berg, H., Bannister, P. R., De Luca, M., Drobnjak, I., Flitney, D. E., Niazy, R. K.,
Saunders, J., Vickers, J., Zhang, Y., De Stefano, N., Brady, J. M., & Matthews, P. M.
(2004). Advances in functional and structural MR image analysis and implementation as
FSL. NeuroImage, 23(SUPPL. 1), S208–S219.
https://doi.org/10.1016/j.neuroimage.2004.07.051
Soares, J. M., Marques, P., Alves, V., & Sousa, N. (2013). A hitchhiker’s guide to diffusion
tensor imaging. Frontiers in Neuroscience, 7(March), 1–14.
https://doi.org/10.3389/fnins.2013.00031
Sohn, Y. H., Jung, H. Y., Kaelin-Lang, A., & Hallett, M. (2003). Excitability of the ipsilateral
motor cortex during phasic voluntary hand movement. Experimental Brain Research.
https://doi.org/10.1007/s00221-002-1292-5
Stephan, K. M., Binkofski, F., Halsband, U., Dohle, C., Wunderlich, G., Schnitzler, A., Tass, P.,
Posse, S., Herzog, H., Sturm, V., Zilles, K., Seitz, R. J., & Freund, H. J. (1999). The role
of ventral medial wall motor areas in bimanual co-ordination. A combined lesion and
activation study. Brain, 122(2), 351–368. https://doi.org/10.1093/brain/122.2.351
Sterr, A., Freivogel, S., & Schmalohr, D. (2002a). Neurobehavioral aspects of recovery:
Assessment of the learned nonuse phenomenon in hemiparetic adolescents. Archives of
Physical Medicine and Rehabilitation. https://doi.org/10.1053/apmr.2002.35660

161
Sterr, A., Freivogel, S., & Schmalohr, D. (2002b). Neurobehavioral aspects of recovery:
Assessment of the learned nonuse phenomenon in hemiparetic adolescents. Archives of
Physical Medicine and Rehabilitation. https://doi.org/10.1053/apmr.2002.35660
Stewart, J. C., Dewanjee, P., Tran, G., Quinlan, E. B., Dodakian, L., McKenzie, A., See, J., &
Cramer, S. C. (2017). Role of corpus callosum integrity in arm function differs based on
motor severity after stroke. NeuroImage: Clinical, 14, 641–647.
https://doi.org/10.1016/j.nicl.2017.02.023
Stewart, J. C., O’Donnell, M., Handlery, K., & Winstein, C. J. (2017). Skilled Reach
Performance Correlates with Corpus Callosum Structural Integrity in Individuals with
Mild Motor Impairment after Stroke: A Preliminary Investigation. Neurorehabilitation
and Neural Repair, 31(7), 657–665. https://doi.org/10.1177/1545968317712467
Stone, K. D., Bryant, D. C., & Gonzalez, C. L. R. (2013). Hand use for grasping in a bimanual
task: Evidence for different roles? Experimental Brain Research, 224(3), 455–467.
https://doi.org/10.1007/s00221-012-3325-z
Sunderland, A., Bowers, M. P., Sluman, S. M., Wilcock, D. J., & Ardron, M. E. (1999a).
Impaired dexterity of the ipsilateral hand after stroke and the relationship to cognitive
deficit. Stroke. https://doi.org/10.1161/01.STR.30.5.949
Sunderland, A., Bowers, M. P., Sluman, S. M., Wilcock, D. J., & Ardron, M. E. (1999b).
Impaired dexterity of the ipsilateral hand after stroke and the relationship to cognitive
deficit. Stroke. https://doi.org/10.1161/01.STR.30.5.949
Takei, T., Lomber, S. G., Cook, D. J., & Scott, S. H. (2021). Transient deactivation of dorsal
premotor cortex or parietal area 5 impairs feedback control of the limb in macaques.
Current Biology, 31(7), 1476-1487.e5. https://doi.org/10.1016/j.cub.2021.01.049

