Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Neural substrates of motor memory consolidation: a double dissociation of primary motor cortex and dorsolateral prefrontal cortex induced by practice structure
(USC Thesis Other)
Neural substrates of motor memory consolidation: a double dissociation of primary motor cortex and dorsolateral prefrontal cortex induced by practice structure
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
NEURAL SUBSTRATES OF MOTOR MEMORY CONSOLIDATION: A DOUBLE
DISSOCIATION OF PRIMARY MOTOR CORTEX AND DORSOLATERAL
PREFRONTAL CORTEX INDUCED BY PRACTICE STRUCTURE
by
Shailesh S. Kantak
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)
May 2010
Copyright 2010 Shailesh S. Kantak
ii
ACKNOWLEDGEMENTS
This thesis is the end of my journey in obtaining my PhD degree in
Biokinesiology. This journey was made richer and more intellectually stimulating by
many people who offered me multiple opportunities to dive in to diverse theories of brain
and behavior. As I was navigating through this tough path, there were many who made
this journey easier with encouragement, and emotional support. It is my pleasure to thank
all those people who were always there for me all along this journey.
I would like to thank my advisor Dr. Katherine J Sullivan. Kathy took me under
her wings at a time I was floundering in my career path. She always guided me during my
research and study at USC. Her insights in to the interactions between cognitive and
motor systems during motor skill acquisition have been very influential in shaping my
research. Her energy, passion for our profession, Physical Therapy has always motivated
me to make a difference. Her support made my research life relatively smooth and
rewarding for me. I owe my deepest gratitude to my co-advisor and mentor, Dr Carolee
Winstein. Her insights in to behavioral and neuroscience research, her attention to minute
details and her ability to question the norms has always motivated me. Dr. Winstein
always challenged me to set my benchmark even higher. She trained me to ask critical
questions that will shape the future of my career, having a life-long impact on my
thinking. She is always accessible and willing to help her students. She always inspired
me to pursue questions in science, and investigate questions carefully and systematically.
iii
Her ability to collaborate across disciplines is just mind-blowing. I am extremely thankful
to her for her ability to inspire me at times when this path felt out of my reach.
I am extremely thankful to Dr. Beth Fisher, who has been my mentor for TMS and
neuroplasticity. In addition to being my mentor, she has been a great person who has
taught me a lot of things from treating patients to neurophysiology of TMS. I am always
amazed at the ease with which Dr. Fisher bridges the gap between motor control/ learning
theory and clinical treatment of patients. I am also indebted to Dr. Barbara Knowlton for
giving me insights in to brain systems for memory. She was always helpful with her
insightful comments and suggestions about our research. I would also like to show my
gratitude to Dr. Stanley Azen for his help in experimental design and statistics. He was
always supportive and encouraged me all along.
Dr. James Gordon deserves special thanks because he was always available to
meet with me to discuss my research ideas right from their inception to publication of my
findings. He always challenged me to question my interpretations of the data, and think
openly to embrace and/ or test alternate explanations. His vision in research and
education has always had an impact me and will shape my career in academia.
All my lab mates at the Motor Behavior and Neurorehabilitation Laboratory made
it a warm friendly place to work. I learned a lot from those who traveled this path, Jool
Tretriluxana and Janice Lin. They helped me during my early days in developing my
thinking, understanding the working in the lab. They were always there to listen and offer
iv
suggestions and help. My buddies, Maureen Whitford, Jill Stewart, Shu-ya Chen, Erica
Pitsch, Alice Lee, Yu-Chen Chung and Hui-Ting Goh made my graduate experience fun
and enriching. Maureen gave me a gift of her friendship that I will cherish all my life.
Hui-ting (Huits) has been a great friend to whom I am highly indebted. If I ever have to
decide on a sabbatical in the future, I will have a hard time deciding whose lab to go to:
Hui-Ting, Jill, Maureen, Janice or Jool.
My deepest gratitude goes to my family for their unflagging love and support
throughout my life; this dissertation is simply impossible without them. My parents,
Shilpa and Shashikant, have always opened doors to my dreams, at times even sacrificing
their own dreams and wants for mine. I can never forget the days when my baba (dad)
used to do small experiments in our kitchen so that I could be able to see what I read in
the textbooks. My aai (mom) always stressed the importance of education, and
encouraged me to dream and achieve big. Both of them always emphasized hard work. It
was this combination that helped me achieve my dream, my doctoral degree. None of my
dreams could have ever realized in truth without my greatest support, my little sister
Shambhavi. She always believed in me, loved me for what I am, supported my dreams
and encouraged me at difficult times. I love you dearly sister; you are the best thing in my
life. My special thanks to my brother-in-law, Amit, who cared for my family in India
when I was busy pursuing my dreams here. I owe you deepest gratitude, Amit. The best
gift my sister and brother-in-law gave me was my niece, Shami. I love her dearly. My
v
other little sister, Pooja, I love you dearly. I hope I am able to help you achieve your
dreams just as you helped me realize and achieve mine. Two of my dearest people whom
I lost on my journey, my aaji (grandmother) and Neeta maushi (aunt) were my biggest
moral supports through out. You are always going to remain with me in spirit.
Last, but not the least, my friends were the greatest source of moral, emotional, and at
times financial support during my stay here in LA. Parind, Ankush, Arun, Komala, Amit-
Nivedita, Niranjan-Priti, Nirav-Poorvangi, Shriniwas, Vishal-Prachi, Arun-Vinita and
many more. Thank you so much!!! My teachers at PTVEM, and KEMH, Mumbai made
me capable enough to take the challenges of the PhD. Thank you very much!
vi
TABLE OF CONTENTS
Acknowledgements ii
List of Figures vii
Abstract xi
Chapter One: Overview 1
Chapter Two: Learning-Performance Distinction and Memory Processes For 4
Motor Skills: A Perspective
Table 1: Summary of motor learning studies that compared the effects 26
of practice/feedback conditions on motor learning by employing
immediate and delayed retention/transfer tests.
Chapter Three: Structure of Practice and Neural Substrates of Motor 34
Skill Acquisition
Chapter Four: Effects of Different Doses of Low Frequency rTMS on 45
Motor Cortical Excitability and Inhibition
Chapter Five: Neural Substrates Of Motor Memory Consolidation 60
Chapter Six: Summary and General Discussion 100
References 114
vii
LIST OF FIGURES
Figure 1: Schematic performance of constant and variable practice groups 7
during acquisition, immediate (IR) and delayed retention (DR) tests.
The benefits of variable practice emerge at the DR, a typical example
of Performance- learning distinction. Time course of memory processes
for motor skills during practice, retention interval and at delayed
retention. motor skills during practice, retention interval and at delayed
retention.
Figure 2: Individual subject data showing change in the MEP amplitude 52
(PRE and POST) with each of the rTMS doses. MEP amplitude is in
µV.
Figure 3: Mean normalized change in the peak-to-peak resting MEP 53
amplitude following each of the 1Hz rTMS doses. The error bars
represent the Standard Error of the Mean (SEM)
Figure 4: Mean normalized change in the peak-to-peak Active MEP 53
amplitude following each of the 1Hz rTMS doses. The error bars
represent the SEM.
Figure 5: Mean normalized change in the Silent period (SP) duration 54
following each of the 1Hz rTMS doses. The error bars represent
the SEM.
Figure 6: A) Participants practiced an arm movement task aimed to match 64
a target presented on the screen. There were 4 targets: A1, A2,
A3, and A4, each with a different absolute amplitude specifications
(30, 45, 60 and 75 degrees, respectively), but a similar movement
structure and same absolute time requirements (800ms). The constant
practice groups practiced 120 rials of target A3, while the variable
practice groups practiced 60 trials of target A3 and 20 trials of targets
A1, A2 and A4 each. The order of presentation of the four targets was
randomized in the variable practice groups. (B) Example of the feedback
display. Participant movement trajectory (thin line) superimposed on
the target trajectory (thick line). The root mean square error (RMSE)
was displayed along with the trajectories after each trial.
viii
Figure 7: Participants practiced the task on Day 1 either under a constant 66
practice condition or variable practice condition. Immediately
following practice, they were tested for end of acquisition (EoA)
performance and immediate transfer. 1day later, participants were
re-tested on a retention/delayed transfer test (RT) to infer learning
of target A3 (Criterion task). Participants from eachpractice condition
(constant and variable) were randomized to a control-no-rTMS group
(CP, VP), M1-interference group (CP-M1, VP-M1), and a
DLPFC-interference group (CP-DLP, VP-DLP). The M1-interference
groups received 1Hz rTMS over M1 and the DLPFC-interference
groups received 1 Hz rTMS over DLPFC, immediately after EoA.
Black dots represent the procedures for measuring motor corticospinal
excitability.
Figure 8: Mean MEP amplitude evoked in the Biceps brachii at three time 73
points: baseline, post-practice and post rTMS in the M1 interference
groups (VP-M1 and CP-M1). There was no difference between the
two groups at any time point. Post-rTMS MEP amplitude was
significantly lower than post-practice MEP amplitude in both groups,
suggesting that rTMS downregulated motor corticospinal excitability
in both the groups.
Figure 9: Practice and End of Acquisition (EoA) performance:Performance 75
of the participants in the control group (open circles), M1
interference group (black filled circles), and DLPFC interference
group (grey filled circles) during the practice phase. The left side
of the figure represents the performance of the participants who
practiced under variable practice structure; the right side of the figure
represents performance of the participants who practiced under constant
practice structure. Each data point for the practice block represents a
mean RMSE of 5 trials on the criterion task (A3). Each data point for
the EoA represents a mean RMSE of the 4-trial EoA test of A3. The
error bars represent the SEM.
Figure 10: End of acquisition (EoA) and retention performance of the 76
participants in the control group (open circles), M1 interference
group (black filled circles), and DLPFC interference group
(grey filled circles). Each data point represents a mean of four trials
of target A3; Error bars represent the SEM.
ix
Figure 11: Data showing change in the performance accuracy from End of 77
Acquisition (EoA) to Retention (R) in individual subjects in all 6
experimental groups.
Figure 12: Retention test performance (Motor learning) Interaction between 79
practice structure (constant practice-striped bars, variable practice-
black filled bars) and rTMS site (no-TMS, M1 and DLPFC) at
retention. rTMS interference to M1, but not DLPFC immediately
after constant practice attenuated retention of the motor skill compared
to control. In contrast, immediately after variable practice, rTMS to
DLPFC, but not M1 attenuated retention of the motor skill compared to
control. Error bars represent the SEM. Note that for the
control-no-rTMS groups, variable practice benefits motor learning more
than constant practice.
Figure 13: Temporal Specificity of rTMS effects End of acquisition (EoA) 81
and retention performance of the participants in the control groups
(open circles), VP-DLP group (grey circle), VP-DLP 4hr group (grey
filled square), CP- M1 group (black filled circle), and CP-M1 4 hr group
(black filled square). Each point represents a mean of 4 trials of EoA
and retention test. The error bars represent SEM.
Figure 14: Immediate and delayed transfer performance for the 50 degree 83
target of the participants in the control group (open squares), M1
interference group (filled squares), and DLPFC interference group
(filled diamonds). The left panel represents the data of the variable
practice groups and the right panel shows the data of the constant
practice groups. Each data point represents a mean of four trials of
target A3; Error bars represent the SEM.
Figure 15: Data showing change in the performance accuracy of individual 84
subjects for the 50˚ Target from Immediate Transfer (IT) to Delayed
Transfer (DT) in all 6 experimental groups.
x
Figure 16: Immediate and delayed transfer performance for the 50 degree 87
target of the participants in the control group (open squares), M1
interference group (filled squares), and DLPFC interference group
(filled diamonds). The left panel represents the data of the variable
practice groups and the right panel shows the data of the constant
practice groups. Each data point represents a mean of four trials of
target A3; Error bars represent the SEM.
Figure 17: Data showing change in the performance accuracy of individual 88
subjects for the 80˚ Target from Immediate Transfer (IT) to Delayed
Transfer (DT) in all 6 experimental groups.
xi
ABSTRACT
Motor practice drives subsequent offline activity within functionally
related resting brain networks. Little is known about how offline neural
networks are modulated by practice structures known to affect motor skill
learning. To investigate the neural correlates of motor memory
consolidation, we applied 1 Hz repetitive Transcranial Magnetic
Stimulation (rTMS) immediately after a bout of constant or variable motor
practice to disrupt either primary motor cortex (M1) or dorsolateral
prefrontal cortex (DLPFC), two putative nodes previously shown to be
engaged in early consolidation. Motor learning was assessed the following
day through a performance-based retention test. Immediately after
constant practice, rTMS to M1, but not DLPFC attenuated retention of the
motor skill. In contrast, immediately after variable practice, rTMS to
DLPFC, but not M1 attenuated retention performance. These findings
provide evidence that for motor skills, the neural substrates of motor
memory consolidation are modulated by practice structure.
1
CHAPTER ONE
OVERVIEW
Acquisition of motor skills involves processes associated with practice or
experience that result in formation of memory representation of the skill (Okano H,
2000). The strength of this motor memory representation is thought to be dependent on
the efficacy of three interdependent processes: encoding, consolidation and retrieval
(Robertson & Cohen, 2006). Motor learning research focuses on understanding these
processes associated with practice or experience that lead to a relatively permanent
change in the capability for motor skill (Schmidt & Lee, 2004). This ―relatively
permanent change‖ distinguishes motor learning from temporary motor performance, a
very well-known phenomenon, the learning-performance distinction. Chapter Two
focuses on how memory processes such as encoding, consolidation and retrieval are
critical to our understanding of the learning-performance distinction different motor
memory processes.
It is well established that when individuals learn multiple tasks or multiple
variants of a task, variable practice leads to better retention and/or transfer of motor skills
compared to constant practice. During practice (encoding), distinct neural substrates are
engaged that differ depending on the practice structure (N. R. Cohen, Cross, Wymbs, &
Grafton, 2009; Cross, Schmitt, & Grafton, 2007; Lin, Fisher, Winstein, Wu, & Gordon,
2008; Lin et al., 2009; Wymbs & Grafton, 2009). However, little is known about how the
2
differences in practice structure affect the neural activity during the immediate post-
practice consolidation phase critical to long-term retention of motor skills. Chapter 3
provides evidence that structure of practice influences the neural activity during different
stages of motor skill acquisition.
This dissertation employs repetitive transcranial magnetic stimulation (rTMS) as a
neuroimaging technique to systematically investigate the neural substrates of motor
memory consolidation. 1Hz rTMS, when administered at an optimal dose, transiently
down regulates cortical excitability. This effect can be used to induce a ―virtual lesion‖
within functionally relevant cortical areas to unravel their role in behavior. Chapter 4
describes the findings of a systematic investigation carried out to identify an optimal 1 Hz
rTMS dose that can significantly and reliably down regulate motor cortical excitability.
This was a crucial step in designing the protocol for this investigation.
The main purpose of this dissertation is to investigate how differences in practice
structure influence the neural activity during the immediate post-practice consolidation
phase. There is evidence to suggest that primary motor cortex (M1) and dorsolateral
prefrontal cortex (DLPFC) are involved during immediate post-practice consolidation.
1Hz repetitive Transcranial Magnetic Stimulation (rTMS) was used to interfere with
processing within one of the two neural substrates (DLPFC or M1) immediately
following variable or constant practice, and the effect of that interference was observed
3
on the retention and transfer test performance 1 day later. Chapter Five summarizes the
results of the following specific aims.
Specific Aims:
Specific Aim 1: To investigate the effects of variability of practice conditions on motor
performance and learning of a discrete ballistic arm movement task. Specifically, we
compare the effects of constant and variable task practice on motor performance and
learning of a discrete ballistic arm movement task in able-bodied young adults.
Specific Aim 2: To investigate the role of M1 during motor memory consolidation in
learning of a discrete ballistic motor skill under constant practice conditions compared
to variable practice conditions using rTMS perturbation
Specific Aim 3: To investigate the effects of rTMS perturbation to DLPFC during the
motor memory consolidation following constant and variable practice, compared to
rTMS perturbation to M1.
Chapter Six will summarize all the results and discuss the implications, limitations and
significance of this work.
4
CHAPTER TWO
LEARNING-PERFORMANCE DISTINCTION AND MEMORY PROCESSES FOR
MOTOR SKILLS: A PERSPECTIVE
Introduction
Acquisition of new movement skills is an essential aspect of life. Right from birth
to old age, we continue to learn different motor skills that enable us to function optimally
in all aspects of life. Be it a marveling performance of Michael Phelps in the Olympic
swimming pool, or recovery of walking after a disabling stroke, both involve acquisition
of pertinent motor skills. Practice is the single most important factor that is critical to
motor skill acquisition. With practice, there are improvements in skill performance.
However, what value would that practice carry if it did not result in a long-term
improvement in the ability of the learner to perform the movement skill? For example, a
patient with stroke may successfully perform a functional task immediately following a
therapeutic practice session. However, critical to determining if the patient has learned
the task is his/her ability to perform the task at a subsequent therapy session or at home
without guidance from the therapist. In other words, the essence of learning lies in its
relative permanence of the capability for motor skill performance. This ―relatively
permanent change‖ lies at the heart of a well-known concept in psychology, the
―learning-performance distinction‖. The learning-performance distinction discriminates
between the observed motor behavior during practice (i.e. motor performance) and the
5
resilience of this behavior that develops over practice and is sustained over time (i.e.
motor learning) (Cahill, McGaugh, & Weinberger, 2001; Schmidt & Bjork, 1992).
Although a well-documented phenomenon, little is known about the mechanisms that
implement the learning-performance distinction.
Practice induces learning-dependent changes in the nervous system such that the
capability for the skilled movement becomes represented as functional networks. These
functional networks represent motor memory. Fuster (1995) defined motor memory as
―representation of motor action in all its forms, from skeletal movement to language,
which is acquired through practice or experience‖ (J. M. Fuster, 1995) The strength of
this motor memory representation is thought to depend on three distinct, but
interdependent memory processes: encoding, consolidation and retrieval (Robertson &
Cohen, 2006). The study of motor learning focuses on understanding these processes that
are involved in the formation and maintenance of motor memory as well as practice-
related factors that influence these memory processes. Recent advances in neuroimaging
techniques such as Transcranial magnetic stimulation (TMS) has allowed direct and
selective perturbation of these memory processes to infer their role in learning(Reis et al.,
2008).
The goal of this article is to integrate the behavioral motor learning phenomenon
of learning-performance distinction with current understanding of memory processes.
Figure 1, below, illustrates a typically observed learning-performance distinction and
6
how memory processes (encoding, consolidation and retrieval) may be operational in
context of motor learning. Figure 1 represents a framework that integrates behavioral
phenomenon (learning-performance distinction) with motor memory processes. Such a
framework may significantly contribute to understanding the mechanisms that implement
the learning-performance distinction. This review is organized in three main sections. In
the first section, we discuss the critical distinction between performance and learning
using behavioral paradigms. We discuss behavioral approaches commonly employed to
make inferences about motor learning, and some limitations of these approaches. In the
second section, we outline the time course of the three motor memory processes during
motor skill acquisition. In the third section, we present findings from recent
investigations which suggest that motor memory processes can be specifically
manipulated to affect performance and learning. Finally, we propose that an
understanding of memory processes in the context of behavioral motor performance and
learning may allow a theoretical framework to generate future hypothesis that help
unravel the brain-behavior relationship during motor learning.
7
Figure 1: Schematic performance of constant and variable practice groups during
acquisition, immediate (IR) and delayed retention (DR) tests. Note that the benefits of
variable practice emerge at the DR, a typical example of the Performance- learning
distinction. Time course of memory processes for motor skills during practice, retention
interval and at delayed retention.
8
Performance-Learning Distinction
In early experiments of learning and memory, Tolman and Hoznik (1930)
observed ―latent learning‖ in rats whose performance did not change with practice on a
maze task until they were motivated with food reward. When provided with motivational
food rewards, the rats were able to demonstrate evidence of having learned the maze.
This suggested that the rats learned the maze task, but their observed behavior was
masked by lack of motivation. These seminal experiments of latent learning were the first
to highlight the important distinction between ‗learning‘, an internal process that is
relatively permanent, and ‗behavior or performance‘, an observable response. In this
case, the observed behavior did not reflect maze learning unless the hunger state was
high.
Human behavior, just like that of animals, is complex and is affected by many
factors such as memory, motivation, and attention. Therefore, in order to draw inferences
about intricate phenomena like memory and learning, one should be careful about the
influence of other factors that might affect behavior (Cahill et al., 2001). Practice of
motor skills can be thought to provide at least two different effects: (1) relatively
permanent effects that we conceptualize as learning, and (2) temporary performance
effects which might be accompanied by a change in mood or motivation (Cahill et al.,
2001; McGaugh, 1989; Schmidt & Bjork, 1992). Performance during acquisition is likely
to be influenced by a number of transient independent variables such as feedback,
9
motivation, or attention that operate in addition to the relatively permanent effects of
practice. Therefore, the performance measures during acquisition may not always reflect
learning, which is an internal process that brings about a relatively permanent change in
the ability to perform the skill. In order to infer learning, it is therefore critical to test the
skill performance after a realistic time interval that allows for dissipation of the
temporary practice effects, and is a better reflection of the permanent effects of practice.