162
Taub, E., Crago, J. E., & Uswatte, G. (1998). Constraint-induced movement therapy: A new
approach to treatment in physical rehabilitation. Rehabilitation Psychology, 43(2), 152–
170. https://doi.org/10.1037/0090-5550.43.2.152
Thrane, G., Emaus, N., Askim, T., & Anke, A. (2011). Arm use in patients with subacute stroke
monitored by accelerometry: Association with motor impairment and influence on self-
dependence. Journal of Rehabilitation Medicine, 43(4), 299–304.
https://doi.org/10.2340/16501977-0676
Todorov, E., & Jordan, M. I. (2002). Optimal feedback control as a theory of motor coordination.
Nature Neuroscience, 5(11), 1226–1235. https://doi.org/10.1038/nn963
Tretriluxana, J., Gordon, J., Fisher, B. E., & Winstein, C. J. (2009). Hemisphere specific
impairments in reach-to-grasp control after stroke: Effects of object size.
Neurorehabilitation and Neural Repair, 23(7), 679–691.
https://doi.org/10.1177/1545968309332733
Uswatte, G., Giuliani, C., Winstein, C., Zeringue, A., Hobbs, L., & Wolf, S. L. (2006). Validity
of Accelerometry for Monitoring Real-World Arm Activity in Patients With Subacute
Stroke: Evidence From the Extremity Constraint-Induced Therapy Evaluation Trial.
Archives of Physical Medicine and Rehabilitation, 87(10), 1340–1345.
https://doi.org/10.1016/j.apmr.2006.06.006
van Beers, R. J. (2009). Motor Learning Is Optimally Tuned to the Properties of Motor Noise.
Neuron, 63(3), 406–417. https://doi.org/10.1016/j.neuron.2009.06.025
Van Den Berg, F. E., Swinnen, S. P., & Wenderoth, N. (2010). Hemispheric asymmetries of the
premotor cortex are task specific as revealed by disruptive TMS during bimanual versus

163
unimanual movements. Cerebral Cortex, 20(12), 2842–2851.
https://doi.org/10.1093/cercor/bhq034
Varghese, R., Kutch, J. J., Schweighofer, N., & Winstein, C. J. (2020). The probability of
choosing both hands depends on an interaction between motor capacity and limb-specific
control in chronic stroke. Experimental Brain Research, 238, 2569–2579.
https://doi.org/10.1007/s00221-020-05909-5
Varghese, R., & Winstein, C. J. (2020). Relationship Between Motor Capacity of the
Contralesional and Ipsilesional Hand Depends on the Side of Stroke in Chronic Stroke
Survivors With Mild-to-Moderate Impairment . In Frontiers in Neurology (Vol. 10, p.
1340).
Vega-González, A., & Granat, M. H. (2005a). Continuous monitoring of upper-limb activity in a
free-living environment. Archives of Physical Medicine and Rehabilitation, 86(3), 541–
548. https://doi.org/10.1016/j.apmr.2004.04.049
Vega-González, A., & Granat, M. H. (2005b). Continuous monitoring of upper-limb activity in a
free-living environment. Archives of Physical Medicine and Rehabilitation, 86(3), 541–
548. https://doi.org/10.1016/J.APMR.2004.04.049
Voineskos, A. N., Rajji, T. K., Lobaugh, N. J., Miranda, D., Shenton, M. E., Kennedy, J. L.,
Pollock, B. G., & Mulsant, B. H. (2012). Age-related decline in white matter tract
integrity and cognitive performance: A DTI tractography and structural equation
modeling study. Neurobiology of Aging, 33(1), 21–34.
https://doi.org/10.1016/j.neurobiolaging.2010.02.009

164
von Monakow, C. (1914). Die Lokalisation im Grosshirn und der Abbau der Funktion durch
Kortikale Herde. In Journal of the American Medical Association: Vol. LXIII (Issue 9).
JF Bergmann. https://doi.org/10.1001/jama.1914.02570090083033
Wang, L. E., Tittgemeyer, M., Imperati, D., Diekhoff, S., Ameli, M., Fink, G. R., & Grefkes, C.
(2012). Degeneration of corpus callosum and recovery of motor function after stroke: A
multimodal magnetic resonance imaging study. Human Brain Mapping, 33(12), 2941–
2956. https://doi.org/10.1002/hbm.21417
Wenderoth, N., Debaere, F., Sunaert, S., van Hecke, P., & Swinnen, S. P. (2004). Parieto-
premotor areas mediate directional interference during bimanual movements. Cerebral
Cortex (New York, N.Y.: 1991), 14(10), 1153–1163.
https://doi.org/10.1093/cercor/bhh075
Wetter, S., Poole, J. L., & Haaland, K. Y. (2005). Functional implications of ipsilesional motor
deficits after unilateral stroke. Archives of Physical Medicine and Rehabilitation, 86(4),
776–781. https://doi.org/10.1016/J.APMR.2004.08.009
Whitall, J., McCombe Waller, S., Silver, K. H., & Macko, R. F. (2000). Repetitive bilateral arm
training with rhythmic auditory cueing improves motor function in chronic hemiparetic
stroke. Stroke, 31(10), 2390–2395. https://doi.org/10.1161/01.str.31.10.2390
White, O., & Diedrichsen, J. (2010). Responsibility Assignment in Redundant Systems. Current
Biology, 20(14), 1290–1295. https://doi.org/10.1016/j.cub.2010.05.069
Winstein, C. (2018). Thoughts About the Negative Results of Clinical Trials in Rehabilitation
Medicine. Kinesiology Review, 7(1), 58–63. https://doi.org/10.1123/kr.2017-0063

165
Winstein, C. J., & Pohl, P. S. (1995). Effects of unilateral brain damage on the control of goal-
directed hand movements. Experimental Brain Research, 105(1), 163–174.
https://doi.org/10.1007/BF00242191
Winstein, C. J., Wolf, S. L., Dromerick, A. W., Lane, C. J., Nelsen, M. A., Lewthwaite, R., Cen,
S. Y., & Azen, S. P. (2016). Effect of a task-oriented rehabilitation program on upper
extremity recovery following motor stroke the ICARE randomized clinical trial. JAMA -
Journal of the American Medical Association, 315(6), 571–581.
https://doi.org/10.1001/jama.2016.0276
Winstein, C., Kim, B., Kim, S., Martinez, C., & Schweighofer, N. (2019a). Dosage Matters: A
Phase IIb Randomized Controlled Trial of Motor Therapy in the Chronic Phase after
Stroke. Stroke, 50(7), 1831–1837. https://doi.org/10.1161/STROKEAHA.118.023603
Winstein, C., Kim, B., Kim, S., Martinez, C., & Schweighofer, N. (2019b). Dosage Matters: A
Phase IIb Randomized Controlled Trial of Motor Therapy in the Chronic Phase after
Stroke. Stroke, 50(7), 1831–1837. https://doi.org/10.1161/STROKEAHA.118.023603
Winstein, C., Kim, B., Kim, S., Martinez, C., & Schweighofer, N. (2019c). Dosage Matters.
Stroke. https://doi.org/10.1161/STROKEAHA.118.023603
Winstein, C., & Varghese, R. (2018). Been there, done that, so what’s next for arm and hand
rehabilitation in stroke? NeuroRehabilitation, 43(1), 3–18. https://doi.org/10.3233/NRE-
172412
Witt, J. K., Proffitt, D. R., & Epstein, W. (2004). Perceiving distance: A role of effort and intent.
Perception. https://doi.org/10.1068/p5090

166
Wolf, S. L., Catlin, P. A., Ellis, M., Archer, A. L., Morgan, B., & Piacentino, A. (2001).
Assessing Wolf Motor Function Test as outcome measure for research in patients after
stroke. Stroke, 32(7), 1635–1639. https://doi.org/10.1161/01.STR.32.7.1635
Wolf, S. L., Winstein, C. J., Miller, J. P., Taub, E., Uswatte, G., Morris, D., Giuliani, C., Light,
K. E., Nichols-Larsen, D., & EXCITE Investigators, for the. (2006). Effect of
Constraint-Induced Movement Therapy on upper extremity function 3 to 9 months after
stroke. JAMA - Journal of the American Medical Association, 296(17), 2095.
https://doi.org/10.1001/jama.296.17.2095
Woodbury, M. L., Velozo, C. A., Richards, L. G., & Duncan, P. W. (2013). Rasch analysis
staging methodology to classify upper extremity movement impairment after stroke.
Archives of Physical Medicine and Rehabilitation, 94(8), 1527–1533.
https://doi.org/10.1016/j.apmr.2013.03.007
Woodbury, M., Nichols-Larsen, D. S., Wolf, S. L., Thompson, P. A., Blanton, S., Morris, D.,
Light, K., Uswatte, G., Velozo, C. A., Taub, E., & Winstein, C. J. (2010). Measurement
structure of the Wolf Motor Function Test: Implications for motor control theory.
Neurorehabilitation and Neural Repair, 24(9), 791–801.
https://doi.org/10.1177/1545968310370749
Woytowicz, E. J., Rietschel, J. C., Goodman, R. N., Conroy, S. S., Sorkin, J. D., Whitall, J., &
McCombe Waller, S. (2017). Determining Levels of Upper Extremity Movement
Impairment by Applying a Cluster Analysis to the Fugl-Meyer Assessment of the Upper
Extremity in Chronic Stroke. Archives of Physical Medicine and Rehabilitation, 98(3),
456–462. https://doi.org/10.1016/j.apmr.2016.06.023