Typically, for motor learning experiments, learning is inferred by observing the learner‘s
performance on a retention or transfer test, typically administered minutes to days after
the practice session, after the temporary effects are thought to have attenuated.
Assessing motor learning
Motor learning involves internal neural processes that cannot be directly
measured. Therefore, learning is typically inferred from observing and quantifying
behavior (Cahill et al., 2001). With motor skill practice, there are improvements in skill
performance evidenced by lower errors and/or increased speed of movement. This
improvement in performance indicates how practice influences the ability to perform the
skill over the acquisition phase.
Assessment of motor performance
Performance curves that examine performance change as a function of practice
trials in acquisition are often used to compare the effect of practice manipulations on
motor performance (Barclay & Newell, 1980; Christina, 1997; Gomez Beldarrain,
10
Grafman, Ruiz de Velasco, Pascual-Leone, & Garcia-Monco, 2002; Shapiro, 1977). In
some studies, however, these performance curves have been used to infer learning
(Christina, 1997; Hernandez et al., 2004). In addition, the net change in the performance
across practice is also used to indicate skill acquisition. For example, in order to measure
learning of an implicit task such as the serial reaction time task (SRTT), a change in the
response time over the acquisition is often measured to give an indication of learning
(Gomez Beldarrain et al., 2002). Some studies report the slope of the performance curves
to indicate the rate of motor skill acquisition. Yet another way to assess the level of
performance in acquisition is by using a dual-task paradigm to determine the degree of
automaticity of performance (Poldrack et al., 2005). All of these performance measures
provide significant insights into the information processing that occurs during the
acquisition phase. However, these performance effects are often short-lived and therefore
may not reflect ―relatively permanent‖ learning effects. Therefore, retention/transfer tests
are often employed to make inferences about learning.
Retention/ Transfer tests assess motor learning
Retention and/or transfer tests involve assessing skill performance under a
common level of independent variable after a certain time interval following the
acquisition phase (Salmoni, Schmidt, & Walter, 1984; Schmidt & Lee, 2004). Retention
and transfer tests differ from each other with respect to the type of test given to the
learner as well as learning-related information that is obtained by the researcher. A test
11
involving the same skill as practiced in the acquisition phase is called retention test. The
purpose of the retention test is to determine the relative permanence of the level of
performance achieved in acquisition over a retention interval. It evaluates the extent to
which the skill is retained by the learner over the retention interval. In effect, retention
test performance reflects the strength of the motor memory representation over time. On
the other hand, transfer tests involve testing the learner on a new variation of the
practiced skill, or on a different, but related skill that was not practiced before. The
transfer test evaluates the generalizability of what is learned during practice. Transfer
tests provide information about the extent to which training in the acquisition phase
produced a level of motor learning that prepares the learner to perform in a post
acquisition situation that differs from the acquisition phase. In other words, transfer tests
may be thought to reflect the flexibility of the motor memory.
Performance and learning are differently affected by practice variables
Research in human motor learning and verbal learning demonstrates that some
training procedures known to enhance immediate performance during acquisition may or
may not benefit long term retention of the skills (learning). Conversely, certain training
procedures that introduce difficulties for the learner and impair acquisition performance
might in fact foster long-term retention of motor skills (T. D. W. Lee, L.R., 2005;
Schmidt & Bjork, 1992). The contextual interference effect (CI) provides a good
example. Practice of multiple tasks (e.g. A, B, C) can be either scheduled in a random-
12
order (e.g. ACBCABCBA...) or in a blocked-order (e.g. AAA…BBB…CCC). During
acquisition, practice in a random-order schedule attenuates performance when compared
to practice under a blocked-order schedule. However, random-order practice enhances
long-term retention compared to that under a blocked-order practice schedule. The
effectiveness of random practice (high interference) over blocked-order practice (low
interference) has been attributed to the contextual interference effect (CI), which refers to
an interference induced by the trial-to- trial variability of the practice schedule. These and
other similar findings highlight the performance-learning distinction. This concept
suggests that what is observed during acquisition may be performance that is localized
and specific for that time and place. In a different context/ learning situation or at a
different time, the learner might perform quite differently and that retention/transfer
performance (indicator of learning) may not be as good or bad as what was observed
during acquisition.
It’s the timing! When to administer the retention/transfer test?
A critical factor in administration of the retention/transfer test is the interval
between the end of acquisition and retention/transfer test. This time interval, called the
retention interval, is extremely variable across studies and depends on the experimenters‘
choice and constraints. Depending on the duration of the retention interval,
retention/transfer tests can be categorized into immediate and delayed retention/transfer
tests. In the literature, the retention interval for immediate retention tests varies from 10
13
seconds to a couple of hours. Delayed tests, on the other hand, are usually conducted 24
hours or more following practice. In fact, it is important to note that these delays (~24
hours) are much shorter than what is seen in many real world examples.
Often it is observed that the effects of practice conditions on motor learning are
robustly evident on the delayed retention/transfer test, but not the immediate
retention/transfer tests. To validate this observation, we reviewed motor learning studies
that employed both immediate and delayed retention/transfer tests (retention interval of
24 hours or more) to evaluate the hypothesis that: in behavioral studies that compared the
effects of different practice and feedback conditions on motor learning, there was a
difference in the findings at delayed compared to immediate retention/transfer tests.
Using Pubmed and PsycINFO databases, we searched the literature to include studies that
investigated the effects of different practice/feedback conditions on motor learning as
inferred by performance on immediate as well as delayed retention/transfer tests. The key
phrases used were ―contextual interference‖, ―variability of practice‖, ―feedback and
motor learning‖ and ―immediate and delayed retention tests, and motor learning‖. We
also used cross references from the research articles that were obtained from the
databases. The selection criteria for articles was intentionally restricted to (1) studies that
investigated the effects of different practice and/or feedback conditions on motor skill
acquisition, and (2) studies that employed both immediate and delayed retention and/or
transfer tests to infer learning. We excluded studies with children or patients with
14
neurological disorders because there is evidence that learning-related cognitive processes
are affected in these populations. The final set of studies consisted of 42 papers.
In the final set of articles, we evaluated whether findings (the effects of
practice/feedback conditions) were similar at immediate and delayed retention/transfer
tests. The studies were categorized into two groups: (1) Studies that demonstrated similar
findings at the immediate and delayed retention tests. We classified them as ‗compatible‘
(C) to indicate that the findings for the two retention tests were compatible with each
other; (2) Studies that had different findings at immediate and delayed retention tests
were classifies as ‗incompatible‘ to reflect that the practice/feedback effects at immediate
and delayed retention tests were not compatible (NC) with each other. We observed that
in 61% of the studies, the findings for immediate and delayed retention tests were not
compatible. That is, the effects of practice/feedback conditions were different as
evidenced by performance at the immediate and delayed retention tests. Furthermore, for
46% of the set, significant differences between the practice/feedback conditions only
emerged during the delayed retention/transfer tests, but not the immediate
retention/transfer tests. In 11% of the set, significant differences between the
practice/feedback conditions were evident at the immediate, but not delayed
retention/transfer test. Finally, in two studies, the findings at the immediate and delayed
retention actually were contradictory to each other. These observations suggest that in at
least 61% of the cases, performance at the immediate retention/transfer tests was not a
15
good predictor of the relatively permanent changes that characterize motor learning.
These observations also suggest that performance at delayed retention/ transfer test,
rather than at practice or immediate retention/transfer test may be a more appropriate
indicator of the relatively permanent change in the capability for the practiced skill
(learning).
Although this distinction between relatively permanent learning and immediate
performance is well-documented, the factors that may contribute to this phenomenon are
not well-understood. Learning constitutes multiple processes associated with practice that
result in formation of memory. In other words, motor memory is a result of the learning
process, and is often conceptualized as a representation of the motor skill embodied in the
networks of the nervous system. In the next section, we discuss different motor memory
processes thought to shed some light on our understanding of the learning-performance
distinction.
Motor memory processes:
Practice and/or experience trigger multiple processes in the nervous system that
constitute learning and result in motor memory formation. Any memory, whether it is for
a fact or motor skill, involves three distinct, yet interdependent processes: encoding,
consolidation and retrieval (Robertson, 2009; Robertson & Cohen, 2006). Although
encoding, consolidation and retrieval are distinct processes, they are interdependent and
may partially overlap in the temporal domain.
16
Encoding of motor memory:
Encoding is a process associated with practice that results in formation of motor
memory. During the encoding phase, the learner processes information related to the task
and makes associations between the goal, movement and movement outcome (Robertson,
2009). The encoding phase involves cognitive processes, whereby, information from the
environment and task is analyzed for its relevance and related to past memories of
movements and respective outcomes. Then, the learner selects a motor response and
employs appropriate force and timing parameters to execute the response. Finally, the
learner evaluates the performance via feedback mechanisms to modulate the future
responses. All these cognitive-motor mechanisms operating during the encoding phase
result in formation of motor memory for the practiced skill. Once a memory
representation is formed through practice, it is stabilized or enhanced during the
consolidation process.
Motor memory consolidation:
Motor memory consolidation is defined as a set of post-acquisition, time-
dependent processes whereby a motor memory (memory representation for a motor skill)
becomes more stable with the passage of time (Krakauer & Shadmehr, 2006; Robertson,
Pascual-Leone, & Miall, 2004). This ―off-line‖ process of consolidation strengthens the
memory representation, which may be behaviorally manifested as an improvement in
performance between practice sessions (offline learning) or increased resistance to
17
retroactive interference (memory stabilization) from a secondary task (Robertson, 2004;
Robertson & Cohen, 2006; Robertson et al., 2004). Recent research indicates that motor
memory consolidation is critical to learning. Interference to memory consolidation
attenuates retention of the practiced skills. Direct evidence for importance of
consolidation in learning comes from studies that selectively and specifically perturb the
consolidation process and assess the effect of that perturbation on retention of motor
skills (Baraduc, Lang, Rothwell, & Wolpert, 2004; Muellbacher et al., 2002; Robertson,
Press, & Pascual-Leone, 2005). For example, low frequency rTMS over primary motor
cortex (M1) immediately following practice of a serial reaction time task blocked any
offline improvements in the skill compared to sham TMS. These disruptive effects of
rTMS on learning were present only when stimulation is applied immediately after
practice, not when rTMS is delayed by a few hours (Robertson et al., 2005). This
indicates that there is an active post-practice neuronal process (consolidation) that is
responsible for offline learning of the skill. Interference to this consolidation impaired
motor learning of the skill, thereby supporting the significance of motor memory
consolidation in learning (retention).
In addition to the processes that immediately follow practice, the motor memory
also undergoes consolidation over the sleep period (Walker et al., 2003; Walker &
Stickgold, 2004). Research in recent years has demonstrated that sleep is critical to
procedural motor skill learning. Walker and colleagues demonstrated that stabilization
18
leads to maintenance of motor performance over time without any further practice, and is
not thought to be affected by sleep. Enhancement or offline learning refers to
improvement in motor skill performance offline without any physical practice, and is
thought to be sleep-dependent. During sleep, motor memories are consolidated through
repeated cycles of non-REM sleep followed by REM sleep (Siengsukon & Boyd, 2009).
Finally, retrieval of motor skills is a critical process in motor skill acquisition and
retention. In the next section, we will review the role of retrieval processes in learning.
Retrieval
Memory retrieval is a fundamental process in learning. Once information is
encoded and stored with practice, it must be retrieved in order to be used. Memory
retrieval is critical to all aspects of daily life, from remembering where you parked your
car to learning a motor skill. In fact, retrieval is the only possible measure of memory
and learning.
Retrieval is defined as the processes involved in accessing information from
stored memories. It encompasses multiple processes such as recall, recognition,
recollection and relearning. Learning and memory are assessed by the ability of the
learner to retrieve the information acquired and stored with practice. Consequently,
retrieval allows us to assess the effectiveness of encoding and consolidation. Despite
being a critical process in memory and learning, few studies in the literature have
specifically and selectively assessed the retrieval processes in motor skill acquisition.
19
Often, it is held ―obvious‖ that retrieval of skills is critical to retention performance.
Some insight may be obtained from studies of verbal learning and memory. Naveh-
Benjamin et al (2006) used concurrent tasks of different difficulties to interfere with
retrieval of a verbal learning task. They demonstrated that divided attention with a
concurrent task affected the performance at retrieval, especially when the testing
conditions were not similar to those during the practice phase (encoding
conditions)(Naveh-Benjamin, Kilb, & Fisher, 2006). This indicates that retrieval is
affected by factors such as attention as well as compatibility with the encoding
conditions.
Integrating memory processes with behavioral motor learning
Figure 1 illustrates how the memory processes discussed above (encoding,
consolidation and retrieval) may be operational during the course of practice, post-
practice interval and long-term retention testing. In this section, we present evidence from
recent studies that used TMS to directly manipulate a specific motor memory process to
affect performance and learning. We use example of the CI effect and present findings
from two experiments: one that specifically manipulated encoding process and other
manipulated the consolidation process. We discuss how the findings from these studies
allow insights in to the mechanisms underlying learning-performance distinction.
20
Selective interference to encoding phase attenuates Contextual Interference effect
Encoding during practice involves cognitive, problem solving processes that
result in formation of a motor memory. Evidence suggests that manipulation of practice
structure provides an opportunity to influence cognitive processes during the encoding
phase (Lee T.D., 1994; Lee TD, 2003; T. D. W. Lee, L.R., 2005; C. H. Shea & Kohl,
1990; J. B. Shea & Wright, 1991). For example, compared to blocked-order practice,
random-order practice engages the learner to cognitively process task-related
information. This deeper cognitive processing leads to a stronger motor memory
representation that enhances learning (retention). Two principal theoretical explanations
for the CI effect are (1) Elaborative-distinctiveness hypothesis and (2) the forgetting
reconstruction hypothesis (Lee TD, 2003). Elaborative-distinctiveness hypothesis
suggests that random-order practice, due to interspersing of the to-be-learned tasks,
provides the learner with an opportunity to compare and contrast the tasks during the
inter-trial interval. In contrast, when the same task is repeated as it is in blocked practice,
the learner has limited opportunity for inter-task comparison. The inter-task comparison
during random-order practice allows the learner to encode critical task-related
information, and leads to a stronger, elaborate memory representation. Forgetting-
reconstruction hypothesis holds that when the same task is practiced repeatedly such as in
blocked-order practice, a previously constructed action plan is likely to be available in the
working memory. In contrast, when multiple tasks are interspersed during practice such
21
as in random-order practice, the learner is forced to ―dump‖ the previously constructed
action plan from the working memory because he/she has to perform a different task on
the subsequent trial. This ―forgetting‖ induced by random practice forces the learner to
reconstruct the action plan again when the initial task is presented again. This added
cognitive processing during random-order practice may be reflected as a poor
performance during practice. However, this process of action plan reconstruction leads to
a stronger motor memory representation that enhances the retention performance
(learning) with random-order practice. Both the above explanations support the notion
that task-related cognitive processing during encoding facilitates a stronger motor
memory representation.
Lin and colleagues (2008) used TMS to selectively interfere with the encoding
processes during blocked- and random- order practice in order to distinguish which of the
two proposed theoretical explanations (Elaborative-distinctiveness or Forgetting-
Reconstruction) predominantly explain the CI effect (Lin et al., 2008). They speculated
that the critical cognitive processes of inter-task comparison and forgetting-
reconstruction occurred mainly during the inter-task interval. Therefore, they
synchronized single TMS pulses applied over the primary motor cortex to each inter-trial
interval to interfere with cognitive processing during blocked- and random-order practice.
They hypothesized that if Elaborative-distinctiveness hypothesis were to better account
for the CI effect, disruptive TMS pulses during inter-trial interval would interfere with
22
the inter-task comparison and diminish the learning benefits of random-order practice;
but would have no effect on learning under blocked-order practice. On the other hand, if
forgetting-reconstruction hypothesis were to hold true, disruptive pulses during intertask
interval would induce forgetting and force the learner to reconstruct the action plan. This
would enhance the learning of blocked practice group such that it would learn similar to
random-order group; with little or no effect on the random group. They found that TMS
disruption over M1 during the inter-trial interval resulted in attenuation of performance
and learning in those who practiced under random-order schedule, thus predominantly
supporting the tenets of the elaborative-distinctiveness hypothesis.
This experiment has significant implications in the context of the current review.
One, it provides evidence that encoding processes can be specifically and directly
manipulated to affect the learning-performance distinction. TMS disruption specifically
timed to the hypothesized encoding processes led to attenuation of the CI effect. In
addition, the direction of this effect of TMS, i.e. diminution of benefits of random-order
practice provides critical insights into the cognitive processing that occurs during
encoding. The inter-task comparison that occurs during the intertrial interval seems
critical to learning benefits reaped by random-order practice. One caveat that this work
presents is that the disruptive effect of TMS on random practice during acquisition was
mirrored at retention. In other words, TMS degraded both acquisition and retention of the
skill under random-order practice.
23
Selective interference to motor memory consolidation provides insights in to the CI effect.
As discussed before, the performance at the immediate retention/transfer test is
often not predictive of the performance at a delayed retention test. One explanation for
this observation is that the memory representation undergoes consolidation during the
time between the immediate and delayed retention test. There is evidence that
consolidation processes evolve over a period of 4-6 hours following practice as well as
over sleep (Brashers-Krug, Shadmehr, & Bizzi, 1996; Muellbacher et al., 2002;
Shadmehr & Holcomb, 1997; Walker & Stickgold, 2004). Often, immediate retention
tests are administered 5-20 min following practice. Therefore performance at the
immediate retention/transfer test may reflect the strength of memory representation
before consolidation has completed. Furthermore, it is possible that different
practice/feedback conditions directly affect the process of motor memory consolidation
(Robertson et al., 2004). If so, this provides a plausible explanation for why subjects
practicing under one condition might perform differently in the delayed retention test but
not in the immediate retention test.
Tanaka and colleagues (2009) applied 1Hz rTMS to primary motor cortex (M1),
Supplementary Motor Area (SMA) and sham immediately after practice of sequential
motor skills under blocked-order and random order training. 1Hz rTMS transiently down
regulates cortical excitability and this effect has been employed as a virtual lesion to
interfere with temporally evolving processes such as motor memory consolidation
24
(Pascual-Leone, Bartres-Faz, & Keenan, 1999; Pascual-Leone, Walsh, & Rothwell,
2000). The effect of the rTMS interference to M1, SMA and sham rTMS on learning
following random- and blocked-order practice was assessed by a retention test a day
following practice. It was found that 1Hz rTMS applied immediately after practice over
supplementary motor area (SMA), but not over control regions or over M1, or over SMA
6 hours post-practiced affected retention of the sequence learning tasks when the tasks
were practiced under blocked-order. When the sequences were practiced in random-
order, rTMS applied over M1, SMA or PMd post-practice did not affect learning
(retention one day later) of the practiced sequences (Tanaka, Honda, Hanakawa, &
Cohen, 2009). One possibility is that by virtue of deeper cognitive processing induced
with random-order practice, the resultant motor memory is more stable and resistant to
rTMS perturbation. This relatively stronger motor memory formed with random-order
practice may implement better performance at the delayed retention test. Motor memory
following blocked practice was susceptible to interference with rTMS applied over SMA.
This may reflect a weaker motor memory that may underlie a relatively poorer
performance with blocked practice compared to random practice. Alternatively, it is also
possible that a different neural substrate (other than those stimulated by the Tanaka
study) may be critical for motor memory consolidation following random practice.
Nevertheless, this study provides evidence that consolidation can be specifically and
directly manipulated to affect the learning performance distinction. In addition, these
25
findings also indicate that structure of practice differently influence the consolidation
processes that impact retention of motor skills.
Conclusion
In this article, we propose a framework that integrates two parallel lines of
research: behavioral motor learning and neuroscience of motor memory. This framework
offers a theoretical basis to design appropriate and timely tests to infer motor
performance and learning. The findings from TMS interference experiments indicate that
the delayed retention/transfer performance likely reflects the efficacy of all the three
motor memory processes i.e. encoding, consolidation and retrieval. Therefore, a delayed
retention/transfer test may be more appropriate reflection of the relatively permanent
change in the capability for the motor skill, i.e. learning. An understanding of the
memory processes in light of behavioral motor skill learning provides an opportunity to
develop specific neurobehavioral hypotheses; operationally define and systematically
manipulate specific processes; and thereby ascertain their unique contribution to learning.
Table 1, below, summarizes the details of each of the 42 studies and the classification (C,
NC) is listed in the last column.
26
Table 1: Summary of the motor learning studies that compared the effects of
practice/feedback conditions on motor learning by employing immediate and delayed
retention/ transfer tests.