167
Woytowicz, E. J., Westlake, K. P., Whitall, J., & Sainburg, R. L. (2018). Handedness results
from complementary hemispheric dominance, not global hemispheric dominance:
Evidence from mechanically coupled bilateral movements. Journal of Neurophysiology,
120(2), 729–740. https://doi.org/10.1152/jn.00878.2017
Woytowicz, E., Whitall, J., & Westlake, K. P. (2016). Age-related Changes in Bilateral Upper
Extremity Coordination. Current Geriatrics Reports, 5(3), 191–199.
https://doi.org/10.1007/s13670-016-0184-7
Wyke, M. (1966). Postural arm drift associated with brain lesions in man. Archives of Neurology,
15(3), 329–334. https://doi.org/10.1001/archneur.1966.00470150107016
Wyke, M. (1968a). The effect of brain lesions in the performance of an arm-hand precision task.
Neuropsychologia, 6(2), 125–134. https://doi.org/10.1016/0028-3932(68)90054-7
Wyke, M. (1968b). The effect of brain lesions in the performance of an arm-hand precision task.
Neuropsychologia, 6(2), 125–134. https://doi.org/10.1016/0028-3932(68)90054-7
Wyke, M. (1971). The effects of brain lesions on the learning performance of a bimanual co-
ordination task. Cortex, 7(1), 59–72. https://doi.org/10.1016/S0010-9452(71)80022-9
Xu, J., Branscheidt, M., Schambra, H., Steiner, L., Widmer, M., Diedrichsen, J., Goldsmith, J.,
Lindquist, M., Kitago, T., Luft, A. R., Krakauer, J. W., Celnik, P. A., Kim, N., Harran,
M. D., Hertler, B., & Cortes, J. C. (2019). Rethinking interhemispheric imbalance as a
target for stroke neurorehabilitation. Annals of Neurology, 85(4), 502–513.
https://doi.org/10.1002/ana.25452
Yadav, G., Haaland, K. Y., & Mutha, P. K. (2019). Laterality of Damage Influences the
Relationship between Impairment and Arm Use after Stroke. Journal of the International

168
Neuropsychological Society, 25(5), 470–478.
https://doi.org/10.1017/S1355617718001261
Yeh, F. C., Tang, P. F., & Tseng, W. Y. I. (2013). Diffusion MRI connectometry automatically
reveals affected fiber pathways in individuals with chronic stroke. NeuroImage: Clinical,
2(1), 912–921. https://doi.org/10.1016/j.nicl.2013.06.014

Abstract (if available)

Abstract Stroke continues to be the leading cause of adult disability in the US and worldwide, with as many as two-thirds of survivors experiencing some degree of contralesional arm and hand weakness. Conventional rehabilitation practices focus on the recovery of the contralesional arm, but several recent clinical trials investigating the effectiveness of variants of contralesional arm training have reported negative or neutral outcomes. Naturally, efforts to recover and rehabilitate the paretic upper extremity in isolation are of little value if it is not engaged in meaningful functional activitiesㅡactivities that are predominantly bilateral in nature, requiring the coordinated engagement of both the paretic and non-paretic limbs. One alternative to traditional paretic limb training is bilateral upper extremity training; however, its effectiveness has been shown to vary by the side of hemispheric lesion. Therefore, the current dissertation seeks to characterize hemisphere-specific deficits in the control of bimanual movements after stroke. To do this, I first conducted a series of retrospective observational studies (Chapter 3, 4, 5) of an existing stroke database from a previous phase-IIb clinical trial (Dose Optimization for Stroke Evaluation, ClinicalTrials.gov ID: NCT01749358). Then, I collected data in a prospective experimental study (Chapter 6) to uncover the differences in bimanual coordination observed between individuals with left and right hemisphere stroke. ❧ Chapter 3 begins with a retrospective observational analysis in which we studied the factors that influence the spontaneous selection of both hands for bimanual tasksㅡtasks that would otherwise, in age-similar able-bodied individuals, naturally elicit the use of both hands. To capture spontaneous, task-specific choices, we covertly observed 50 pre-stroke right-handed chronic stroke survivors (25 each of left paresis, and right paresis) and 11 age-similar control adults and recorded their hand use strategies for two pairs of bimanual tasks with distinct demands: one with greater precision requirements (photo-album tasks), and another with greater stabilization requirements (letter-envelope tasks). We found that the probability of choosing a bimanual strategy is greater for those with less severe motor impairment and in those with right paresis. However, the influence of these factors, i.e., impairment severity and side of lesion, on bimanual choice varied based on task demands. ❧ In Chapter 4, we further analyze these differences in spontaneous bimanual use by examining the relationship between unimanual performance of the upper extremities in a subset of 42 of the sample of 50 chronic stroke survivors examined in Chapter 3. The purpose of this retrospective analysis was to test the idea that those with right paresis were more likely to use both hands together more because their less-affected left hand would be slower on its own. To do this, we looked at the relationship between the degree of impairment in the contralesional hand, quantified using the Upper Extremity Fugl-Meyer score and the extent of ipsilesional hand deficits, quantified by the distal component of the Wolf Motor Function Test. We found that in those with right paresis, the speed of performance with the ipsilesional hand was proportionally slower to the degree of contralesional impairment. However, this was not the case in those with left paresis. This interaction between ipsilesional hand and side of lesion was observed not only with a measure of contralesional impairment but also contralesional hand function. ❧ In Chapter 5, given the well-established role of the CC for bimanual coordination, especially fibers connecting the larger sensorimotor networks such as prefrontal, premotor, and supplementary motor regions, we examine the relationship between the microstructural status of the CC and bimanual performance in chronic stroke survivors (n = 41). We used movement times for two self-initiated and self-paced bimanual tasks (quantified in Chapter 3) to capture bimanual performance. Using publicly available control datasets (n = 52), matched closely for acquisition parameters, including sequence, diffusion gradient strength and number of directions, we also explored the effect of age and stroke on callosal microstructure. We found that callosal microstructure was significantly associated with bimanual performance in chronic stroke survivors such that those with lower callosal FA were slower at completing the bimanual task. Notably, while the primary sensorimotor regions (CC3) showed the strongest relationship with bimanual performance, this was closely followed by the premotor/supplementary motor (CC2) and the prefrontal (CC1) regions. Furthermore, chronic stroke survivors presented with significantly greater loss of callosal fiber orientation (lower mean FA) compared to neurologically intact, age-similar controls, who in turn presented with lower callosal FA compared to younger controls. The effect of age and stroke were observed for all regions of the CC except the splenium. These findings suggest that in chronic stroke survivors with relatively localized lesions, callosal microstructure can be expected to change beyond the primary sensorimotor regions and might impact coordinated performance of self-initiated and cooperative bimanual tasks. ❧ Lastly, the purpose of Chapter 6 was to understand the principles of responsibility assignment in a bimanual task and thereby uncover mechanisms underlying the previously observed hemisphere-specific effects of stroke. To do this, I used a prospective experimental design and studied a redundant bimanual task wherein we tested the predictions of a leading theory in motor control, known as the Optimal Feedback Control model (OFC). The OFC model suggests that responsibility assignment is a flexible process such that errors are assigned to and corrected for by the limb that is most likely to produce those errors. In right-handed adults, this is often the less-skilled, non-dominant left limb. The flexibility of this process can be probed by examining corrections made by each limb after they have acquired alternative use-dependent experiences, e.g., if the left limb became more skilled, experiencing fewer errors or if the right limb became less skilled, experiencing more frequent errors. Such is the case of stroke affecting the right side of the body, wherein we would predict that the left limb corrects less while the paretic right limb corrects more for task error. ❧ In this prospective study, we tested this prediction in 20 individuals with an intact sensorimotor system, as well as 23 chronic stroke survivors (12 right hemiparesis). Consistent with previous studies, correction gains were asymmetric between the limbs in the non-disabled young control group such that the left limb corrected more than the right limb. Our data also supported our predictions in those with right hemiparesis, but not in those with left hemiparesis. Those with right hemiparesis not only corrected more with their paretic right limb within a trial, but also corrected less with their now more skilled left limb. They also systematically adapted to these errors in a feedforward manner over trials. The extent of correction in stroke survivors did not appear to vary with the degree of motor impairment of the paretic extremity. These findings lead us to conclude that responsibility assignment is not entirely flexible but may be limited to some extent by the hemispheric specialization of motor control processes. Prior to this experiment, I performed an experiment that used an interlimb interference paradigm, motivated by early models of bimanual control. The main findings of this experiment are presented in the Appendix. ❧ Collectively, these studies show that mechanisms of deficits in bimanual coordination after stroke are distinct between individuals with left and right hemisphere damage. For those with left paresis (right hemisphere stroke), error detection or correction is impaired, and it may be why these individuals do not spontaneously use both hands. Large inter-individual variability might suggest sensitivity to the type of error or different mechanisms for correcting it, however we did not explicitly test this idea in this dissertation. Potential rehabilitation strategies might include accuracy training for these individuals. Conversely, in those with right paresis (left hemisphere stroke), error assignment and adaptation seem to be intact. In fact, extensive experience helps the once-less-skilled left limb to calibrate its forward model. However, these individuals were slower than controls. Therefore, potential rehabilitation approaches might include speed training for these individuals.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Relationship between interhemispheric inhibition and bimanual coordination in musicians