Authors Independen
t
variable
Task IR
conditions
(FB, time)
DR
conditions
(FB, time)
IR
findings
DR
findings
Category:
C/ NC
Albaret and
Thon, 1998
Practice
schedule
(random,
blocked)an
d task
complexity
Drawing
task
No KR,
5 min
No KR,
48 hours
Bidirectional
variable
error: No
difference
between the
groups on
retention and
transfer
Bidirectional
variable
error:
significant
group
differences
on retention
and transfer
NC
Anderson
et al, 1994
KR delay:
immediate
vs. after 2
trials
Aiming
task
No KR,
10 min
No KR,
Next day
No
difference
between the
groups on
CE and VE
No
difference
on VE,
Significant
difference
b/w groups
on CE
NC
Anderson
et al, 2001
KR delay +
altered
intrinsic FB
with spring
Aiming
task
No KR,
1min
No KR,
24 hours
No
significant
main effect
or
interaction
Significant
interaction
between KR
delay and
intrinsic FB
NC
Anderson
et al, 2005
KR Delay Aiming
task- 2
days of
practice
No KR,
1 min
No KR,
24 hours
No
difference
between the
groups on
CE and VE
No
difference
on between
the group on
CE and VE;
moderate
effect size
(0.48)
C
Badets and
Blandin,
2005
KR
bandwidth
frequency
with
Observatio
nal learning
Sequenc
e key
pressing
task
No KR,
10 min
No KR,
24 hours
Significant
group effect
for Total
error and
Constant
Error
No
significant
group
differences
NC
27
Table 1, Continued
Brisson
and
Alain,
1996
Type of
augmented
FB (KP,
KR,
KP+KR)
Coinciden
t timing
task
No FB,
5min
No FB,
24 hours
Significant
group
difference
on the score
Significant
group
difference
on the score
C
Brisson
and
Alain,
1997
Type of
augmented
FB
Template
matching
task
No FB,
5min
No FB,
24 hours
No effect for
timing error
Significant
differences
between
groups for
timing error
NC
Brydges
et al,
2007
Practice
schedule
(random/bl
ocked/
whole)
Surgical
skills
FB,
5 min
(retention)
FB,
1 week
(transfer)
No
significant
differences
between
groups
No
significant
differences
between
groups
C
Butki et
al, 2003
KR
frequency
(continuous
, 50%, 100
KR)
Modified
putting
task
No KR, KR
5 min
No KR,
KR,
24 hours
Significant
group
differences
No
significant
group
differences
NC
Butler et
al,
1996
Instructiona
l set during
bandwidth
FB
Two
segment
tapping
task
No FB,
10 min
No FB,
1 day later
Bandwidth
FB group
significantly
better than
other groups
Bandwidth
FB group
significantly
better than
other groups
C
Goodwin
and
Meeu-
wsen,
1995
Bandwidth
KR and
frequency
Putt a golf
ball at a
distance
of 4.57m
No KR and
KR
10 min
No KR and
KR
48 hours
No
difference
between
groups on
no-KR test
Significant
difference
between
groups on
no-KR test
NC
Green
and
Sher-
wood,
2000
Random
and
Blocked
order
practice
Rapid
aiming
task
NO KR,
Immediate
No KR
24 hours
No
significant
effects for
spatial and
temporal
accuracy
No
significant
effects for
spatial and
temporal
accuracy
C
Guay et
al, 1999
KR
summary
and
Spacing
Angular
positionin
g task
No KR,
10 min
No KR,
2 days later
No effect on
CE
Significant
main effect
on CE in
spatial goal
NC
28
Table 1, Continued
Hansen
et al,
2005
Part Vs.
Whole task
practice
Four
componen
t aiming
task
No KR,
Immediate
No KR,
1 day later
Group main
effect
Group main
effect
C
Ishikura.
2005
KR
frequency
and task
complexity
Barrier
knock
down task
No KR,
10 min
No KR,
24 hours
Significant
group
differences
Significant
group
differences
C
Ishikura,
2008
Frequency
of KR
(100%,
33%)
Putting a
golf ball
KR,
10 min
KR,
24 hours
No
significant
group
differences
Significant
group
difference
(100% KR
group had
higher error
than 33%
KR group)
NC
Lai and
Shea,
1999
Frequency
of KR
Movemen
t timing
task
No KR,
10 min
No KR,
24 hour
No effect on
CE/VE
Significant
main effect
of KR
condition on
CE
NC
Lane et
al, 2000
KR and
sensory
information
Coinciden
t
anticipatio
n timing
task
No KR,
5 min
No KR,
24 hours
ACE:
significant
effect of KR;
AE: No
effect; For
VE:
significant
effect of
sensory
information
ACE:
significant
effect of KR;
AE:
Significant
effect of KR;
VE: no
effect of KR
or sensory
information
NC
29
Table 1, Continued
Lee et al,
1997
Practice
condition
(blocked,
random and
random
with
model)
Key
pressing
task in
different
patterns
No KR,
3 min
No KR,
1 day later
For AE:
Random
group
significantly
better than
other two
groups
VE:
random+
model
significantly
more
consistent
than other
two groups
For AE:
Random
group
significantly
better than
other two
groups
VE: random
group
significantly
more
consistent
than the
other two
groups
NC
Li and
Lima,
2002
Practice
schedule:
random as.
blocked
Pass a
soccer
ball at
different
distances
KR,
Immediate
KR,
24 hours
No
significant
effect on the
score
No
significant
effect on the
score
C
Lin et al,
2008
Practice
order
(random,
blocked)
and TMS
interference
Arm
movement
task
No KR
Immediate
No KR
1 day
No
significant
effects
Significant
stimulation
(TMS) and
practice
order
interaction
NC
Liu and
Wris-
berg,
1997
KR delay
and
subjective
estimation
Throwing
a ball at a
target
No KR,
5 min
No KR,
24 hours
Performance
accuracy:
No
significant
effects
Performance
accuracy:
Significant
main effect
of subjective
estimation
NC
Liu and
Wris-
berg,
2005
Practice vs.
control
Targeted
throwing
task
KR?
10 min
KR?
24 hours
Experimenta
l group
significantly
better than
control
group
Experimenta
l group
significantly
better than
control
group
C
Maslovat
et al,
2009
Feedback
present-
ation
(contin-
uous,
discreet)
bimanual
coordinati
on pattern
learning
KR, no KR
Immediate
KR, No KR
1 week
Significant
groupX
testing
condition
interaction
Significant
groupX
testing
condition
interaction
C
30
Table 1, Continued
MuCull-
agh and
Little,
1990
KR
frequency
and
demonstrati
on with
model
(KR-100%,
KR-33%,
and KR+
model)
Timing
task
No KR,
2 min
No KR,
24 hours
Significant
group X
block
interaction
for AE, not
for VE
Significant
group X
block
interaction
for VE, not
for AE
NC
Pollatou
et al,
1997
*** 2
weeks
pracice
Practice
schedule
(random/bl
ocked)
Throwing
and
kicking
tasks
KR,
Immediate
KR, 2
weeks
No effect of
practice
condition on
task
accuracy
Significant
effect of
practice
schedule
NC
Schmidt
et al,
1989
Summary
KR after 1,
5, 10, 15
trials
Simple
ballistic-
timing
task
No KR,
10 min
No KR,
2 days
For CE: No
significant
difference
between
groups
For CE:
Significant
difference
between
groups (15-
trial
summary
group lesser
error than 1
trial
summary)
NC
Sher-
wood,
1996
Practice
schedule
(blocked/ra
ndom)
Rapid arm
reversal
movement
s to either
20/40/60
degrees
No KR,
2 min
No KR
24 hour
CE: Blocked
group had
significantly
higher error
than
Random
group
(typical CI
effect)
CE: Blocked
group had
significantly
higher error
than
Random
group
(typical CI
effect)
C
31
Table 1, Continued
Sidaway
et al,
1992
Summary
KR and
movement
time (MT)
Linear
positionin
g task
with
specific
times
No KR
10 min
No KR,
2days later
No effect of
MT on CE
Main effect
of MT on
CE
NC
Sidaway
et al,
2008
FB
frequency
(100%,
33%) and
type of FB
(KR,
guidance)
Weight-
bearing
task
No KR,
10min
No KR,
24 hours
KR-100%
group most
accurate
than other
groups
KR-33%
group most
accurate
than other
groups
NC
Ste-
Marie et
al, 2004
Practice
schedule
(blocked,
random)
Handwriti
ng skill
No error
correction
information
, 20 min
No error
correction
information
, 24 hours
Random
practice
significantly
better than
blocked on
transfer test;
Exp 2:
Significant
group
difference
between
groups
(random
practice
better than
blocked)
Random
practice
significantly
better than
blocked on
transfer test;
Exp 2: No
significant
group
difference
between
groups
NC
Vander-
Linden
et al,
1993
Kinetic FB
frequency:
100%, 50
% and
concurrent
Isometric
elbow
extension
task to
match a
template
No KR,
Immediate
No KR,
48 hours
Significant
main effect
of group:
RMSE:
concurrent<
100% <50%
Significant
main effect
of group:
RMSE:
concurrent<
100% <50%
C
VanLoon
et al,
1998
Erroneous
KR
Adjustme
nts of
biphasic
movement
s to a
coinciden
ce
anticipatio
n task
No KR,
1 min
No KR,
1week
CE, VE: no
effects
Kinematic
parameters:
No effecf
CE, VE: No
effect;
Kinematic
parameters:
Variablity
significantly
lower in the
no KR group
NC
32
Table 1, Continued
Vera and
Montilla
(2003)
18 weeks
practice
Practie
schedule
(random-
variable,
constant)
Throwing
task
KR
Immediate
KR
2 weeks
Sig
difference
between
groups
Sig
difference
between
groups
C
Weeks
and
Sher-
wood,
1994
Average/
summary/
every trial
KR
Force
productio
n task
No KR,
immediate
No KR,
48 hours
Significant
main effect
of group for
CE;
No
significant
effect on VE
No
significant
effects for
CE;
Significant
main group
effect on VE
NC
Weeks
and
Ander-
son,
2000
Observatio
nal learning
and
practice
Volleyball
serves
No
modeling,
immediate
No
modeling,
48 hours
Significant
group effect
on the
accuracy and
form score
Significant
group effect
on the
accuracy and
form scores
C
Weeks
and
Kordus,
1998
KP
frequency
(100%,
33%)
Multi-
limb
closed
sport skill
No KP,
Immediate
(KR
present)
No KP,
24 hours
(KR
present)
No group
differences
on accuracy
measures
No group
differences
on accuracy
measures
C
Weirnick
et al,
2005
Feedback
on the
virtual
reality
simulation
system
Manual
dexterity:
drilling
grade 1
dental
cavity
NO KR
2 min
No KR
72 hours
No
significant
difference
between the
groups
Significant
group effect:
No FB group
better than
control and
FB
NC
33
Table 1, Continued
Winstein
and
Schmidt,
1990
Exp 2
Relative
frequency
of feedback
Template
matching
task
No KR,
5 min
No KR,
24 hours
No
difference
between
groups
Significant
group
difference,
lower KR
frequency
better
NC
Winstein
et al,
1994
Two forms
of FB
(physical
guidance
and KR),
and
Frequency
Angular
positionin
g task
with a
lever
No KR,
20 min
No KR
1 day
No
significant
effect of FB
form or
Frequency
Significant
interaction
effect
between FB
form and
frquency
NC
Wright
et al,
1997
KR
precision
and
bandwidth
Force
productio
n task
No KR,
20 sec
No KR
24 hours
ACE: no
effect
ACE:
significant
group effect
NC
Note. KR: Knowledge of Results; FB: Feedback; C: Compatible findings at the immediate and
delayed retention/transfer tests; NC: Non-compatible: the findings at immediate retention test
were different that those at the delayed retention test. CE Constant error, ACE: Absolute
Constant error; VE: Variable Error; KP: Knowledge of Performance; RMSE: Root Mean Square
Error.
34
CHAPTER THREE
STRUCTURE OF PRACTICE AND NEURAL SUBSTRATES OF
MOTOR SKILL ACQUISITION
Motor skill acquisition phases:
Motor skill learning is a process that leads to acquisition of complex goal-oriented
movement skills through prolonged practice. Once the skill is learned, it is relatively
well-retained for long periods of time. This enduring nature of motor memory acquired
through practice is unique to procedural skills and makes it a very fascinating process to
study. Research suggests that motor skill acquisition is characterized by performance
changes that occur over two distinct time scales: a fast within practice stage when
considerable performance improvement is observed during practice session, and a slow,
delayed stage that occurs post-practice leading to further gains observed during the
subsequent testing (retention test)(Karni et al., 1998). Motor memory consolidation
occurs during this post-practice period and is known to be time- and sleep dependent
(Krakauer & Shadmehr, 2006; Robertson & Cohen, 2006; Walker & Stickgold, 2004).
During consolidation, motor memory may become more stable which is behaviorally
manifested as an increased resistance to retroactive interference with time (memory
stabilization). Alternatively, motor memory may be enhanced during consolidation as
evidenced by an improvement in performance between practice sessions (offline
learning) (Robertson, 2009).
35
Behavioral studies of motor learning provide evidence that the structure of
practice session strongly influences motor skill acquisition. Motor practice structure can
be characterized on a continuum with a simple structure such as constant practice on one
end and a more complex structure such as variable practice on the other. A constant task
practice structure is drill-like with multiple repetitions of the same task in a row while
variable practice structure is one in which a motor task can be randomly interleaved with
trials of other motor tasks. Compared to constant practice, strong evidence exists that
variable practice enhances long-term retention of a motor skill, although it may degrade
performance during practice. This differential effect of practice structure on practice
performance and delayed retention testing is a typical example of learning-performance
distinction. This effect also raises critical questions about how practice structure itself
affects behavior and neural processes during the two stages of motor skill acquisition.
A wealth of behavioral studies in experimental psychology provides evidence for
distinct cognitive processes that may underlie behavioral differences with each practice
structure. Recently, neuroimaging studies have begun to explore how different practice
structures drive the neural activity during the two stages of motor learning. The thesis of
the present review is to discuss evidence for differences in behavior and neural substrates
between variable and constant practice across two distinct stages of motor skill
acquisition. We highlight how changes in behavior and neural activity associated with
36
each of the practice structure allow critical insights about brain-behavior relationship
during motor learning.
Behavioral basis of motor skill acquisition: Effects of practice structure
When practicing different versions of a motor skill, or multiple different motor
skills, ample evidence exists to suggest that a variable practice structure enhances long-
term retention of the practiced skill compared to constant practice. Often, this beneficial
effect of variable practice is not evident during practice, but emerges more robustly when
tested after some time delay following practice. Behavioral research in motor skill
learning demonstrates that variable practice usually results in less effective performance
during acquisition than a constant practice condition (with-in practice stage). However,
when learning is assessed using retention and transfer tests, typically 24 hours post-
practice (post-practice stage), the variable-practice group demonstrates superior
performance compared to the constant-practice group. This exemplifies the well-known
distinction between temporary changes in motor performance from relatively permanent
learning changes, the learning-performance distinction phenomenon (Cahill et al., 2001).
The performance-learning distinction also suggests that it is likely that practice structure
may influence performance and associated neural substrates differently during each of the
two phases of learning.
37
Effects of practice structure on with-in practice performance change
Theoretical accounts to explain the learning-performance distinction phenomenon
propose that variable practice invokes a higher level of cognitive effort than constant
practice (Guadagnoli & Lee, 2004; Lee T.D., 1994; Sherwood & Lee, 2003). Cognitive
effort refers to the mental work involved in problem-solving and decision making
processes during practice such as anticipation, planning (Lee T.D., 1994). Variable
practice, where a different task/ task version is presented at each trial, forces the learner
to actively process information to select, plan and execute a different skill at every trial.
In contrast, during constant practice, where a single task/task version is practiced at each
trial, the learner requires relatively little cognitive effort because the same plan is
executed every trial. Increased cognitive effort during variable practice may be reflected
as a relatively poorer within-practice performance compared to constant practice (with-in
practice phase). Further, when variable practice involves different versions of the same
task, the learner learns an abstract ―rule‖ that typifies the relationship between the goal
and the parameters (e.g. force, speed) of the movement. This abstract higher-level
representation is thought to underlie a better transfer performance to a novel task
(Schmidt, 1975; Sherwood & Lee, 2003). For example, learning a ―movement rule‖ such
as how to throw a ball overhand under various ―movement parameter specifications‖ such
as different distances (variable practice) will provide increased likelihood of generating
an accurate movement in novel movement situations, compared to if one practices to
38
throw at a single distance (constant practice). Thus, the deeper cognitive processing, and
a higher-level memory representation developed during variable practice is thought to
enhance the capability of the learner to retain the skill better (post-practice stage)
compared to constant practice.
Effects of practice structure on post-practice performance change
Recently researchers have specifically investigated how different practice
structures affect the performance over the post-practice period that involves motor
memory consolidation. Wymbs and Grafton (2009) investigated the effect of blocked-
and random- order practice on changes in post-practice motor performance (Wymbs &
Grafton, 2009). During the training phase, participants practiced 3 motor sequences under
blocked-order schedule, and 3 other motor sequences under random-order schedule such
that by the end of training, all six sequences (3 blocked, 3 random) were practiced an
equal number of times. Participants were tested on a retention test one day after the
training phase. Off-line learning was characterized by a motor performance change at the
retention test phase compared to the end of training. The effect of practice structure on
offline learning was determined by comparing the motor performance change on the
block-trained sequences to those of the random-trained sequences. Random-trained
sequences were executed faster at the retention test compared to the end of training,
providing a strong indication of offline learning following random-order training. In
contrast, blocked-trained sequences did not demonstrate any offline improvements. These
39
findings suggest that, in addition to with-in practice effects, the structure of practice itself
may influence post-practice processes to affect motor learning.
Neural basis of motor skill acquisition: Effects of practice structure
It is evident from animal and human research that a broad neural network
including cortical and sub-cortical structures is involved during motor skill acquisition.
However, learning-associated brain networks differ depending on the type of motor skill
(procedural or declarative) (Boyd & Winstein, 2006; Destrebecqz et al., 2005; Honda et
al., 1998), the task (sequence learning, adaptation, motor skill) (Ghilardi et al., 2000), and
learner characteristics (novice or expert) (Landau & D'Esposito, 2006; Meister et al.,
2005). Despite ample behavioral evidence supporting differences in cognitive processes
associated with different practice structures, little is known about how the different
practice structures affect the brain networks during motor skill acquisition. However, a
distinct effect of practice structure on motor performance during two stages of motor skill
acquisition has prompted researchers to investigate the neural changes that are associated
with different practice structures.
Neural basis of with-in practice performance changes: Effects of practice structure
Recently, investigators have begun to explore how differences in practice
structure influence the brain networks during motor practice. Evidence for differences in
brain activation during blocked and random practice was first reported using a functional
magnetic resonance imaging (fMRI) approach by Cross and Colleagues (Cross et al.,
40
2007). One group of participants practiced three key-pressing sequences in a blocked-
order and the other group practiced the sequences in random-order. Brain activation was
assessed with fMRI during response preparation phase and response execution phase
across each of the practice condition (blocked and random). During the course of
practice, participants in random-practice group demonstrated an increased activation in
the superior and middle frontal gyri during response execution. It is likely that task-
switching during random practice may invoke cognitive processing activities such as
attending to the new sequence, working memory and selecting a correct action plan to
match the sequence, all of which may involve processing within the prefrontal cortex.
Further, as practice proceeded, participants in the random-practice group showed
increased activity in the premotor and motor cortices during response preparation
compared to those in the blocked-practice group. This increased activation may reflect
more cognitive processing during the preparation phase of the random-order trials.
A more definitive support for increased engagement of motor cortex during
random practice compared to blocked practice comes from TMS experiments. When
single pulses of TMS were applied over the motor cortex during the inter-trial interval
(Lin et al., 2008; Lin et al., 2009), or during the preparation phase of every trial (N. R.
Cohen et al., 2009), the learning benefits of random practice were attenuated compared to
when sham stimulation was applied over the motor cortex. However, the stimulation had
no effect on learning when participants practiced the tasks in blocked-order. This
41
provides clear evidence that motor cortex may be more actively engaged during random
practice, compared to blocked practice. Recently, Wymbs and Grafton demonstrated that
activity in ipsilateral primary motor cortex during practice significantly correlated with
post-practice offline learning of sequences that were practiced in random-order (Wymbs
& Grafton, 2009). These findings, taken together, suggest that random-order practice may
increase the activity within the primary motor cortices during practice, which may in part,
enhance retention by influencing the post practice phase of learning.
Neural basis of post-practice performance changes: Effects of practice structure
Recently, investigators have begun to investigate the neural basis of post-practice
processes such as motor memory consolidation that contribute to motor learning. Alberts
and colleagues used fMRI to characterize changes within resting neural networks after
practice of a visuomotor adaptation task (Albert, Robertson, & Miall, 2009). In their
study, test-group participants practiced moving a cursor with a joystick to targets
presented radially from the starting position where they learned a novel relationship
between cursor and joystick. The control group performed similar movements but with
veridical cursor feedback of the joystick. Both groups underwent functional MRI during
rest before and after the motor task practice (test group) or performance (control group).
Analysis of the BOLD signal at the two time points (pre- and post-) demonstrated that
motor practice, and not just performance modulated the activity within fronto-parietal and
cerebellar neural networks. This post-practice modulation of the resting fronto-parietal
42
neural network may reflect, in part, early processes of motor memory consolidation.
Shadmehr and Holcomb (1996) reported that as participants practiced to adapt to a force-
field during reaching movements, there was a significant increase in PET regional
cerebral blood flow (rCBF) within DLPFC (BA 46). When participants were tested 6
hours after practice, the rCBF within DLPFC had decreased significantly compared to its
immediate post-practice level despite no apparent change in performance (Shadmehr &
Holcomb, 1997). This time-dependent modulation of DLPFC activity during the post-
practice period may imply its role in motor memory consolidation.
Evidence from animal and human research suggests that primary motor cortex is
involved in early motor memory consolidation following practice. Inhibition of protein
synthesis in contralateral M1 immediately following practice significantly affected
learning of a skilled reaching task without affecting basic motor execution, indicating that
post-practice protein synthesis in M1 is an important step to motor memory consolidation
(Luft, Buitrago, Kaelin-Lang, Dichgans, & Schulz, 2004; Luft, Buitrago, Ringer,
Dichgans, & Schulz, 2004). Human studies using transcranial magnetic stimulation have
also demonstrated the role of M1 in motor memory consolidation following practice of
ballistic tasks (Classen, Liepert, Wise, Hallett, & Cohen, 1998; Muellbacher et al., 2002;
Robertson et al., 2005). (Muellbacher et al., 2002) demonstrated that low frequency
repetitive transcranial magnetic stimulation (LF-rTMS) of M1, but not other brain areas,
disrupted the retention of the behavioral improvement on skilled finger-movement tasks
43
if it was applied immediately after training. Basal motor behaviors, task performance,
motor learning by subsequent practice were not affected. These studies indicate that M1-
mediated motor memory consolidation is critical to retention of practiced motor skills.
What remains to be elucidated is how different practice structures may influence the
neural substrates critical for motor memory consolidation. Until now, only one study has
investigated how SMA activity is differentially modulated by different practice structures
(Tanaka et al., 2009). Tanaka and colleagues (2010) demonstrated that 1Hz rTMS over
supplementary motor area (SMA), but not over control regions or over primary motor
cortex immediately after practice or over SMA 6hours post-practiced affected retention
of the sequence learning tasks when the tasks were practiced under blocked-order. When
the sequences were practiced in random-order, rTMS applied over M1, SMA or PMd
post-practice did not affect learning (retention) of the practiced sequences. These findings
suggest that SMA is more engaged during early motor memory consolidation following
blocked practice. In addition, these findings may imply that consolidation may
concurrently occur as the practice proceeded to yield a stronger motor memory
representation that may not require post-practice consolidation. Alternatively, motor
memory consolidation following random practice may involve neural substrates other
that those stimulated in the Tanaka et al study.
In the present dissertation work, we aimed at systematically investigating how
differences in practice structures affected the neural substrates of motor memory
44
consolidation. We plan to use rTMS to interfere with processing of either M1 or DLPFC
post practice following constant and variable practice and observe the effect of this
perturbation on retention of the practiced motor skill.
45
CHAPTER FOUR
EFFECTS OF DIFFERENT DOSES OF LOW FREQUENCY rTMS ON MOTOR
CORTICAL EXCITABILITY AND INHIBITION
Introduction
Low Frequency (1 Hz) repetitive transcranial magnetic stimulation (LFrTMS/
1Hz rTMS) is a useful method to study brain-behavior relationships by modulating
cortical excitability. In addition it is also used to influence learning and rehabilitation in
patients with neurological disorders. 1Hz rTMS has been shown to suppress motor
corticospinal excitability (Chen et al., 1997; Fitzgerald, Fountain, & Daskalakis, 2006).
However, the suppressive effect of 1 Hz rTMS shows considerable variability across
studies. Many studies have failed to show any modulation in corticospinal excitability
with LFrTMS while in other studies, the reported decrease in motor cortcospinal
excitability ranges in magnitude from a 16-30% (Chen et al., 1997; Maeda, Keenan,
Tormos, Topka, & Pascual-Leone, 2000b; Todd, Flavel, & Ridding, 2006) and in
duration from 10 minutes to 1 hour (Iyer, Schleper, & Wassermann, 2003; Romero,
Anschel, Sparing, Gangitano, & Pascual-Leone, 2002).
There are a number of factors that contribute to the observed variability in
LFrTMS studies. First, there are some general features of TMS methodology that result
in experimental variability such as inconsistent coil position and angle. Second, inter-
subject variability may arise from individual factors such as attention, age, or differences
46
in resting muscle tone. Indeed, Maeda et al (2000 b) reported considerable inter-
individual variability in the modulation of cortical responses to rTMS. Similarly,
Gangitano et al. (2002) identified two subpopulations of subjects with different patterns
of cortico-spinal modulation after application of rTMS. One group showed a suppression
of corticospinal excitability with 1Hz rTMS and an increase in the corticospinal
excitability after 20 Hz rTMS; while the other group showed the opposite pattern of
modulation (increase in cortical excitability after 1Hz rTMS and decrease in cortical
excitability after 20 Hz rTMS)(Gangitano et al., 2002).
Finally, the specific stimulation parameters of rTMS such as frequency, intensity
and duration can influence the nature of its effects on corticospinal excitability. For
example, 1 Hz rTMS delivered at a suprathreshold intensity (115%RMT) resulted in
MEP suppression, whereas 1 Hz rTMS delivered at a subthreshold intensity (85% RMT)
had no effect on Motor Evoked Potential (MEP) amplitude (Fitzgerald, Brown,
Daskalakis, Chen, & Kulkarni, 2002). Other studies have reported MEP suppression
following subthreshold low frequency rTMS (Gangitano et al., 2002; Maeda, Keenan,
Tormos, Topka, & Pascual-Leone, 2000a; Maeda et al., 2000b; Romero et al., 2002;
Touge, Gerschlager, Brown, & Rothwell, 2001). Further difficulty in comparing rTMS
effects across studies arises from differences in dependent measures employed to probe
cortical excitability. Some studies measured MEP amplitude tested with suprathreshold
single pulse TMS, while others used input-output curves, stimulating at varying intensity
47
levels (Gangitano et al., 2002; Maeda et al., 2000a, 2000b; Romero et al., 2002; Touge et
al., 2001). Finally, the MEPs recorded by EMG in response to the same stimulation
intensity at a particular motor cortex site can be highly variable because of fluctuations in
the cortical as well as spinal segmental motoneuron excitability levels (Kiers, Cros,
Chiappa, & Fang, 1993).
The overall goal of this study was to find an optimal dose of 1 Hz rTMS that
reliably and consistently down regulates motor cortical excitability. We systematically
compared the effects of different 1Hz rTMS doses on motor corticospinal excitability,
while trying to minimize the other sources of variability described. A neuronavigation
system was employed to minimize the variability in the coil position and angle. We
attempted to minimize the intersubject variability by using a with-in subject design to
systematically compare the effects of four different LFrTMS doses on motor
corticospinal excitability. We also standardized the time of testing for all individuals to
reduce the variability due to diurnal factors. In addition, we ensured that the participants
were alert during the entire testing session. The four 1Hz rTMS doses were designed with
a combination of two intensities (subthreshold- 90%MT and suprathreshold-110%) and
two durations (10 min and 20 min).
48
Methods
Subjects
Nine right-handed volunteers (3 men, 6 women), aged 23- 34 (mean, 26.2 + 2.9
years) participated in the study. All participants gave written informed consent and the
protocol was approved by the Institutional Review Board of the University of Southern
California. All subjects were screened for TMS safety prior to the experiment, and
showed no contraindication to TMS in their medical, personal or family history (Kleim,
Kleim, & Cramer, 2007; Rossi, Hallett, Rossini, & Pascual-Leone, 2009; Wassermann,
1998).
EMG recording
Surface EMG was recorded from the right first dorsal interosseous (FDI) muscle
with disc electrodes placed in a tendon-belly arrangement over the bulk of the muscle and
the metacarpo-phalyngeal joint of the index finger. The EMG signal was filtered with a
bandpass of 1–1000 Hz, amplified, and digitized at 2000 Hz. The data was graphically
displayed and stored for offline analysis
Experimental design
We employed a within-subject design in which participants underwent one of four
1Hz rTMS sessions, at least a week apart. At each session, participants received one of
the following 1Hz rTMS doses (intensity-duration combinations): Subthreshold
(90%MT) rTMS for 10 minutes, Subthreshold (90%MT) rTMS for 20 minutes,
49
Suprathreshold (110%MT) rTMS for 10 minutes, and Suprathreshold (110% MT) rTMS
for 20 minutes. The order of the doses was counterbalanced across participants.
Measures of cortical excitability and inhibition
Motor corticospinal excitability was quantified by measuring the average peak-to-
peak amplitude of motor evoked responses (MEP amplitude) to suprathreshold (120% of
resting motor threshold) TMS pulses prior to, and immediately after each rTMS session.
MEP amplitude was measured at rest and during active contraction. Motor cortical
inhibition was quantified by silent period.
A brain navigation system, the Brainsight frameless stereotaxy system, was used
to precisely guide the position of the coil over the motor cortex. A sample MRI brain
image was used for all participants. A three dimensional (3D) image of the cortex was
reconstructed in Brainsight by processing a two-dimensional MR image. The
neuronavigation system allowed for systematic sampling of the stimulation sites as well
as accurate placement of the coil during stimulation (Kleim et al., 2007). Markers were
then placed at specific anatomical landmarks on the MR image. The participant‘s
anatomical landmarks were then coregistered with the anatomical landmarks on the MR
image such that it ensured an optimal transformation between actual skin points on the
participant and the skin surface of the MRI reconstructed human model. Then the TMS
coil was calibrated. This allowed a real time display of the relative positions of the coil
50
and the participant‘s head and brain surface, which was critical in guiding the placement
of the coil over motor cortex.
Participants were seated in a comfortable chair with their forearm supported in a
prone position and hand resting on arm support. Each session lasted approximately 2
hours. Single TMS pulses were applied over the left motor cortex with a 70mm figure of
eight coil attached to Magstim Rapid Stimulator (The Magstim Company). The coil was
held tangentially to the scalp with the handle pointing posteriorly away from the midline
at an angle of 45 . Current induced from this position is directed approximately
perpendicular to the central sulcus (Brasil-Neto, 1992; Mills, 1992). A ―hot-spot‖ for FDI
was determined as the site at which the largest MEP was obtained from FDI at lowest
TMS intensity. The coil was then fixed over the hotspot for the rest of the experiment and
the intensity was systematically reduced to determine the resting motor threshold (RMT).
RMT is the minimum TMS intensity required to invoke MEP amplitude of at least 50µV,
in 5 out of 10 consecutive trials.
Motor corticospinal excitability was assessed prior to- (pre) and immediately after
(post) the rTMS protocol. Ten pulses of TMS were then applied over the hot-spot at
120%MT intensity under two conditions: resting and active. For the resting condition, the
participants maintained relaxation of the FDI muscle. For the active contraction
condition, the participant adducted the index finger to the transducer‘s force pad (Jamar
hydraulic dynamometer) and pressed to a force of 10 % of the maximum voluntary force.
51
Care was taken to avoid fatigue during the entire session. The recorded motor evoked
potentials (MEPs) from FDI were amplified, digitized and stored for off-line analysis.
rTMS procedure
1Hz repetitive TMS (rTMS) was delivered to the left motor cortex with a 7 cm
figure of eight coil held tangential to the scalp in a posterior-anterior direction and
applied using the Magstim Rapid stimulator (The Magstim Company). At each of the
four sessions separated by at least a week, every participant received one of the following
four doses (intensity-duration combinations) of 1Hz rTMS: (1) 90% RMT intensity for 10
min, (2) 90% RMT for 20 min, (3) 110% RMT for 10 min, and (4) 110% RMT for 20
min. The measures of corticospinal excitability and inhibition were obtained before and
after each rTMS session.
Data Analysis
MEPs were analyzed offline using DataWizard, a MATLAB-based program.
Peak-to-peak amplitude was measured for each recorded MEP. For both resting and
active conditions, mean peak-to-peak amplitude was calculated for 10 MEPs pre- and
post-rTMS. Silent period (SP) was calculated from the TMS pulse to the return of EMG
period after the MEP (Saisanen et al., 2008). Mean change from baseline (pre-rTMS) in
MEP amplitude and SP was compared among the four doses using Repeated measures
ANOVA. A paired t test was used to compare the pre-TMS MEP amplitude and SP to
post-TMS MEP amplitude and SP respectively.
52
Results:
Resting MEP amplitude
Figure 2, below, summarizes the effect of four rTMS doses on resting MEP
amplitude in individual subjects. Out of the four 1Hz rTMS doses, three significantly
downregulated resting motor cortical excitabilty (Figure 3). 1 Hz rTMS at 90%MT for 20
min (t= 4.324, p= 0.003), 110%MT for 10 (t= 6.274, p< 0.001) and 20 minutes (t= 3.352,
p= 0.012) led to significant reduction in the resting MEP amplitude compared to baseline.
There was no significant difference in mean MEP amplitude reduction between the three
doses (p=0.721). 1Hz rTMS at 90%MT applied over M1 for 10 min did not significantly
affect MEP amplitude compared to baseline (t= 1.362, p= 0.21).
10 minutes 20 minutes
Figure 2: Individual subject data showing change in the MEP amplitude (PRE and
POST) with each of the rTMS doses. MEP amplitude is in µV.
90 %
RMT
110%
RMT
Duration
53
Figure 3: Mean
normalized
change in the
peak-to-peak
resting MEP
amplitude
following each
of the 1Hz
rTMS doses.
The error bars
represent the
Standar Error of
the Mean (SEM)
Active MEP amplitude
There was no significant effect of any of the rTMS doses on active MEP amplitude (90%
RMT for 10 min: t= 1.003, p= 0.349; 90% RMT for 20 min: t= 1.287, p= 0.234; 110%
RMT for 10 min: t= .207, p= 0.842; 110% RMT for 20 min: t= 0.54, p= 0.604; Figure 4).
Figure 4: Mean
normalized change
in the peak-to-peak
Active MEP
amplitude following
each of the 1Hz
rTMS doses. The
error bars represent
the SEM.
54
Silent period
Of the four 1Hz doses, only subthreshold (90% RMT) rTMS applied for 10 min
significantly lengthened the silent period (t= 3.669, p= 0.008; Figure 5). The other three
doses had no significant effect on the silent period duration (90% RMT for 20 min:
t=0.51, p= 0.624; 110% RMT for 10 min: t= 0.958, p= 0.37; 110% RMT for 20 min: t=
1.08, p= 0.308).
Figure 5: Mean normalized change in the Silent period (SP) duration following each of
the 1Hz rTMS doses. The error bars represent the SEM.
Discussion
The present study revealed three main findings. First, 1Hz rTMS applied at
suprathreshold intensity (110% RMT) reduced corticospinal excitability at rest,
irrespective of the duration of stimulation. Second, subthresold 1Hz rTMS decreased
corticospinal excitability at rest only when applied for a longer duration (20 min
55
compared to 10 min). Third, interestingly, subthreshold rTMS when applied for 10 min
was able to induce cortical inhibition as evidenced by lengthening of the silent period.
Multiple studies have investigated the effects of intensity and duration of 1 Hz rTMS on
measures of motor cortical excitability, facilitation and inhibition(Chen et al., 1997;
Fitzgerald et al., 2002; Heide, Witte, & Ziemann, 2006; Houdayer et al., 2008; Maeda et
al., 2000a, 2000b; Muellbacher, Ziemann, Boroojerdi, & Hallett, 2000; Romeo et al.,
2000). The results of these are highly variable, and there is little consensus about the
optimal dose to effectively downregulate cortico-spinal excitability. In this study, we
systematically manipulated intensity and duration of 1 Hz rTMS to yield four doses of
rTMS, and compared the response to these doses in the same individuals. In good
accordance with previous studies (Lang et al., 2006; Muellbacher et al., 2000; Plewnia,
Lotze, & Gerloff, 2003), suprathreshold rTMS at 1Hz applied for 10 min or 20 min was
able to reduce the resting MEP amplitude. However, there was no significant difference
in the magnitude of downregulation between these two doses of suprathreshold 1Hz
rTMS. This is a new finding which demonstrates that corticospinal excitability can be
downregulated with suprathreshold 1Hz rTMS even at durations as short as 10 min. In
contrast, subthreshold (90% RMT) rTMS at 1 Hz was able to reduce the resting MEP
amplitude only when applied for longer duration, but not when applied for shorter
duration. This is consistent with some of the previous findings which indicate that it may
56
be necessary to apply longer trains of sub threshold, 1 Hz rTMS for consistent and
significant downregulation of motor corticospinal excitability (Maeda et al., 2000a).
Consistent with the previous literature, suprathreshold 1Hz rTMS led to significant
downregulation of corticospinal excitability compared to subthreshold 1Hz rTMS,
specifically for shorter duration of stimulation. Multiple mechanisms may underlie the
increased efficacy of suprathreshold rTMS effects to downregulate corticospinal
excitability. It is likely that suprathreshold rTMS influences a larger pool of motor
cortical neurons compared to subthreshold rTMS, thereby inducing a larger effect in the
global excitability measure (MEP amplitude). Functional imaging studies have
demonstrated that suprathreshold rTMS at 1 Hz may, through neuronal connections, may
influence other non-primary motor areas such as the dorsal premotor cortex thus yielding
a stronger suppression of motor cortical excitability (Speer et al., 2003). Finally, afferent
feedback generated by suprathreshold stimulation-evoked muscle twitches may also
suppress the motor cortical excitability, and enhance the downregulating effects of
suprathreshold rTMS at 1 Hz on corticospinal excitability(Lang et al., 2006).
None of the four 1Hz rTMS doses had a significant effect on MEP amplitude
under the active condition (Active MEP). This finding is consistent with previous reports
that did not demonstrate any change in active MEP with subthreshold (Touge 2001) or
suprathreshold 1Hz rTMS (Romeo, 2000). It is likely that tonic voluntary contraction
increases the overall excitability of the corticospinal system and may mask the effects of
57
LFrTMS on the corticospinal excitability. Therefore, it is likely that in our study, the
downregulation effect of rTMS on MEP amplitude may not be detected in active
contraction conditions. These findings however differ from those of Fitzgerald (2002)
who demonstrated a significant decrease in active MEP following supra, but not sub-
threshold 1Hz rTMS (Fitzgerald et al., 2002). In the Fitzgerald study (2002), active MEPs
were recorded with TMS pulses at an intensity of 125% of Active Motor Threshold
(AMT). AMT is the lowest intensity required to produce at least 1 MEP of 100µV in 5
trials as the subjects sustained a low intensity contaction (5-10% of MVC). Further, if
AMT changed in individual subjects following rTMS, post rTMS measures were
acquired with this new AMT. In contrast, in our study, stimulation intensity for the
active condition was unvarying at 120% of the pre-rTMS resting motor threshold (RMT)
both before and after rTMS. This may have resulted in relatively higher stimulation
intensities for MEP recording in our study compared to those used in the Fitzgerald study
(2002), since RMT is typically higher that AMT. It is likely that relatively higher
intensity of stimulation in our study may have also masked the reduced corticospinal
excitability in active contraction condition. These differences in methods between
Fitzgerald (2002) and our study may likely contribute to the differences in the findings
between the two studies.
A rather surprising finding of our study was that the only subthreshold 1Hz rTMS
applied for shorter duration(10 min) significantly lengthened silent period (SP) in our
58
participants. There was no significant change in SP duration with remaining rTMS doses.
Daskalakis et al 2006 demonstrated an increase in SP duration with subthreshold
(90%RMT) for 15 min (Daskalakis et al., 2006). In contrast, several others did not show
an effect of subthreshold 1 Hz rTMS on SP duration (Fitzgerald et al., 2002; Modugno et
al., 2003; Romeo et al., 2000). What made this finding rather surprising was the fact that
a 10 min dose of subthreshold 1 Hz rTMS did not significantly affect the resting MEP
amplitude, but significantly lengthened SP. This finding supports previous observations
that imply that rTMS may produce different effects on MEPs and SPs. MEP amplitude is
a global measure of motor corticospinal excitability, while SP is predominantly mediated
by GABA-Bergic inhibitory mechanisms. Our current findings, together with previous
literature suggest that different doses of 1 Hz rTMS (intensity and duration) may
distinctly affect activity within multiple excitatory and inhibitory circuits of the motor
cortex. Suprathreshold 1 Hz rTMS and subthreshold 1 Hz rTMS for longer durations (20
min) may lead to reduction in the global excitability without significantly affecting the
GABA-Bergic inhibition. In contrast, it is likely that subthreshold 1Hz rTMS for 10 min
increase GABA-Bergic inhibition, but may fail to modulate the global corticospinal
excitability. Further research is warranted to precisely investigate the nature of the
change within specific excitatory and inhibitory circuits with rTMS at different doses.
In summary, the current study aimed to identify an optimal 1 Hz dose (intensity-duration
combination) to down regulate motor cortical excitability. We demonstrated that down
59
regulation of corticospinal excitability with 1Hz rTMS is dose-dependent with supra-
threshold rTMS being more effective, even with smaller duration to achieve the same
effect as with longer duration of stimulation. Further, our study also suggests that
different doses of 1Hz rTMS may differentially affect the excitatory and inhibitory
circuits within the motor cortex. This information has significant implications of the
potential use of 1Hz rTMS for research and therapy. Future work is needed to investigate
the effects of different rTMS doses on specific excitatory and inhibitory neural circuits
within the motor cortex. Information obtained from these studies will be critical for
effective use of rTMS in research and clinical application.
60
CHAPTER FIVE
NEURAL SUBSTRATES OF MOTOR MEMORY CONSOLIDATION
Introduction
Motor skill acquisition has been characterized by at least two distinct phases: a
fast within-training phase leading to performance improvement during practice, and a
delayed, latent phase that occurs after the practice session (Buitrago, Ringer, Schulz,
Dichgans, & Luft, 2004; Karni et al., 1998; Shadmehr & Brashers-Krug, 1997). Motor
memory consolidation occurs in this post-acquisition phase and is thought to result in a
more stable motor memory over time (Krakauer & Shadmehr, 2006; Robertson et al.,
2004). Interference during the immediate post-acquisition period often degrades retention
of motor skills, the implication being that consolidation is critical for motor learning
(Luft, Buitrago, Ringer et al., 2004; Muellbacher et al., 2002; Shadmehr & Brashers-
Krug, 1997). Recent evidence suggests that time-dependent consolidation processes
actively engage a fronto-parietal network during the post-acquisition period (Albert N.B.,
Robertson E.M., & R.C., 2009).
The structure of motor practice has a strong influence on motor skill learning
(Schmidt & Bjork, 1992; Sherwood & Lee, 2003). For example, repetitive motor skill
practice of the same task, known as constant practice, results in degraded motor
performance during retention testing compared to variable practice where practice trials
of a task are randomly interleaved with trials of other motor tasks. Consistently,
61
behavioral motor learning studies demonstrate that variable practice enhances long-term
retention of a motor skill, compared to constant practice, although it may not show any
performance benefits during practice or at the end of practice (Sekiya, Magill, &
Anderson, 1996; C. H. Shea & Kohl, 1990, 1991; C. H. Shea, Lai, Wright, Immink, &
Black, 2001; Wymbs & Grafton, 2009). Although variable and constant practice lead to
different levels of retention, little is known about how differences between the two
practice structures specifically affect post-practice consolidation processes that lead to
enduring motor memories for long-term retention.
The neural substrates of motor skill acquisition have been investigated using
functional imaging. Recently, it has been shown that during practice of motor skills,
distinct neural substrates are engaged that differ depending on the practice structure (N.
R. Cohen et al., 2009; Cross et al., 2007; Lin et al., 2008; Lin et al., 2009; Wymbs &
Grafton, 2009). Cross and colleagues (2007) demonstrated that during variable practice,
BOLD signal activation of the motor cortex, pre-motor, and prefrontal areas was
increased, suggesting greater cognitive processing in these cortical areas compared to
constant practice conditions. It is not known how differences in practice structure
influence neural network activity post-practice during the consolidation phase. Given that
variable practice leads to better retention than constant practice, we hypothesized that
differences in the practice structure may drive subsequent offline activity within different
brain networks critical to motor memory consolidation.
62
To test this hypothesis, we used 1 Hz repetitive transcranial magnetic stimulation (rTMS)
immediately following constant or variable motor practice to directly interfere with post-
practice processes in primary motor cortex (M1) or dorsolateral prefrontal cortex
(DLPFC). Both regions have previously been shown to be engaged in early motor
memory consolidation. The effect of rTMS interference on motor learning was assessed
through motor performance in a retention test administered 1 day after practice. Here, we
demonstrate that practice structure during motor skill acquisition does affect subsequent
offline activity within different neural networks that mediate motor memory
consolidation.
Material and Methods:
Participants: Fifty-nine volunteers (25 men, 34 women; mean age 26.12 years;
range: 23-34 years), naïve to the purpose of the experiments, participated. All participants
were free of any history of neurological disorders, and met the safety criteria for
repetitive Transcranial Magnetic Stimulation (rTMS) and Magnetic Resonance Imaging
(MRI). A written informed consent, approved by the Institutional Review board of the
University of Southern California was obtained prior to participation.
The participants were randomized in to six experimental groups: Constant
practice no-rTMS group (CP), Variable practice no-rTMS group (VP), Constant practice
M1 Interference group (CP-M1), Variable practice M1 interference group (VP-M1),
Constant practice DLPFC interference group (CP-DLP) and Variable practice DLPFC
63
interference group (VP-DLP). The M1 interference groups (CP-M1 and VP-M1) received
1 Hz rTMS over the biceps representation of M1 during the post-practice consolidation
phase. The DLPFC interference groups (CP-DLP and VP-DLP) received 1Hz rTMS over
the DLPFC similarly, immediately post-practice during the consolidation phase (Figure
7).
Motor task:
Participants practiced moving a light-weight lever with their dominant arm to
replicate a target trajectory (a position-time trace) presented on a computer monitor. This
lever was affixed to a frictionless vertical axle restricting movement of the lever to the
horizontal plane above the surface of a table. The handle at the end of the lever was
adjusted to accommodate the length of the participant‘s forearm. A linear potentiometer,
attached to the base of the vertical axle recorded lever-position information with
movements of the lever away from the body reflecting upward movements on the
64
computer monitor and movements toward the body reflecting downward movements on
the computer monitor. Signals from the potentiometer were converted to digital signal by
an A/D board of a Compact 466v computer and sampled at 1000Hz to provide FB on the
computer monitor. The template software program (A.Weekly, 2004) was used for
manipulation of the movement trajectory and the interval duration, and data storage for
off-line analysis of each trial.
Figure 6:
A) Participants practiced an arm movement
task aimed to match a target presented on the
screen. There were 4 targets: A1, A2, A3 and
A4, each with different absolute amplitude
specifications (30, 45, 60 and 75 degrees,
respectively), but a similar movement
structure and same absolute time
requirements (800ms). The constant practice
groups practiced 120 trials of target A3, while
the variable practice group practiced 60 trials
of target A3 and 20 trials of targets A1, A2
and A4 each. The order of presentation of the
four targets was randomized in the variable
practice groups.
(B) Example of the feedback display. Participant movement trajectory (thin line)
superimposed on the target trajectory (thick line). The root mean square error (RMSE)
was displayed along with the trajectories after each trial.
There were 4 target trajectories (position-time traces): A1, A2, A3 and A4 (Figure 6a),
each having different peak amplitude specifications (30, 45, 60 and 75 degrees)
respectively, but with a similar movement structure and temporal duration requirements
(800ms). The coordinated arm movement skill was to replicate a target trajectory that
A
B
65
consisted of two elbow extension-flexion reversal movements, each of specific
amplitude, performed in the horizontal plane. This was a fast movement with a total
duration of 800 ms. One of the 4 target trajectories was displayed on the computer
monitor at the beginning of each trial for a 2000 ms duration, after which the trajectory
disappeared from the screen. After 1000 ms, a ―go‖ signal was displayed at which point,
the subject was instructed to move the lever in a manner to replicate the target trajectory
as closely as possible. After a 2000 ms delay following movement, post-response
feedback was displayed on the computer screen for a 5000 ms duration. Feedback
consisted of: 1) an overall numeric error score (root mean square error, RMSE) and 2) a
graphic representation of the participant‘s response, time-locked to onset and
superimposed on the target movement pattern. Figure 6b shows an example of an
individual trial and post-response FB display with RMSE and superimposed trajectory.
66
Experimental Design:
Figure 7: Experimental design: Participants practiced the task on Day 1 either under a
constant practice condition or variable practice condition. Immediately following
practice, they were tested for end of acquisition (EoA) performance and immediate
transfer. 1day later, participants were re-tested on a retention/delayed transfer test (RT) to
infer learning of target A3 (Criterion task). Participants from each practice condition
(constant and variable) were randomized to a control-no-rTMS group (CP, VP), a M1-
interference group (CP-M1, VP-M1), and a DLPFC-interference group (CP-DLP, VP-
DLP). The M1-interference groups received 1Hz rTMS over M1 and the DLPFC-
interference groups received 1 Hz rTMS over DLPFC, immediately after EoA. Black dots
represent the procedures for measuring motor corticospinal excitability.
The experimental design is illustrated in Figure 7. The experiment was conducted on two
consecutive days. On day 1, prior to randomization, motor cortex excitability was
assessed in all participants using TMS. Ten minutes following this procedure, each
participant practiced the motor task for a total of 120 trials during the acquisition phase.
The structure of the practice condition was different for the constant practice and variable
67
practice groups. The constant practice groups (CP, CP-M1 and CP-DLP) practiced 120
trials of target A3 (criterion target). In the variable practice groups (VP, VP-M1 and VP-
DLP), the participants practiced target A3 for 60 trials and 20 additional trials of targets
A1, A2 and A4 each. The presentation order of the four targets was pseudo-randomized.
During practice, the participants received post-response feedback about their performance
after every trial. Immediately following practice, they performed 4-trials of target A3
without post-response feedback as end of acquisition (EoA) performance. In addition,
they were also tested on an immediate transfer (IT) test which comprised of consisted of
an 8-trial block where participants were required to replicate the targets that were not
presented during the acquisition phase (4 trials each of 50 and 80 degree amplitudes). On
Day 2 (24+3 hours later), participants were re-tested on a 4-trial no-feedback retention
test (R) to measure learning of the criterion target A3. This was followed by a delayed
transfer (DT) test, which was similar to immediate transfer (IT) test in its composition.
Following practice and EoA testing on Day 1, 1 Hz rTMS was applied over the
representation of biceps brachi muscle in the contralateral M1 for subjects in the CP-M1
and VP-M1 groups. 1 Hz rTMS transiently down regulates local cortical function and this
effect has been employed to interfere with cortical function beneath the stimulation site.
In the CP-DLP and VP-DLP groups, 1Hz rTMS was applied over the contralateral
DLPFC to interfere with its processing immediately post-acquisition. Methods used to
localize MI and DLPRC are described below. Retention and transfer test performance on
68
the next day was used to determine the effect of rTMS interference on motor skill
learning (Figure 7). Contrasting learning for the M1 and DLPFC interference groups with
the control groups allowed us to assess the role of M1 and DLPFC in early motor
memory consolidation following each of the two practice conditions.
Assessment of Motor cortex (M1) excitability: TMS was used to assess M1 excitability.
Stimulation was delivered with a 70 mm figure of eight coil attached to Magstim
Rapid2 magnetic stimulator. Single pulse TMS was used to identify the ‗hot-spot‘ which
is the optimal scalp position for consistently eliciting the largest motor evoked potential
(MEP) from the biceps brachii muscle. The coil was held tangentially to the scalp
contralateral to the dominant arm with the coil-handle pointing posteriorly away from the
midline at an angle of 45 (Brasil-Neto, McShane, Fuhr, Hallett, & Cohen, 1992; Mills,
Boniface, & Schubert, 1992). Next, resting motor threshold (MT) was determined by
systematically decreasing the stimulus intensity over the hotspot. MT is defined as the
lowest intensity level required to induce MEP peak-to-peak amplitude of at least 50 V,
in 5 out of 10 consecutive trials (Maeda et al., 2000b). Motor cortical excitability was
assessed by applying 10 TMS pulses at 120%MT intensity over the hot-spot.
Practice procedure: During practice, participants sat comfortably in front of the computer
monitor with their dominant forearm along the arm of the lever and their hand grasping
the lever handle. A sample trajectory was used to orient the participant to the task.
Sample goal movement and feedback were explained carefully to each participant. The
69
experimenter and participants reviewed templates of a sample target trajectory and
superimposed feedback trajectory to ensure maximal understanding of the computer
displayed feedback. The participants were instructed that during practice, they were to
practice to replicate the goal trajectory and make their movements as accurate as possible
(i.e. lower RMSE). When the experimenter determined that the participant was
adequately oriented to the task and feedback, practice was begun. Immediately following
practice, each participant was tested for the end of acquisition performance (EoA) and
immediate transfer (IT) test. One day later (24 +3 hours), the participants returned for a
retention test (R) and a delayed transfer (DT) test. The EoA and 1 day retention test (R)
consisted of a no-feedback, 4-trial block of the criterion target trajectory (A3). Retention
test was used to determine the participant‘s recall and reflects the strength of the motor
skill memory representation developed during practice. The transfer tests (IT and DT)
consisted of an 8-trial block where participants were required to replicate the targets that
were not presented during the acquisition phase (50 and 80 degree amplitudes). Transfer
tests assess the generalizability of what has been learned.
rTMS procedure: 1 Hz rTMS was applied with Magstim 70 mm figure of eight
coil attached to a Magstim Rapid² magnetic stimulator. Immediately after EoA, subjects
in the CP-M1 and VP-M1 groups received a 1 Hz train of rTMS for 10 min (600 pulses)
at 110% MT intensity over the hotspot of biceps brachii in contralateral M1. Immediately
before and after rTMS over M1, we measured motor cortical excitability to examine
70
rTMS-induced changes in motor cortical excitability. In the CP-DLP and VP-DLP
groups, 1Hz rTMS at 110% MT intensity for 10 min (600 pulses) was applied over the
contralateral DLPFC. A structural MRI-based, stereotaxic neuronavigation system,
Brainsight Frameless (Rogue Research), was used to precisely localize the TMS coil over
the DLPFC. First, a three dimensional (3D) image of the cerebral cortex was
reconstructed in Brainsight by processing a two-dimensional MR image. This 3D image
allowed identification of the sulci and gyri on the individual brain surface. Markers were
placed on the MR image at specific anatomical landmarks: the tip and base of nose, right
and left tragus of the ear. Then optical markers were attached to the TMS coil which was
calibrated and registered to the Brainsight system by means of a Polaris optical position
sensor and a coil-tracker device. Next, the participant put on tracker glasses with
reflective markers. The participant‘s anatomical landmarks were then coregistered with
the anatomical landmarks on the MR image. Coregistration entails an optimal
transformation between actual skin points on the participant and the skin surface of the
MRI reconstructed human model. This allowed a real time display of the relative
positions of the coil and the participant‘s head and brain surface, which was critical in
guiding the placement of the coil over the DLPFC. Stimulation over DLPFC was directed
at the middle third of the middle frontal gyrus corresponding with the region of
Broadmann area 46, and posterior part of area 9 (Farzan et al., 2009).
71
Data Analysis:
Motor behavior: Performance accuracy was assessed for practice, end of
acquisition (EoA), retention test (R), and transfer tests (IT and DT). Dependent measure
for accuracy included RMSE which is the average difference between the goal movement
trajectory and the participant‘s response, calculated over the participant‘s total movement
time (Schmidt & Lee, 2004). RMSE was calculated for each trial. For the acquisition
phase, only the trials on criterion task (A3) from the variable practice group, and
corresponding trials in the constant practice group were included in behavioral analysis.
Thus for each participant, we included the performance on the sixty trials of A3 for the
acquisition phase analysis. For each participant, RMSE for these sixty trials were
averaged into twelve 5-trial blocks for the acquisition phase. For the EoA and 1 Day
retention test, an average RMSE was computed over each set of 4 criterion task trials. For
each of the transfer tests (IT and DT), average RMSE for 4 trials was calculated
separately for 50˚ amplitude and 80˚ amplitude targets.
For motor skill practice data during the acquisition phase, a 2 practice condition
(constant, variable) X 3 rTMS site (no rTMS, M1-rTMS and DLP-rTMS) X 12 Block (1-
12 acquisition blocks) ANOVA with repeated measures on the last factor was used. 2
practice (constant, variable) X 3 rTMS site (no rTMS, M1-rTMS and DLP-rTMS) X 2
test (EoA, R) Repeated measures ANOVA with repeated measures on the test was used
to characterize the between-group differences in offline memory stabilization. Separate 2
72
practice condition (constant, variable) X 3 rTMS site (no rTMS, M1-rTMS and DLP-
rTMS) ANOVA was used for each, EoA and 24-hour retention test performance. For the
transfer tests, 2 practice (constant, variable) X 3 rTMS site (no rTMS, M1-rTMS and
DLP-rTMS) X 2 transfer test (IT, DT) Repeated measures ANOVA with repeated
measures on the last factor was used separately for 50˚and 80˚ target block each. Also,
separate 2 practice condition (constant, variable) X 3 rTMS site (no rTMS, M1-rTMS and
DLP-rTMS) ANOVA was used for each, IT and DT performance of 50˚ and 80˚ trial
blocks. For all statistical tests, the significance
level was set at P<.05. SPSS version 15.0
statistical software
was used for all statistical analyses.
TMS data: MEPs were analyzed offline with a customized MATLAB
(Mathworks, MA, USA) software tool DataWizard (version 0.8.5, Dr. Allan D. Wu,
UCLA). Peak-to-peak amplitude was computed for each recorded MEP. For the M1-
interference groups (CP-M1, VP-M1), mean MEP amplitude of 10 trials was calculated
for baseline, pre- and post-rTMS. A 2 practice (Constant, variable) X 3 time points
(baseline, pre- and post rTMS) Repeated measures ANOVA with repeated measures on
time points was used to compare the effects of practice and rTMS on motor corticospinal
excitability.
73
Results:
Motor cortical excitability:
We assessed the effect of practice conditions and 1Hz rTMS over M1 on motor
corticospinal excitability by examining the change in the MEP amplitude at three time
points: baseline, pre- and post- rTMS in CP-M1 and VP-M1 groups. There was no
significant difference between the two groups across the three time points (2X3 Repeated
measures ANOVA, F
1, 18
= 0.152,p = 0.702; Figure 8). Further, 1Hz rTMS at 110%MT
applied over M1 for 10 min reduced motor corticospinal excitability as evidenced by a
significant reduction in MEP amplitude post rTMS (Mean MEP amplitude V + SEM:
210.79 + 28.69) compared to pre-rTMS (Mean MEP amplitude V + SEM: 480.99 +
80.55) (Paired sample t test, t (1,19) = 4.377; p< 0.05; Figure 8). This finding confirmed
that 1Hz rTMS was a valid tool to down regulate motor corticospinal excitability and
induced a ―virtual lesion‖ that could be used to
interfere with cortical processing during early
motor memory consolidation.
Figure 8: Mean MEP amplitude evoked in the
Biceps brachii at three time points: baseline,
post-practice and post rTMS in the M1
interference groups (VP-M1 and CP-M1). There was no difference between the two
groups at any time point. Post-rTMS MEP amplitude was significantly lower than post-
practice MEP amplitude in both groups, suggesting that rTMS downregulated motor
corticospinal excitability in both the groups.
74
Motor skill performance during practice and End of Acquisition (EoA)
Participants in all groups improved across the acquisition phase as evidenced by a
reduction in the global error (root mean square error, RMSE) in performing the target
matching task (practice effect; ANOVA, F
1, 58
= 351.66, p< 0.001). However, there was
no significant difference between the groups across the entire acquisition phase (Figure 9,
ANOVA, F
2, 53
= 0.458, p=0.635). To confirm that the experimental groups were not
different at the beginning of acquisition, we used a separate 2 (practice) X 3 (TMS site)
ANOVA on the first practice block. At the first practice block, there was no significant
difference in the skill performance on the criterion task (A3) between the six
experimental groups (Figure 9, ANOVA; F
(5, 58)
= 1.324; p= 0.268). Additionally, end of
acquisition (EoA) performance on the no-feedback test was not significantly different
between the experimental groups (Figure 9, ANOVA, F
2, 58
= 0.463, p=0.632).
75
Figure 9: Practice and End of Acquisition (EoA) performance: Performance of the
participants in the control group (open circles), M1 interference group (black filled
circles), and DLPFC interference group (grey filled circles) during the practice phase.
The left side of the figure represents the performance of the participants who practiced
under variable practice structure; the right side of the figure represents performance of
the participants who practiced under constant practice structure. Each data point for the
practice block represents a mean RMSE of 5 trials on the criterion task (A3). Each data
point for the EoA represents a mean RMSE of the 4-trial EoA test of A3. The error bars
represent the SEM.
Motor Memory consolidation and Motor skill learning:
Offline memory stabilization was determined by measuring the performance
change from the EoA to the retention for each group. 2 (practice structure) X 3 (rTMS
site) X 2 test (EoA, R) ANOVA with repeated measures on the last factor yielded a
significant interaction (ANOVA, F
2, 53
= 5.419, p=0.007), suggesting that the effect of
practice structure on offline memory stabilization depended on the site of rTMS
76
stimulation (Figure 10). In those participants who practiced under constant conditions,
rTMS over M1 attenuated the offline stabilization of motor memory; while in those who
practiced under variable practice structure, rTMS over DLPFC attenuated offline
stabilization of motor memory. Figure 11 shows the change in the performance accuracy
from End of Acquisition (EoA) to Retention (R) in individual subjects in all 6
experimental groups.
Figure 10: End of
acquisition (EoA) and
retention performance of
the participants in the
control group (open
circles), M1 interference
group (black filled circles),
and DLPFC interference
group (grey filled circles).
Each data point represents a
mean of four trials of target
A3; Error bars represent the
SEM.
77
Figure 11: Data showing change in the performance accuracy from End of Acquisition
(EoA) to Retention (R) in individual subjects in all 6 experimental groups.
78
Performance on the next day retention test was used as an indicator of skill learning. At
the retention test, performance differed significantly between groups (ANOVA, F
2, 58
=
6.44, p=0.003). A significant effect of practice condition (ANOVA, F
1,58
=13.51,
p=0.001) indicated that participants who practiced under variable conditions retained the
criterion skill better (i.e. lower error) than those who practiced under constant conditions.
However, the beneficial effect of variable practice depended on rTMS condition. In other
words, post-practice rTMS over M1 or DLPFC affected retention differently for the two
practice conditions (Figure 12; 2 practice condition X 3TMS site ANOVA, F
2, 58
= 6.44,
p=0.003). In those participants who practiced under constant practice conditions, post-
practice rTMS interference over DLPFC did not affect learning compared to the no-rTMS
control CP group (Figure 12; RMSE: CP-DLP = CP; ANOVA, F
1, 18
= 0.011, p=0.918). In
contrast, when rTMS was applied over M1 following constant practice, learning was
significantly attenuated (Figure 12; RMSE: CP- M1> CP; ANOVA, F
1, 19
= 4.681,
=0.044). The opposite pattern emerged for participants who trained under a variable
practice condition. Post-practice rTMS over M1 following variable practice did not
significantly affect learning compared to the no-rTMS control VP group (Figure 12;
RMSE: VP-M1 rTMS= VP, ANOVA, F
1, 20
= 0.024, p=0.879). In contrast, when rTMS
was applied over DLPFC following variable practice, learning was significantly
attenuated (Figure 12; RMSE: VP- DLP rTMS > VP; ANOVA, F
1, 20
= 5.425, p=0.014).
79
Thus, a double-dissociation was observed 1 day after practice between practice condition
structure (variable vs. constant) and the neural locus of the rTMS interference.
Figure 12: Retention test
performance (Motor
learning)
Interaction between
practice structure
(constant practice-striped
bars, variable practice-
black filled bars) and
rTMS site (no-TMS, M1
and DLPFC) at retention.
rTMS interference to M1,
but not DLPFC
immediately after constant
practice attenuated
retention of the motor skill
compared to control. In
contrast, immediately after
variable practice, rTMS to
DLPFC, but not M1 attenuated retention of the motor skill compared to control. Error
bars represent the SEM. Note that for the control-no-rTMS groups, variable practice
benefits motor learning more than constant practice.
Temporal specificity of rTMS effects:
One possibility is that the differential effects of post-training rTMS over M1 or
DLPFC are due to some residual or lingering effects of rTMS effects on later recall of the
skill, rather than on the consolidation process itself. This is unlikely since the transient
effects of rTMS are known to be operational for up to 15-30 min post stimulation.
However, to rule out the aforementioned possibility, we recruited 12 additional
80
participants and randomized them into one of two groups which received delayed rTMS
following practice: CP M1-4 hr and VP DLP-4 hr. CP M1-4 hr group practiced the skill
under constant conditions and had 1Hz rTMS applied over M1 4 hours after practice. The
VP DLP-4hr group practiced the task under variable conditions and rTMS was applied
over DLPFC 4 hours after practice. We observed that retention performance of the CP
M1-4hr group was similar to the CP-no-rTMS group (Figure 13 right panel; ANOVA, F
1,
15
= 0.739, p=0.404). Similarly, the VP DLP-4hr group retained the skill as well as the
VP-no-rTMS group (Figure 13 left panel; ANOVA, F
1, 16
= 0.879, p=0.363). The failure
of the delayed rTMS to interfere with retention of the motor skill in these groups
demonstrates that the practiced skill had become resistant to the interference after 4
hours. This finding provides strong evidence in support of the conclusion that the rTMS
interference effects on motor learning observed for the CP-M1 and VP-DLP groups were
temporally specific to the immediate post-practice consolidation phase.
81
Figure 13: Temporal Specificity of rTMS effects
End of acquisition (EoA) and retention performance of the participants in the control
groups (open circles), VP-DLP group (grey circle), VP-DLP 4hr group (grey filled
square), CP- M1 group (black filled circle), and CP-M1 4 hr group (black filled square).
Each point represents a mean of 4 trials of EoA and retention test. The error bars
represent SEM.
82
Transfer of motor learning:
Transfer to 50º target:
Offline memory stabilization was determined by measuring the performance
change from the immediate transfer (IT) to the delayed transfer for each group. A 2
practice(constant, variable) X 3 rTMS site (M1, DLPFC, and M1) X 2 test (IT, DT)
ANOVA with repeated measure on the last factor for transfer to the 50º target yielded a
significant main effect of practice (F1,53= 7.201; p=0.01, Figure 14). Similar to previous
findings, variable practice structure benefited offline memory stabilization for transfer
compared to constant practice structure. There were no significant interaction effects or
effect of rTMS site. Further analysis revealed that the participants in the variable practice
groups performed with significantly lower error on the delayed retention test compared to
those in the constant practice groups (ANOVA, F
1,58
= 6.351; p= 0.015). Figure 15 change
in the performance accuracy of individual subjects for the 50˚ Target from Immediate
Transfer (IT) to Delayed Transfer (DT) in all 6 experimental groups.
83
Figure 14: Immediate and delayed transfer performance for the 50 degree target of the
participants in the control group (open squares), M1 interference group (filled squares),
and DLPFC interference group (filled diamonds). The left panel represents the data of the
variable practice groups and the right panel shows the data of the constant practice
groups. Each data point represents a mean of four trials of target A3; Error bars represent
the SEM.
84
Figure 15: Data showing change in the performance accuracy of individual subjects for
the 50˚ Target from Immediate Transfer (IT) to Delayed Transfer (DT) in all 6
experimental groups.
85
Transfer to 80 º target:
A comparison of the changed performance on 80 º target from the IT to DT across
all groups indicated how rTMS interference to the neural substrates following constant
and variable practice affected the offline stabilization. Since there was a significant effect
of practice structure at the IT for 80º target (2X3ANOVA, F1, 58= 20.462; p< 0.001), we
used the performance at the IT as a covariate for 2 practice(constant, variable) X 3 rTMS
site (M1, DLPFC, and M1) X 2 test (IT, DT) ANOVA with repeated measure on the last
factor. A significant main effect of practice (F1, 52= 8.22; p= 0.006) indicated that
offline stabilization was enhanced by variable practice compared to constant practice. A
significant interaction of rTMS site and practice structure (2X3X2 RM ANOVA, F2, 52
= 6.371; p=0.003) suggested that the post-practice rTMS over M1 or DLPFC affected
offline stabilization differently for the two practice conditions. rTMS over the DLPFC,
but not over M1 attenuated offline stabilization only when the practice structure was
variable. In contrast, rTMS over M1, but not over DLPFC attenuated offline stabilization
only when the practice structure was constant. Further analysis (2 practice structure X 3
rTMS site ANOVA) on the DT for 80º target yielded a significant interaction between the
rTMS site and practice structure. In those participants who practiced under constant
practice conditions, post-practice rTMS interference over DLPFC did not affect transfer
to 80º target compared to the no-rTMS control CP group (Figure 16; RMSE: CP-DLP =
CP; ANOVA, F
1, 18
= 0.008, p=0.930). In contrast, when rTMS was applied over M1
following constant practice, delayed transfer to 80 º was significantly attenuated (Figure
86
16; RMSE: CP- M1> CP; ANOVA, F
1, 19
= 4.495, p=0.048). In contrast, post-practice
rTMS over M1 following variable practice did not significantly affect delayed transfer
performance on 80 º target compared to the no-rTMS control VP group (Figure 16;
RMSE: VP-M1 rTMS= VP, ANOVA, F
1, 20
= 0.015, p=0.904). In contrast, when rTMS
was applied over DLPFC following variable practice, delayed transfer performance was
significantly attenuated (Figure 16; RMSE: VP- DLP rTMS > VP; ANOVA, F
1, 20
=
4.487, p=0.048). Thus, similar to retention, a double-dissociation was observed 1 day
after practice between practice condition structure (variable vs. constant) and the neural
locus of the rTMS interference for transfer performance on 80 º target. Figure 17 shows
change in the performance accuracy of individual subjects for the 80˚ Target from
Immediate Transfer (IT) to Delayed Transfer (DT) in all 6 experimental groups.
87
Figure 16: Immediate (IT) and delayed (DT) transfer performance for the 80 degree
target of the participants in the control group (open squares), M1 interference group
(filled squares), and DLPFC interference group (filled diamonds). The left panel
represents the data of the variable practice groups and the right panel shows the data of
the constant practice groups. Each data point represents a mean of four trials of target A3;
Error bars represent the SEM.
88
Figure 17: Data showing change in the performance accuracy of individual subjects for
the 80˚ Target from Immediate Transfer (IT) to Delayed Transfer (DT) in all 6
experimental groups.
89
Discussion
The double dissociation between rTMS site of neural stimulation and practice
structure on motor learning suggests that engagement of specific neural networks for
motor memory consolidation depend on practice structure. When practice structure is
constant, rTMS applied over M1 but not DLPFC interferes with motor skill learning. In
contrast, when practice structure is variable, rTMS applied over DLPFC, but not M1
attenuates motor skill learning.
Role of primary motor cortex in motor memory consolidation
Immediately following constant task practice, rTMS applied over M1 impaired
motor skill performance at the 24-hour retention test. This rTMS-induced interference
effect following constant practice was temporally and spatially specific to the primary
motor cortex. In contrast, rTMS applied over DLPFC immediately after constant practice
did not attenuate the retention of the motor skill. Similarly, rTMS applied over M1 4
hours post-practice had no interference effect.
One concern may be that the effects of rTMS are possibly not restricted to M1.
Functional imaging studies have demonstrated that, due to M1‘s neuronal connections, a
distributed network of cortical and subcortical areas is activated by rTMS over M1 (Paus
et al., 1998; Van Der Werf & Paus, 2006a). This possibility cannot be ruled out in our
study. Therefore, we suggest that M1 is a critical part of a putative neural network that
mediates consolidation processes during the immediate post-constant practice period.
90
M1 is one of the important neural structures involved in formation and maintenance of
motor memory. Practice of a motor skill leads to reorganization within M1 suggesting
that the resultant ―motor map‖ may represent the neural substrate of motor memory
(Monfils et al, 2005). Animal and human studies demonstrate that perturbation to M1
processing post-practice affects retention of a newly practiced motor skill, indicating its
crucial role in motor memory consolidation. Here we extend our understanding of M1
function by demonstrating that its role during immediate consolidation is modulated by
the structure of practice. M1 processing is critical during memory consolidation
following constant practice, and this consolidation involving M1 contributes to skill
retention.
Our results may also help explain some apparent conflicting results in the
literature regarding the role of M1 in motor memory consolidation. While many studies
have demonstrated that rTMS interference to M1 processing immediately post-practice
attenuates learning (Muellbacher et al., 2002; Robertson et al., 2005), some studies have
shown no effect (Baraduc et al., 2004; Shemmell, Riek, Tresilian, & Carson, 2007). In
light of our current findings, additional evidence found here may provide a different
perspective of these apparently contradictory effects. Indeed, M1 is a critical neural
substrate that implements motor memory consolidation following constant/ blocked
practice (Muellbacher et al., 2002; Robertson et al., 2005). However, when two or more
tasks were practiced using a variable/interleaved structure, rTMS over M1 post-practice
91
failed to interfere with consolidation of the practiced tasks (Baraduc et al., 2004;
Shemmell et al., 2007). Our findings suggest that variable/interleaved practice may
engage other neural substrates during consolidation more than MI such as the DLPFC.
DLPFC in motor memory consolidation following variable practice structure
In the present study, rTMS over DLPFC immediately following variable but not
constant practice impaired motor skill retention; however, interference to motor skill
learning was not evident when rTMS was applied over DLPFC 4-6 hours later. Similarly,
when rTMS was applied over M1 immediately following variable practice, there was no
effect on motor learning. This indicates that the interference effect of DLPFC stimulation
after variable practice was spatially and temporally specific. Studies using functional
imaging and rTMS together have demonstrated stimulation-dependent increases in rCBF
in distant cortical and subcortical areas with prefrontal rTMS (Ohnishi et al., 2004; Paus,
Castro-Alamancos, & Petrides, 2001). Consequently, when rTMS is applied over DLPFC
as it was in our study, a perturbation effect likely resulted in a circuit of brain areas
including DLPFC. Our results indicate that this circuit is critical for post-practice
consolidation processes that mediate the retention of motor skills acquired under variable
practice structure.
DLPFC is actively engaged in cognitive mechanisms that implement long-term
memory (Blumenfeld & Ranganath, 2006, 2007). Shadmehr and Holcomb (1996)
reported a significant increase in PET regional cerebral blood flow within DLPFC (BA
92
46) with practice. When participants were tested 6 hours after practice, this activation had
decreased significantly compared to its immediate post-practice level despite no apparent
change in performance. This time-dependent modulation of DLPFC activity during the
post-practice period supports a role in motor memory consolidation. Recent
investigations using TMS to directly modulate DLPFC function have added further
support to its role in consolidation (Galea, Albert, Ditye, & Miall, 2009). In the present
study, we directly perturbed DLPFC processing using 1Hz rTMS applied immediately
post-practice and demonstrated its critical role in motor memory consolidation when
practice was conducted under a variable, but not constant practice structure. To our
knowledge, this is the first study to directly perturb DLPFC immediately post-practice
and demonstrate its role in motor memory consolidation during wake period.
Practice structure and neural circuits in memory consolidation
There is evidence to indicate that practice structure influences motor learning. As
is often evident in the literature, our data from the control groups (VP and CP)
demonstrate that variable practice enhances long-term retention of the skill compared to
constant practice even when there are no obvious performance differences during, or at
the end of the practice phase. This delayed emergence of practice structure benefits is an
example of the learning-performance distinction, a well-known phenomenon in cognitive
neuroscience and behavioral motor learning (Cahill et al., 2001). Our findings further
offer an insight in to some mechanisms that may underlie this learning-performance
93
distinction. Our control groups (VP and CP) data suggest that variable practice may lead
to better retention (learning), in part, through enhanced post-practice processes resulting
in better motor memory stabilization.
Our results add to the body of evidence that consolidation processes are critical
for long-term retention of motor skills. We observed that immediately following variable
practice, rTMS interference over DLPFC, but not M1, attenuated retention of the motor
skill. DLPFC is associated with cognitive aspects of motor control and learning. Animal
and human imaging studies suggest that the DLPFC is involved in mechanisms of
attention, processing of working memory (J. M. Fuster, 2000), representation of goal-
related spatial information (Genovesio, Brasted, & Wise, 2006), selection and planning of
upcoming actions (Pochon et al., 2001; Rowe, Toni, Josephs, Frackowiak, & Passingham,
2000) as well as the formation of long-term memory (Blumenfeld & Ranganath, 2006,
2007). In our study, participants in the variable practice group were required to select and
execute a different movement plan at each trial as prompted by the target display. This
variable practice structure may require a relatively higher engagement of the DLPFC
compared to M1. Indeed, an increased activation in the prefrontal regions (specifically
the superior and middle frontal gyrus) was reported by Cross and colleagues (2008)
during variable, but not constant practice conditions. Our findings further extend the role
of DLPFC to the post-practice consolidation phase. It is likely that after variable practice,
memory consolidation is mediated to a greater extent by DLPFC than by MI.
94
Constant practice requires the learner to move repeatedly to only one target across
practice. Therefore, unlike variable practice, the learner may not be forced to rely as
much on working memory, action selection, and planning that garner the unique
processing capabilities of DLPFC. Instead, repeated practice on a single task may result
in memory formation within M1 and this may be a critical node for motor memory
consolidation after constant practice. Our findings of a double dissociation for the role of
M1 and DLPFC in motor memory consolidation demonstrate that different neural
networks make different post-training contributions to subsequent retention depending on
practice structure. Recent functional imaging evidence suggests that motor skill practice
drives offline activity in fronto-parietal networks during the immediate post-practice
period. Our findings support this evidence and suggest that the structure of practice can
differentially modulate the role of specific nodes (M1 and DLPFC) within this network
that are important for motor skill learning.
What could be the mechanisms underlying these observed differences in the
neural substrates of motor memory consolidation with differing practice structures? One
possibility is that each practice structure predominantly facilitates memory of a specific
component of the motor skill that has a distinct neural substrate for consolidation.
Memory for motor skill has been thought to have two main components: one component
that represents the spatial goal of the skill, and the other represents the movements
needed to achieve that goal (D. A. Cohen, Pascual-Leone, Press, & Robertson, 2005;
95
Robertson, 2009). It is likely that variable practice may give rise to preponderance of
goal-component of the motor skill. This is in accordance with the schema theory of motor
learning, which suggests that with higher variability in practice structure, the learner
learns an abstract relationship or ―schema‖ between the goal and action parameters
(Schmidt, 1975). This ―schema‖ is likely to emerge when the learner practices a
particular skill version in the context of other versions. In contrast, repetitive constant
practice may predominantly give rise to a movement-component of the motor skill. The
goal-component of a motor skill is known to be represented within a neural circuit that
includes the DLPFC, while movement-component of motor memory is primarily encoded
within a circuit that included M1 (Robertson, 2009). Further, evidence indicates that an
embedded contextual element within task practice typical of variable structure practice
may promote a goal-based motor memory representation (Spencer, Sunm, & Ivry, 2006).
Our findings of double dissociation of DLPFC and M1 invoked by the differences in
practice structure may reflect the relative preponderance of specific motor skill
component that is acquired through variable and constant practice. Variable practice may
lead to a memory representation that is predominantly goal-based, and requires
processing within DLPFC for offline memory stabilization. In contrast, constant practice
may give rise to a predominantly movement-based memory representation that relies on
M1 for consolidation.
96
A second possible explanation stems from the evidence that rTMS effects are
dependent on the state of brain activity (Silvanto & Pascual-Leone, 2008). Therefore, the
differences in the effects of rTMS in each practice group may be related to differences in
the brain activity induced by the practice structure. This seems unlikely in our study
because of two reasons. One, the motor corticospinal excitability data suggest that the
state of the brain activity, as assessed by post-practice motor corticospinal excitability
change was not differentially affected by practice structure. Further, the rTMS applied
over M1 after variable and constant practice significantly downregulated the motor
corticospinal excitability. Second, the observed double dissociation findings may also
provide evidence against any global differences in brain state induced by practice
conditions.
Another possibility may be that the difference is practice condition may actually
trigger different post-practice processes that may be implemented by distinct neural
mechanisms. For example, reconsolidation may occur following variable practice, while
consolidation may follow constant practice. It is speculated that some aspect of the motor
memory may consolidate online during the course of variable practice. It is likely that
reconstruction (retrieval) of the action plan induced by the interspersing of the task in
variable practice may render the established memory labile and sensitive to disruption
and trigger reconsolidation process post-practice to make it more stable. Thus, after
variable practice, reconsolidation processes may be predominant rather than
97
consolidation. Constant repetitive practice that does not force reconstruction (retrieval) of
the skill during practice may lead to motor memory that requires post-practice
stabilization to retain the practiced skill. Distinct cellular mechanisms have been
demonstrated for consolidation and reconsolidation (J. L. Lee, Everitt, & Thomas, 2004).
It may be likely that our findings of differences in neural substrates may reflect these
different processes. This possibility may require further investigation.
Previous investigations have reported M1activation during variable practice that
significantly correlated with post-practice behavioral gains (Wymbs & Grafton, 2009). In
contrast, we show that rTMS applied over M1 following variable practice did not impair
motor learning compared to the no-rTMS condition. How does one explain these
apparently contradictory findings? It is critical to note that Wymbs and Grafton (2009)
investigated M1 activity during the practice phase. It is likely that motor memory
encoding during variable practice involves processing within M1 that is critical for
retention. In contrast, we directly perturbed M1 processing immediately following
variable practice structure and found no differential effect on retention compared with the
no-rTMS condition. Similar lack of effect of M1 interference with rTMS post-variable
practice has been recently reported by Tanaka and colleagues (Tanaka et al., 2009). Our
findings taken together with previous studies suggest a dynamic nature of motor memory
processing mediated by M1 that may, in part, depend on the practice structure.
98
In some respect, our findings differ from those of Tanaka et al (2009) who demonstrated
no effect of M1 perturbation following blocked (constant) practice of a sequence learning
task. Additionally, rTMS over dorsal premotor and SMA immediately following variable
practice did not affect retention performance of the sequences one day later. These
differences could be likely due to differences in the motor task. Our task was a template
matching skill that required a precise scaling of a specific movement pattern in space and
time. Tanaka et al used an explicit sequence learning task. Previous research has
demonstrated that the time course and underlying neural substrates of motor memory
consolidation differ depending in the motor task (Baraduc et al., 2004; D. A. Cohen et al.,
2005; Krakauer, Ghilardi, & Ghez, 1999; Muellbacher et al., 2002; Shadmehr &
Holcomb, 1997; Shemmell et al., 2007). Nevertheless, our research and that of Tanaka et
al provide evidence that the structure of practice may also influence distinct neural
substrates for motor memory consolidation.
This is the first study, to our knowledge, to demonstrate that the structure of
motor practice, variable or constant, previously known to engage different degrees of
cognitive processing, also depends on different neural substrates for motor memory
consolidation. Our work complements previous functional imaging experiments which
suggest recruitment of distinct neural substrates for encoding during different practice
schedules. Our findings extend that work to suggest that practice structures that are more
cognitively challenging (i.e., variable) may engage higher-order motor areas like the
99
dorsolateral prefrontal cortex for motor memory consolidation; while less cognitively
challenging constant practice structures may depend more heavily on primary motor
cortex to mediate motor memory consolidation.
100
CHAPTER SIX
SUMMARY AND GENERAL DISCUSSION
The overall goal of this dissertation was to understand the neural mechanisms that
implement acquisition of motor skills with different practice structures. It is well-
established that practice structure affects motor skill acquisition. Practice structures that
cognitively challenge the learner is often detrimental to immediate performance during
practice compared to relatively simple practice structures. However, challenging practice
structure often leads to better performance at a delayed retention or transfer test
compared to simple practice structure, indicating their superiority in promoting learning.
This delayed benefit of challenging practice conditions, despite a poor practice
performance characterizes a well-known phenomenon in cognitive neuroscience, the
learning-performance distinction. One of the mechanisms that underlie this phenomenon
may relate to how the practice structure may affect certain processes such as
consolidation that evolve post-practice, and modulate the strength of motor memory
representation. This dissertation investigated how differences in practice conditions that
distinctly affect motor learning influence neural substrates of motor memory
consolidation.
Specifically, we aimed to determine how constant and variable practice conditions
modulate the activity of two neural substrates: primary motor cortex (M1) and
dorsolateral prefrontal cortex (DLPFC) during the early motor memory consolidation
101
phase. In this dissertation, we used 1 Hz repetitive transcranial magnetic stimulation
(rTMS) as a neuroimaging tool to study the neural substrates of motor memory
consolidation. 1 Hz rTMS has been used to interfere with specific neurocognitive
processes (Baraduc et al., 2004; Muellbacher et al., 2002; Robertson et al., 2005),
although there are few that have used high frequency rTMS (Shemmell et al., 2007).
There is evidence to suggest that LF-rTMS applied over the cortex suppresses cortical
excitability. This down regulation of cortical excitability may be used to interfere with
specific neural processes. The suppressive effects of LF-rTMS are dependent on its
parameters such as frequency, intensity and duration (Fitzgerald et al., 2002; Lang et al.,
2006). Prior to using rTMS as a tool to induce interference and create a ―virtual lesion‖, it
was critical to establish its validity to down regulate cortical excitability and identify an
optimal dose to do so. In order to identify parameters that most effectively suppress
cortical excitability, we compared the effects of 4 different 1 Hz rTMS doses on motor
cortical excitability (Chapter 4). We chose the most optimal and safe dose to apply a
focal perturbation to M1 or DLPFC immediately after task practice conducted under
either constant or variable conditions. We measured the effect of that perturbation on
learning 24 hours later with a retention and transfer tests.
This chapter begins by summarizing the main results of this investigation with
reference to each a priori hypothesis. Then, I will discuss how this dissertation adds to the
body of knowledge that spans multiple disciplines: neurophysiology, behavioral and
102
cognitive neuroscience in particular. I will also present alternative hypothesis that emerge
from this work, and may form the basis of future investigation. Finally, limitations and
clinical implications of this work are also discussed.
Summary of main results
Three specific a priori hypotheses corresponding to three specific aims were
proposed (Chapter 1). These hypotheses were based on an extensive review of previous
literature, and a set of pilot experiments conducted prior to beginning of these
experiments. Hypothesis 1: Practice of the task under variable conditions will attenuate
performance during practice, but will lead to better performance at the delayed retention
and transfer, compared to practice under constant condition. Hypothesis 2: 1 Hz rTMS
induced interference to M1 immediately post-practice will interfere with the
consolidation process, and attenuate the learning benefits of the constant practice group,
but not the variable practice group. Hypothesis 3: 1 Hz rTMS induced interference to
DLPFC immediately post-practice will interfere with the consolidation process, and
attenuate the learning benefits of the variable practice group, but not the constant practice
group.
Hypothesis 1 was partially supported by the data from the control groups (CP,
VP; Chapter Five). Our results suggested that was no significant difference in the
performance accuracy between the CP and VP groups during practice and at the end of
acquisition. However, variable practice benefitted learning as demonstrated by a
103
significantly more accurate performance in the VP group compared to CP group at
delayed retention and transfer tests. Specifically, it was clear that variable practice led to
better motor memory stabilization as demonstrated by little decay in performance from
the end of acquisition to retention and transfer testing 1 day later. This was in contrast to
constant practice following which participants demonstrated considerable decay in
performance one day following practice. Hypothesis 2 was supported by the data from
the control and M1 interference groups (Chapter 5). rTMS interference applied over M1
immediately post-practice attenuated delayed retention and transfer performance in those
who practiced the skill under constant practice structure, but not those who practiced
under variable practice structure. This result had two plausible implications. One, it may
be that variable practice led to a stronger memory representation that may be resistant to
rTMS interference during the consolidation phase. Alternatively, M1 is critical during
motor memory consolidation only following constant practice. The locus for
consolidation following variable practice may be different than M1. To explore these two
possibilities, we employed rTMS interference over a different neural substrate (DLPFC)
immediately following variable and constant practice and measured the effect of that
perturbation on learning 1 day later with a retention and transfer test (Hypothesis 3). We
observed that interference to DLPFC immediately post practice attenuated learning
(delayed retention and transfer) in those who practiced under variable practice structure
but not in those who practiced under constant practice structure. These findings supported
104
hypothesis 3 (Chapter Five). Our findings of double dissociation provide support for the
idea that constant and variable practice structure not only engages specific cognitive
processes for motor learning, but that the neural substrates which mediate these effects
for motor memory consolidation are specific and unique for the primary motor cortex and
dorsolateral prefrontal cortex, respectively.
Brain-behavior relationship: A two-way street
This dissertation extends our understanding of brain-behavior relationship by
drawing on three distinct, yet interrelated disciplines: neurophysiology, cognitive
neuroscience and behavioral neuroscience. Our results support the notion of experience
dependent neuroplasticity which suggests that the nature of behavioral interventions
(practice) dictate the nature of neuroplasticity. Conversely, we also demonstrate that
time- and substrate-specific interference to neural processes such as consolidation affects
subsequent motor behavior (learning).
The results unravel the nature of motor memory consolidation.
This study was designed to directly probe specific neural substrates that were
previously implied to be involved in early motor memory consolidation during the
immediate post-practice period, and observe the effect of this interference on motor
learning. The results supported the notion that consolidation processes are critical for
retention of practiced motor skills. Further, our findings also supported the role of M1
and DLPFC in early motor memory consolidation. To our knowledge, this is the first
105
study to directly demonstrate that DLPFC is involved in motor memory consolidation.
More importantly, our results demonstrate that the relative contribution of M1 and
DLPFC during early motor memory consolidation is modulated by differences in practice
structure. This finding highlights how distinct behavioral interventions that invoke
different cognitive processes drive distinct brain networks during motor memory
consolidation.
The results provide insights into mechanisms of learning-performance distinction.
Little is known about the cognitive mechanisms that underlie the well-known
phenomenon of learning-performance distinction. Much of the previous work focused on
how differences in encoding processes invoked by different practice structures may result
in behavioral differences during practice and at retention. Our control group results
demonstrate the delayed retention/ transfer benefits of variable practice, a typical finding
that highlights the learning performance distinction. However, little is known about how
post-practice processes such as consolidation contribute towards a delayed emergence of
variable practice benefits. In this study, we directly perturbed the brain structures that
implement motor memory consolidation processes, and modulated the delayed benefits
following both, constant and variable practice. Our findings shed light on how differences
in practice structure influence motor memory consolidation to impact delayed retention/
transfer performance (learning), thereby impacting the learning-performance distinction.
Our work used neurophysiological technique (TMS) within a framework that integrates
106
constructs of cognitive neuroscience (motor memory) and behavioral motor learning
phenomenon (learning-performance distinction). The findings of this study demonstrate
that the framework proposed in Chapter Two allows development and testing of
neurobehavioral hypotheses to advance our understanding of the brain-behavior
relationship during motor skill acquisition.
The study provides information about what is learned with different practice
structures.
Our results demonstrate double dissociation between M1 and DLPFC during
motor memory consolidation that is induced by differences in the practice structure. From
an experimental psychology perspective, what do these findings inform us about the
different practice structures? One explanation may be based on the concept of cognitive
effort. It has been proposed that variable practice engages the learner in a higher
cognitive effort compared to blocked practice. Cognitive effort refers to the mental work
that is involved in processes such as planning, anticipation, processing of post-response
feedback. Increased cognitive effort during variable practice may engage DLPFC more
that M1, and therefore DLPFC may be necessary for consolidation following variable
practice. Further, our results may imply a difference in the nature of motor memory
representation that is formed with each of the practice structure. It is likely that variable
practice may result in a relatively ―higher-order‖ motor memory that is represented in
DLPFC, a higher-level structure in the motor hierarchy. In contrast, simple rote-repetition
107
such as in constant practice may result in a motor memory that is represented at a
relatively lower-level (M1) in the motor hierarchy.
Alternatively, it may be likely that each practice structure may lead to a motor
memory that predominantly represents some specific aspect of the skill (goal or
movement). For example, variable practice may generate a predominantly goal-based
motor memory representation that is known to involve DLPFC. In contrast, constant
practice may result in a movement-based representation that is often M1-mediated. This
interpretation fits well with theoretical frameworks in both cognitive neuroscience
(Robertson, 2009) as well as behavioral motor learning literature (Schmidt, 1975). This
may further provide some validation to our framework presented in Chapter 2 and foster
generation of more hypotheses that form a basis for future investigations.
Finally, as we suggested in Chapter Five, our results may suggest that
fundamentally different processes may occur immediately following constant and
variable practice. One possibility is that reconsolidation may occur immediately post-
variable practice and consolidation may occur after constant practice. Behavioral
literature suggests that variable practice, due to interspersing of tasks, may cause the
learner to ―forget‖ an action plan for a task because he/ she has to perform a new task on
the next trial. So, when the initial task is presented again, it provides opportunity for
retrieval or reconstruction of action plan during practice. Constant practice does not
provide an opportunity for retrieval because the action plan stays in the working memory.
108
The retrieval processes that occur during variable practice may render the motor memory
relatively unstable, and trigger reconsolidation processes post-practice. In contrast,
consolidation processes may follow constant practice, and engage M1. The differences in
neural substrates critical to post-practice memory processing following constant and
variable practice may reflect a distinction in the basic processes i.e. consolidation and
reconsolidation. Consolidation may follow constant practice and rely on M1 for learning,
while reconsolidation may follow variable practice and may rely on DLPFC for retention.
One of the ways to probe this may be to introduce a perturbation (e.g. a distractor task or
TMS in a manner used by Lin et al, 2009) during inter-trial interval with constant
practice. If this explanation were to hold true, a perturbation during the inter-trial interval
of constant practice would induce ―forgetting‖, and prompt retrieval/ reconstruction. This
may lead to two changes: (1) the constant practice group may retain skill similar to the
variable practice control group, and (2) rTMS over DLPFC following this perturbed-
constant practice may attenuate the retention of the skill. Such an experimental finding
would help investigate the exact nature of post-practice processes following constant and
variable practice.
Clinical implications
This dissertation work has significant clinical implications for understanding the
neural basis of learning and recovery. At the heart of this investigation, is a need to
understand the mechanisms of a clinically relevant phenomenon of ―carry over,‖ or
109
retention of practiced functional motor skills following therapy. We confirmed that
variable practice structure promotes long-term retention and transfer of motor skill
compared to constant practice structure. Clinical strategies that promote these
consolidation processes may be critical for ―carry over‖ following rehabilitation
interventions. In addition, using non-invasive brain stimulation with rTMS, we unraveled
the role of specific neural substrates that implement motor memory consolidation
following each of the practice structures. Our approach of combining motor practice with
substrate-specific non-invasive brain stimulation provides a basis for potentially effective
therapeutic interventions to enhance cognitive-motor recovery and learning. Finally, our
work strongly supports the notion that motor skill acquisition is a cognitive-motor
process. This has significant implications on interventions that are targeted at promoting
motor learning and recovery in healthy individuals as well as patients with brain damage.
Limitations, Alternative Interpretations and Future directions
Role of Sleep
Considerable research has demonstrated that motor memory consolidation is time
as well as a sleep dependent process. In our study, we focused on an early phase of this
process that occurs immediately post-practice, and results in a more stable motor memory
representation over time. However, in this study, we did not directly probe the
consolidation processes during sleep. All participants did have 6-7 hours of sleep
between the practice and retention testing. Therefore, we are confident that the observed
110
results were not a result of differences in hours of sleep between groups. However, one
cannot rule out the possibility that rTMS immediately post-practice may have, in some
way, affected characteristic of motor memory, there by affecting its sleep-mediated
consolidation process. Alternatively, rTMS may have modulated brain-activity during
sleep such as slow-frequency oscillations of nonrapid eye movement sleep known to
facilitate consolidation within the prefrontal cortex. The second possibility is unlikely
because when rTMS applied ~4 h post practice did not affect learning compared to the
control groups. Further research is warranted to understand the interaction between time-
dependent and sleep-dependent consolidation processes.
TMS Effects
TMS perturbation technique allows us to directly interfere with the neural
processing at specific neural substrates. However, there are some caveats of TMS-
induced perturbation that need to be considered making conclusions about its effects.
First, TMS may influence more than one brain region. In addition to a direct local effect
upon neuronal elements (axons, dendrites and cell bodies) by the electrical currents
induced within the underlying brain tissue (by the magnetic fields set up around the
stimulating coil), secondary indirect effects are generated through synaptic actions of
these excited elements. The local effects may represents a disruption or facilitation in the
cortical network that is underneath the site of stimulation while the distant effects
represents a transmission of neural activity to other brain sites anatomically connected to
111
the site of stimulation (Van Der Werf & Paus, 2006b). The distant effects may also be
instrumental in driving behavioral effects of rTMS. Therefore, we suggest that M1 and
DLPFC are parts of the neural circuits that are involved during motor memory
consolidation, and their engagement is modulated by practice conditions.
Another important aspect of TMS that needs to be considered is the possible
movement of the coil relative to the subject‘s head and the effect of such a misalignment
on neuronal activity. This was effectively controlled by using a brain navigation system
to localize and monitor the position of the coil over the brain area of interest.
Task
As discussed in Chapter Five, characteristics of the motor task may influence the
neural substrates of motor skill acquisition. Previous studies investigating the neural
substrates of motor memory consolidation have employed a multitude of tasks such as
ballistic finger tapping task, implicit sequence learning task, explicit finger sequence task,
a visuomotor adaptation task, and dynamic force adaptation tasks (Baraduc et al., 2004;
D. A. Cohen et al., 2005; Krakauer et al., 1999; Muellbacher et al., 2002; Shadmehr &
Holcomb, 1997; Shemmell et al., 2007). Our findings are in accordance with those of
Muellbacher and Robertson who demonstrated the role of M1 in motor memory
consolidation following repetitive constant practice. However, our findings differ from
Tanaka who did not show any effect of rTMS over M1 immediately post-constant
practice on learning. These differences in the findings could be attributed to differences
112
in motor task. Our task was a template matching skill that required a precise scaling of a
specific movement pattern in space and time. Tanaka et al used an explicit sequence
learning task. Previous research has demonstrated that the time course and underlying
neural substrates of motor memory consolidation differ depending in the motor task
(Baraduc et al., 2004; D. A. Cohen et al., 2005; Krakauer et al., 1999; Muellbacher et al.,
2002; Shadmehr & Holcomb, 1997; Shemmell et al., 2007). Lastly, the laboratory tasks
we used are analogs to real world skills, bringing into question the generalizability of task
learning in a real-world environment with its meaningful motor skills. Future work
which applies more ―real-life‖ and motivating motor tasks and practice conditions is
necessary to test the generalizability of our findings.
Summary
The present research adds a new dimension to our understanding of motor
memory consolidation and the underlying neural substrates. More importantly, this work
used a dualistic approach to the study of motor learning—one behavioral approach in
which practice conditions were investigated for their beneficial effects on learning and
two, key nodes in putative neural networks were subjected to a virtual lesion using rTMS
for its effect on learning. Our findings compliment, and extend previous experimental
imaging work on motor memory consolidation (Albert N.B. et al., 2009; Robertson et al.,
2004). Finally, our data provide empirical support for a recent theoretical model of motor
memory consolidation (Criscimagna-Hemminger S.E. & R, 2008). Our data, together
113
with an understanding of the behavioral and theoretical basis of human motor memory
may stimulate future research with a goal to enhance learning through direct modulation
of specific human cortical regions during memory consolidation. Our results have
theoretical implications for understanding the architecture of human motor memory and
the neural substrates that implement memory consolidation. In addition, this work also
has significant practical implications for learning motor skills; and offers new insights for
enhancing the effectiveness of rehabilitation interventions.
114
REFERENCES
Albert N.B., Robertson E.M., & R.C., M. (2009). The resting human brain and motor
learning. Current Biology, Jun 23(19(12)), 1023-1027
Albert, N. B., Robertson, E. M., & Miall, R. C. (2009). The resting human brain and
motor learning. Curr Biol, 19(12), 1023-1027.
Baraduc, P., Lang, N., Rothwell, J. C., & Wolpert, D. M. (2004). Consolidation of
dynamic motor learning is not disrupted by rTMS of primary motor cortex. Curr
Biol, 14(3), 252-256.
Barclay, C. R., & Newell, K. M. (1980). Children's processing of information in motor
skill acquisition. J Exp Child Psychol, 30(1), 98-108.
Blumenfeld, R. S., & Ranganath, C. (2006). Dorsolateral prefrontal cortex promotes
long-term memory formation through its role in working memory organization. J
Neurosci, 26(3), 916-925.
Blumenfeld, R. S., & Ranganath, C. (2007). Prefrontal cortex and long-term memory
encoding: an integrative review of findings from neuropsychology and
neuroimaging. Neuroscientist, 13(3), 280-291.
Boyd, L., & Winstein, C. (2006). Explicit information interferes with implicit motor
learning of both continuous and discrete movement tasks after stroke. J Neurol
Phys Ther, 30(2), 46-57; discussion 58-49.
Brashers-Krug, T., Shadmehr, R., & Bizzi, E. (1996). Consolidation in human motor
memory. Nature, 382(6588), 252-255.
Brasil-Neto, J. P., McShane, L. M., Fuhr, P., Hallett, M., & Cohen, L. G. (1992).
Topographic mapping of the human motor cortex with magnetic stimulation:
factors affecting accuracy and reproducibility. Electroencephalogr Clin
Neurophysiol, 85(1), 9-16.
Buitrago, M. M., Ringer, T., Schulz, J. B., Dichgans, J., & Luft, A. R. (2004).
Characterization of motor skill and instrumental learning time scales in a skilled
reaching task in rat. Behav Brain Res, 155(2), 249-256.
115
Cahill, L., McGaugh, J. L., & Weinberger, N. M. (2001). The neurobiology of learning
and memory: some reminders to remember. Trends Neurosci, 24(10), 578-581.
Chen, R., Classen, J., Gerloff, C., Celnik, P., Wassermann, E. M., Hallett, M., et al.
(1997). Depression of motor cortex excitability by low-frequency transcranial
magnetic stimulation. Neurology, 48(5), 1398-1403.
Christina. (1997). Concerns and Issues in Studying and Assessing Motor Learning.
Measurement in physical education and exercise science, 1(1), 19-38.
Classen, J., Liepert, J., Wise, S. P., Hallett, M., & Cohen, L. G. (1998). Rapid plasticity
of human cortical movement representation induced by practice. J Neurophysiol,
79(2), 1117-1123.
Cohen, D. A., Pascual-Leone, A., Press, D. Z., & Robertson, E. M. (2005). Off-line
learning of motor skill memory: a double dissociation of goal and movement.
Proc Natl Acad Sci U S A, 102(50), 18237-18241.
Cohen, N. R., Cross, E. S., Wymbs, N. F., & Grafton, S. T. (2009). Transient disruption
of M1 during response planning impairs subsequent offline consolidation. Exp
Brain Res, 196(2), 303-309.
Criscimagna-Hemminger S.E., & R, S. (2008). Consolidation patterns of human motor
memory. The journal of Neuroscience, Sept 24(28(39)), 9610-9618.
Cross, E. S., Schmitt, P. J., & Grafton, S. T. (2007). Neural substrates of contextual
interference during motor learning support a model of active preparation. J Cogn
Neurosci, 19(11), 1854-1871.
Daskalakis, Z. J., Moller, B., Christensen, B. K., Fitzgerald, P. B., Gunraj, C., & Chen, R.
(2006). The effects of repetitive transcranial magnetic stimulation on cortical
inhibition in healthy human subjects. Exp Brain Res, 174(3), 403-412.
Destrebecqz, A., Peigneux, P., Laureys, S., Degueldre, C., Del Fiore, G., Aerts, J., et al.
(2005). The neural correlates of implicit and explicit sequence learning:
Interacting networks revealed by the process dissociation procedure. Learn Mem,
12(5), 480-490.
116
Farzan, F., Barr, M. S., Wong, W., Chen, R., Fitzgerald, P. B., & Daskalakis, Z. J.
(2009). Suppression of gamma-oscillations in the dorsolateral prefrontal cortex
following long interval cortical inhibition: a TMS-EEG study.
Neuropsychopharmacology, 34(6), 1543-1551.
Fitzgerald, P. B., Brown, T. L., Daskalakis, Z. J., Chen, R., & Kulkarni, J. (2002).
Intensity-dependent effects of 1 Hz rTMS on human corticospinal excitability.
Clin Neurophysiol, 113(7), 1136-1141.
Fitzgerald, P. B., Fountain, S., & Daskalakis, Z. J. (2006). A comprehensive review of
the effects of rTMS on motor cortical excitability and inhibition. Clin
Neurophysiol, 117(12), 2584-2596.
Fuster, J. M. (1995). Memory in the Cerebral Cortex: An Empirical Approach to Neural
Networks in the Human and Nonhuman Primate. Cambridge, MA: MIT Press.
Fuster, J. M. (2000). Executive frontal functions. Exp Brain Res, 133(1), 66-70.
Galea, J. M., Albert, N. B., Ditye, T., & Miall, R. C. (2009). Disruption of the
Dorsolateral Prefrontal Cortex Facilitates the Consolidation of Procedural Skills. J
Cogn Neurosci.
Gangitano, M., Valero-Cabre, A., Tormos, J. M., Mottaghy, F. M., Romero, J. R., &
Pascual-Leone, A. (2002). Modulation of input-output curves by low and high
frequency repetitive transcranial magnetic stimulation of the motor cortex. Clin
Neurophysiol, 113(8), 1249-1257.
Genovesio, A., Brasted, P. J., & Wise, S. P. (2006). Representation of future and previous
spatial goals by separate neural populations in prefrontal cortex. J Neurosci,
26(27), 7305-7316.
Ghilardi, M., Ghez, C., Dhawan, V., Moeller, J., Mentis, M., Nakamura, T., et al. (2000).
Patterns of regional brain activation associated with different forms of motor
learning. Brain Res, 871(1), 127-145.
Gomez Beldarrain, M., Grafman, J., Ruiz de Velasco, I., Pascual-Leone, A., & Garcia-
Monco, C. (2002). Prefrontal lesions impair the implicit and explicit learning of
sequences on visuomotor tasks. Exp Brain Res, 142(4), 529-538.
117
Guadagnoli, M. A., & Lee, T. D. (2004). Challenge point: a framework for
conceptualizing the effects of various practice conditions in motor learning. J Mot
Behav, 36(2), 212-224.
Heide, G., Witte, O. W., & Ziemann, U. (2006). Physiology of modulation of motor
cortex excitability by low-frequency suprathreshold repetitive transcranial
magnetic stimulation. Exp Brain Res, 171(1), 26-34.
Hernandez, J. D., Bann, S. D., Munz, Y., Moorthy, K., Datta, V., Martin, S., et al. (2004).
Qualitative and quantitative analysis of the learning curve of a simulated surgical
task on the da Vinci system. Surg Endosc, 18(3), 372-378.
Honda, M., Deiber, M. P., Ibanez, V., Pascual-Leone, A., Zhuang, P., & Hallett, M.
(1998). Dynamic cortical involvement in implicit and explicit motor sequence
learning. A PET study. Brain, 121 ( Pt 11), 2159-2173.
Houdayer, E., Degardin, A., Cassim, F., Bocquillon, P., Derambure, P., & Devanne, H.
(2008). The effects of low- and high-frequency repetitive TMS on the
input/output properties of the human corticospinal pathway. Exp Brain Res.
Iyer, M. B., Schleper, N., & Wassermann, E. M. (2003). Priming stimulation enhances
the depressant effect of low-frequency repetitive transcranial magnetic
stimulation. J Neurosci, 23(34), 10867-10872.
Karni, A., Meyer, G., Rey-Hipolito, C., Jezzard, P., Adams, M. M., Turner, R., et al.
(1998). The acquisition of skilled motor performance: fast and slow experience-
driven changes in primary motor cortex. Proc Natl Acad Sci U S A, 95(3), 861-
868.
Kiers, L., Cros, D., Chiappa, K. H., & Fang, J. (1993). Variability of motor potentials
evoked by transcranial magnetic stimulation. Electroencephalogr Clin
Neurophysiol, 89(6), 415-423.
Kleim, J. A., Kleim, E. D., & Cramer, S. C. (2007). Systematic assessment of training-
induced changes in corticospinal output to hand using frameless stereotaxic
transcranial magnetic stimulation. Nat Protoc, 2(7), 1675-1684.
Krakauer, J. W., Ghilardi, M. F., & Ghez, C. (1999). Independent learning of internal
models for kinematic and dynamic control of reaching. Nat Neurosci, 2(11),
1026-1031.
118
Krakauer, J. W., & Shadmehr, R. (2006). Consolidation of motor memory. Trends
Neurosci, 29(1), 58-64.
Landau, S. M., & D'Esposito, M. (2006). Sequence learning in pianists and nonpianists:
an fMRI study of motor expertise. Cogn Affect Behav Neurosci, 6(3), 246-259.
Lang, N., Harms, J., Weyh, T., Lemon, R. N., Paulus, W., Rothwell, J. C., et al. (2006).
Stimulus intensity and coil characteristics influence the efficacy of rTMS to
suppress cortical excitability. Clin Neurophysiol, 117(10), 2292-2301.
Lee, J. L., Everitt, B. J., & Thomas, K. L. (2004). Independent cellular processes for
hippocampal memory consolidation and reconsolidation. Science, 304(5672),
839-843.
Lee T.D., S. S. P., Serrien D.J. (1994). Cognitive effort and motor learning. Quest(46),
328-344.
Lee TD, S. D. (2003). Contextual Interference, Information processing perspectives (Vol.
2003).
Lee, T. D. W., L.R. (2005). Motor Learning Conundrums (and possible solutions). Quest,
57, 67-78.
Lin, C. H., Fisher, B. E., Winstein, C. J., Wu, A. D., & Gordon, J. (2008). Contextual
interference effect: elaborative processing or forgetting-reconstruction? A post
hoc analysis of transcranial magnetic stimulation-induced effects on motor
learning. J Mot Behav, 40(6), 578-586.
Lin, C. H., Fisher, B. E., Wu, A. D., Ko, Y. A., Lee, L. Y., & Winstein, C. J. (2009).
Neural correlate of the contextual interference effect in motor learning: a
kinematic analysis. J Mot Behav, 41(3), 232-242.
Luft, A. R., Buitrago, M. M., Kaelin-Lang, A., Dichgans, J., & Schulz, J. B. (2004).
Protein synthesis inhibition blocks consolidation of an acrobatic motor skill.
Learn Mem, 11(4), 379-382.
Luft, A. R., Buitrago, M. M., Ringer, T., Dichgans, J., & Schulz, J. B. (2004). Motor skill
learning depends on protein synthesis in motor cortex after training. J Neurosci,
24(29), 6515-6520.
119
Maeda, F., Keenan, J. P., Tormos, J. M., Topka, H., & Pascual-Leone, A. (2000a).
Interindividual variability of the modulatory effects of repetitive transcranial
magnetic stimulation on cortical excitability. Exp Brain Res, 133(4), 425-430.
Maeda, F., Keenan, J. P., Tormos, J. M., Topka, H., & Pascual-Leone, A. (2000b).
Modulation of corticospinal excitability by repetitive transcranial magnetic
stimulation. Clin Neurophysiol, 111(5), 800-805.
McGaugh, J. L. (1989). Dissociating learning and performance: drug and hormone
enhancement of memory storage. Brain Res Bull, 23(4-5), 339-345.
Meister, I., Krings, T., Foltys, H., Boroojerdi, B., Muller, M., Topper, R., et al. (2005).
Effects of long-term practice and task complexity in musicians and nonmusicians
performing simple and complex motor tasks: implications for cortical motor
organization. Hum Brain Mapp, 25(3), 345-352.
Mills, K. R., Boniface, S. J., & Schubert, M. (1992). Magnetic brain stimulation with a
double coil: the importance of coil orientation. Electroencephalogr Clin
Neurophysiol, 85(1), 17-21.
Modugno, N., Curra, A., Conte, A., Inghilleri, M., Fofi, L., Agostino, R., et al. (2003).
Depressed intracortical inhibition after long trains of subthreshold repetitive
magnetic stimuli at low frequency. Clin Neurophysiol, 114(12), 2416-2422.
Muellbacher, W., Ziemann, U., Boroojerdi, B., & Hallett, M. (2000). Effects of low-
frequency transcranial magnetic stimulation on motor excitability and basic motor
behavior. Clin Neurophysiol, 111(6), 1002-1007.
Muellbacher, W., Ziemann, U., Wissel, J., Dang, N., Kofler, M., Facchini, S., et al.
(2002). Early consolidation in human primary motor cortex. Nature, 415(6872),
640-644.
Naveh-Benjamin, M., Kilb, A., & Fisher, T. (2006). Concurrent task effects on memory
encoding and retrieval: further support for an asymmetry. Mem Cognit, 34(1), 90-
101.
Ohnishi, T., Matsuda, H., Imabayashi, E., Okabe, S., Takano, H., Arai, N., et al. (2004).
rCBF changes elicited by rTMS over DLPFC in humans. Suppl Clin
Neurophysiol, 57, 715-720.
120
Okano H, H. T., Balaban E. (2000). Learning and memory. Proceedings of the National
Academy of Sciences of the United States of America, 97(23), 12403-12404.
Pascual-Leone, A., Bartres-Faz, D., & Keenan, J. P. (1999). Transcranial magnetic
stimulation: studying the brain-behaviour relationship by induction of 'virtual
lesions'. Philos Trans R Soc Lond B Biol Sci, 354(1387), 1229-1238.
Pascual-Leone, A., Walsh, V., & Rothwell, J. (2000). Transcranial magnetic stimulation
in cognitive neuroscience--virtual lesion, chronometry, and functional
connectivity. Curr Opin Neurobiol, 10(2), 232-237.
Paus, T., Castro-Alamancos, M. A., & Petrides, M. (2001). Cortico-cortical connectivity
of the human mid-dorsolateral frontal cortex and its modulation by repetitive
transcranial magnetic stimulation. Eur J Neurosci, 14(8), 1405-1411.
Paus, T., Jech, R., Thompson, C. J., Comeau, R., Peters, T., & Evans, A. C. (1998). Dose-
dependent reduction of cerebral blood flow during rapid-rate transcranial
magnetic stimulation of the human sensorimotor cortex. J Neurophysiol, 79(2),
1102-1107.
Plewnia, C., Lotze, M., & Gerloff, C. (2003). Disinhibition of the contralateral motor
cortex by low-frequency rTMS. Neuroreport, 14(4), 609-612.
Pochon, J. B., Levy, R., Poline, J. B., Crozier, S., Lehericy, S., Pillon, B., et al. (2001).
The role of dorsolateral prefrontal cortex in the preparation of forthcoming
actions: an fMRI study. Cereb Cortex, 11(3), 260-266.
Poldrack, R. A., Sabb, F. W., Foerde, K., Tom, S. M., Asarnow, R. F., Bookheimer, S.
Y., et al. (2005). The neural correlates of motor skill automaticity. J Neurosci,
25(22), 5356-5364.
Reis, J., Robertson, E., Krakauer, J. W., Rothwell, J., Marshall, L., Gerloff, C., et al.
(2008). Consensus: "Can tDCS and TMS enhance motor learning and memory
formation?" Brain Stimulat, 1(4), 363-369.
Robertson, E. M. (2004). Skill learning: putting procedural consolidation in context. Curr
Biol, 14(24), R1061-1063.
121
Robertson, E. M. (2009). From creation to consolidation: a novel framework for memory
processing. PLoS Biol, 7(1), e19.
Robertson, E. M., & Cohen, D. A. (2006). Understanding consolidation through the
architecture of memories. Neuroscientist, 12(3), 261-271.
Robertson, E. M., Pascual-Leone, A., & Miall, R. C. (2004). Current concepts in
procedural consolidation. Nat Rev Neurosci, 5(7), 576-582.
Robertson, E. M., Press, D. Z., & Pascual-Leone, A. (2005). Off-line learning and the
primary motor cortex. J Neurosci, 25(27), 6372-6378.
Romeo, S., Gilio, F., Pedace, F., Ozkaynak, S., Inghilleri, M., Manfredi, M., et al. (2000).
Changes in the cortical silent period after repetitive magnetic stimulation of
cortical motor areas. Exp Brain Res, 135(4), 504-510.
Romero, J. R., Anschel, D., Sparing, R., Gangitano, M., & Pascual-Leone, A. (2002).
Subthreshold low frequency repetitive transcranial magnetic stimulation
selectively decreases facilitation in the motor cortex. Clin Neurophysiol, 113(1),
101-107.
Rossi, S., Hallett, M., Rossini, P. M., & Pascual-Leone, A. (2009). Safety, ethical
considerations, and application guidelines for the use of transcranial magnetic
stimulation in clinical practice and research. Clin Neurophysiol, 120(12), 2008-
2039.
Rowe, J. B., Toni, I., Josephs, O., Frackowiak, R. S., & Passingham, R. E. (2000). The
prefrontal cortex: response selection or maintenance within working memory?
Science, 288(5471), 1656-1660.
Saisanen, L., Pirinen, E., Teitti, S., Kononen, M., Julkunen, P., Maatta, S., et al. (2008).
Factors influencing cortical silent period: optimized stimulus location, intensity
and muscle contraction. J Neurosci Methods, 169(1), 231-238.
Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984). Knowledge of results and motor
learning: a review and critical reappraisal. Psychol Bull, 95(3), 355-386.
Schmidt, R. A. (1975). A schema theory of discrete motor skill learning. Psychological
Review, 82, 225-260.
122
Schmidt, R. A., & Bjork, R. A. (1992). New conceptualization of practice: Common
principles in three paradigms suggest new concepts for training. Psychological
Science, 3, 207-217.
Schmidt, R. A., & Lee, T. D. (2004). Motor control and learning: A behavioral emphasis
(Fourth edition ed.): Human Kinetics.
Sekiya, H., Magill, R. A., & Anderson, D. I. (1996). The contextual interference effect in
parameter modifications of the same generalized motor program. Res Q Exerc
Sport, 67(1), 59-68.
Shadmehr, R., & Brashers-Krug, T. (1997). Functional stages in the formation of human
long-term motor memory. J Neurosci, 17(1), 409-419.
Shadmehr, R., & Holcomb, H. H. (1997). Neural correlates of motor memory
consolidation. Science, 277(5327), 821-825.
Shapiro, D. C. (1977). Knowledge of results and motor learning in preschool children.
Res Q, 48(1), 154-158.
Shea, C. H., & Kohl, R. M. (1990). Specificity and variability of practice. Res Q Exerc
Sport, 61(2), 169-177.
Shea, C. H., & Kohl, R. M. (1991). Composition of practice: influence on the retention of
motor skills. Res Q Exerc Sport, 62(2), 187-195.
Shea, C. H., Lai, Q., Wright, D. L., Immink, M., & Black, C. (2001). Consistent and
variable practice conditions: effects on relative and absolute timing. J Mot Behav,
33(2), 139-152.
Shea, J. B., & Wright, D. L. (1991). When forgetting benefits motor retention. Res Q
Exerc Sport, 62(3), 293-301.
Shemmell, J., Riek, S., Tresilian, J. R., & Carson, R. G. (2007). The role of the primary
motor cortex during skill acquisition on a two-degrees-of-freedom movement
task. J Mot Behav, 39(1), 29-39.
Sherwood, D. E., & Lee, T. D. (2003). Schema theory: critical review and implications
for the role of cognition in a new theory of motor learning. Res Q Exerc Sport,
74(4), 376-382.
123
Siengsukon, C. F., & Boyd, L. A. (2009). Does sleep promote motor learning?
Implications for physical rehabilitation. Phys Ther, 89(4), 370-383.
Silvanto, J., & Pascual-Leone, A. (2008). State-dependency of transcranial magnetic
stimulation. Brain Topogr, 21(1), 1-10.
Speer, A. M., Willis, M. W., Herscovitch, P., Daube-Witherspoon, M., Shelton, J. R.,
Benson, B. E., et al. (2003). Intensity-dependent regional cerebral blood flow
during 1-Hz repetitive transcranial magnetic stimulation (rTMS) in healthy
volunteers studied with H215O positron emission tomography: I. Effects of
primary motor cortex rTMS. Biol Psychiatry, 54(8), 818-825.
Spencer, R. M., Sunm, M., & Ivry, R. B. (2006). Sleep-dependent consolidation of
contextual learning. Curr Biol, 16(10), 1001-1005.
Tanaka, S., Honda, M., Hanakawa, T., & Cohen, L. G. (2009). Differential Contribution
of the Supplementary Motor Area to Stabilization of a Procedural Motor Skill
Acquired through Different Practice Schedules. Cereb Cortex.
Todd, G., Flavel, S. C., & Ridding, M. C. (2006). Low-intensity repetitive transcranial
magnetic stimulation decreases motor cortical excitability in humans. J Appl
Physiol, 101(2), 500-505.
Touge, T., Gerschlager, W., Brown, P., & Rothwell, J. C. (2001). Are the after-effects of
low-frequency rTMS on motor cortex excitability due to changes in the efficacy
of cortical synapses? Clin Neurophysiol, 112(11), 2138-2145.
Van Der Werf, Y. D., & Paus, T. (2006a). The neural response to transcranial magnetic
stimulation of the human motor cortex. I. Intracortical and cortico-cortical
contributions. Exp Brain Res, 175(2), 231-245.
Van Der Werf, Y. D., & Paus, T. (2006b). The neural response to transcranial magnetic
stimulation of the human motor cortex. I. Intracortical and cortico-cortical
contributions. Exp Brain Res.
Walker, M. P., Brakefield, T., Seidman, J., Morgan, A., Hobson, J. A., & Stickgold, R.
(2003). Sleep and the time course of motor skill learning. Learn Mem, 10(4), 275-
284.
124
Walker, M. P., & Stickgold, R. (2004). Sleep-dependent learning and memory
consolidation. Neuron, 44(1), 121-133.
Wassermann, E. M. (1998). Risk and safety of repetitive transcranial magnetic
stimulation: report and suggested guidelines from the International Workshop on
the Safety of Repetitive Transcranial Magnetic Stimulation, June 5-7, 1996.
Electroencephalogr Clin Neurophysiol, 108(1), 1-16.
Wymbs, N. F., & Grafton, S. T. (2009). Neural substrates of practice structure that
support future off-line learning. J Neurophysiol, 102(4), 2462-2476.
Abstract (if available)
Abstract
Motor practice drives subsequent offline activity within functionally related resting brain networks. Little is known about how offline neural networks are modulated by practice structures known to affect motor skill learning. To investigate the neural correlates of motor memory consolidation, we applied 1 Hz repetitive Transcranial Magnetic Stimulation (rTMS) immediately after a bout of constant or variable motor practice to disrupt either primary motor cortex (M1) or dorsolateral prefrontal cortex (DLPFC), two putative nodes previously shown to be engaged in early consolidation. Motor learning was assessed the following day through a performance-based retention test. Immediately after constant practice, rTMS to M1, but not DLPFC attenuated retention of the motor skill. In contrast, immediately after variable practice, rTMS to DLPFC, but not M1 attenuated retention performance. These findings provide evidence that for motor skills, the neural substrates of motor memory consolidation are modulated by practice structure.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Contextual interference in motor skill learning: an investigation of the practice schedule effect using transcranial magnetic stimulation (TMS)
PDF
Neural substrates associated with context-dependent learning
Asset Metadata
Creator
Kantak, Shailesh S.
(author)
Core Title
Neural substrates of motor memory consolidation: a double dissociation of primary motor cortex and dorsolateral prefrontal cortex induced by practice structure
School
School of Dentistry
Degree
Doctor of Philosophy
Degree Program
Biokinesiology
Publication Date
05/11/2011
Defense Date
03/03/2010
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
motor memory consolidation,neural substrates,OAI-PMH Harvest,practice structure
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Sullivan, Katherine J. (
committee chair
), Winstein, Carolee J. (
committee chair
), Azen, Stanley Paul (
committee member
), Fisher, Beth (
committee member
), Knowlton, Barbara (
committee member
)
Creator Email
kantak@usc.edu,sskantak@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m3074
Unique identifier
UC1303367
Identifier
etd-Kantak-3629 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-340348 (legacy record id),usctheses-m3074 (legacy record id)
Legacy Identifier
etd-Kantak-3629.pdf
Dmrecord
340348
Document Type
Dissertation
Rights
Kantak, Shailesh S.
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
motor memory consolidation
neural substrates
practice structure