PDF

Relationship between brain structure and motor behavior in chronic stroke survivors

PDF

Deficits and rehabilitation of upper-extremity multi-joint movements in individuals with chronic stroke

PDF

Learning reaching skills in non-disabled and post-stroke individuals

PDF

The effects of fast walking, biofeedback, and cognitive impairment on post-stroke gait

PDF

Arm choice post-stroke

PDF

The structure of kinematic variability during hopping: a window into motor control factors in overuse injury

PDF

Hemispheric specialization of reach-to-grasp actions

PDF

The footprint of pain: investigating persistence of altered trunk control in recurrent low back pain

PDF

Using ecological momentary assessment to study the impact of social-cognitive factors on paretic hand use after stroke

PDF

The brain and behavior of motor learning: the what, how and where

PDF

Planning of unconstrained reach actions after unilateral sensorimotor stroke

PDF

Virtual surgeries as a tool for studying motor learning

PDF

Trunk control during dynamic balance: effects of cognitive dual-task interference and a history of recurrent low back pain

PDF

Computational model of stroke therapy and long term recovery

PDF

Development and implementation of a modular muscle-computer interface for personalized motor rehabilitation after stroke

PDF

Experimental and computational explorations of different forms of plasticity in motor learning and stroke recovery

PDF

Modeling motor memory to enhance multiple task learning

PDF

Design of adaptive automated robotic task presentation system for stroke rehabilitation

PDF

Understanding associations between whole-body dynamic balance and spatiotemporal asymmetry during healthy and pathological gait

Asset Metadata

Creator Varghese, Rini (author)

Core Title Hemisphere-specific deficits in the control of bimanual movements after stroke

School School of Dentistry

Degree Doctor of Philosophy

Degree Program Biokinesiology

Degree Conferral Date 2021-08

Publication Date 07/28/2021

Defense Date 06/03/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag bimanual,hemisphere-specific,interlimb,laterality,OAI-PMH Harvest,stroke

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Winstein, Carolee (committee chair), Gordon, James (committee member), Kutch, Jason (committee member), Schweighofer, Nicolas (committee member)

Creator Email rini.matthew9@gmail.com,rinivarg@pt.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15657563

Unique identifier UC15657563

Legacy Identifier etd-VargheseRi-9905

Document Type Dissertation

Format application/pdf (imt)

Rights Varghese, Rini

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu