University of Southern California Dissertations and Theses

Neural substrates associated with context-dependent learning

(USC Thesis Other)

Neural substrates associated with context-dependent learning

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
NEURAL SUBSTRATES ASSOCIATED WITH CONTEXT-DEPENDENT
LEARNING

by

Ya-Yun (Alice) Lee

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)

December 2013

Copyright 2013 Ya-Yun (Alice) Lee
ii

ACKNOWLEDGEMENT
This dissertation could not be completed without the guidance, support, and assistance of
many individuals. I am fortunate to have the guidance and help from these individuals
throughout the past five years so that I am able to complete the dissertation smoothly.
First and foremost, I would like to express my deepest appreciation to my primary advisor,
Dr. Beth Fisher. I feel extremely fortunate and am honored to be Dr. Fisher’s first PhD student.
She teaches me from every aspect she could, and guides me and accompanies me through the
PhD process. Dr. Fisher is full of wisdom and always provides me the wisest suggestions when I
am not sure how to make my decisions. She is also extremely patient and spent extensive amount
of time with me in discussing research ideas, figuring out the best study design, interpreting the
results, practicing presentations, and editing manuscripts. She is always very supportive and
helps me overcome difficulties. She is a role model for me in every aspect, especially as a mentor,
and I hope I could be as good as her one day. I am indebted to Dr. Carolee Winstein, my co-chair
and co-advisor, for her the guidance and mentorship. I would like to thank Dr. Winstein for
always challenging my thinking process from different perspectives and encouraging me to think
outside of the box. She always brings new thoughts and ideas into our discussion and teaches me
the knowledge outside of my own realm. I value every chance to meet with Dr. Winstein, and she
is the role model for me as a scientist. I am very grateful to Dr. James Gordon who leads me to
read into the past and think about the future. Dr. Gordon is the person who inserted this
dissertation idea into my head. I am thankful to Dr. Gordon for keeping me on the right track and
providing excellent suggestions for the dissertation work. He always foresees the interesting
parts of the results and encourages me to explore the data. His expertise in analyzing the data and
plotting graphs tremendously enriched the finding of this dissertation work. I have learned so
iii

much from Dr. Gordon’s wise thinking process. I would like to thank Dr. Giselle Petzinger for
being always so enthusiastic about this dissertation work. I learned a lot from her expertise in
Parkinson’s disease, neuroanatomy, and medication usage. I would also like to thank Dr.
Petzinger for her support and help in recruiting participants so that I am able to complete the
studies so smoothly. I would like to acknowledge my last committee member, Dr. Liz Zelinski. I
am grateful for her time discussing motor learning and neuroplasticity papers with me while
bringing in valuable insights from the psychology and gerontology fields. Dr. Zelinski also
provides extraordinary ideas in data analysis and statistical methods for this dissertation work.
In addition to my committee members, I would like to thank all the faculty and staff from
the Division of Biokinesiology and Physical Therapy. Especially, I would like to thank Dr.
Lucinda Baker, who is a role model for me not only as a scientist but also as an instructor. I
really learned a lot from Dr. Baker. I would also like to acknowledge Lydia Vazquez, Narissa
Casebeer, Jennifer Bandich, Janet Burney, Matt Sandusky, David Donaldson, and Chad Louie for
always providing the most immediate and emergent help that I need. Their help and support
made the whole PhD process possible.
I would like to acknowledge all the previous and current members of the Motor Behavior
and Rehabilitation Laboratory (MBNL) and Neuroplasticiy and Imaging Laboratory (NAIL). I
am grateful to Dr. Jill Stewart, Dr. Shailesh Kantak, Dr. Hui-Ting Goh, Dr. Hsiang-Han Huang,
Dr. Shuya Chen, Dr. Eric Wade, and Dr. Sue Duff for helping and guiding me through every step
of the process. They provided me valuable life and school experiences, and brainstormed ideas
with me. They are good mentors as well as good friends that I will always cherish. I am also
grateful to have colleagues Matt Konersman, Clarisa Martinez, Bobby Charalambos, Yi-An Chen,
Yu-Chen Chung, Bokkyu Kim, and Helen Bacon for accompany me through the years down in
iv

the basement. In addition to MBNL members, I would also like to thank friends and colleagues
from the department – Dr. Barbara Sargent, Dr. Kate Havens, Dr. Szu-Ping Lee, Sharon Teng,
Eugene Chang, and Jo Smith – for constantly giving me advises and discussing about fields that I
am not familiar with.
Most importantly, I would like to acknowledge the support from my parents, Dr. Wei-
Cheng Lee and Mei-Chen Liu. Their unlimited mental support and encouragement is the force
for keeping me stay at USC and pursue this degree. This whole process is not possible without
their unlimited love and faith in me. I would also like to thank my grandmother, aunts and uncles,
Dr. Yung-Tsung Hsiao, and all my friends from Taiwan for constantly encouraging me
throughout the whole PhD process.
This dissertation work is supported by the funding from Division of Biokinesiology and
Physical Therapy at the University of Southern California, Ministry of Education in Taiwan,
North American Society for Psychology of Sport and Physical Activity, and Team Parkinson and
Parkinson’s Alliance.
v

TABLE OF CONTENTS

LIST OF FIGURES ...................................................................................................................... vii

LIST OF TABLES ......................................................................................................................... ix

ABSTRACT .....................................................................................................................................x

CHAPTER 1: BACKGROUND AND OVERVIEW ......................................................................1

CHAPTER 2: CONTEXT-DEPENDENT LEARNING IN PEOPLE WITH PARKINSON’S
DISEASE .........................................................................................................................................9
Introduction .................................................................................................................................9
Methods .....................................................................................................................................11
Results .......................................................................................................................................15
Discussion .................................................................................................................................20
Clinical Implications/ Conclusions ...........................................................................................23

CHAPTER 3: RELATIONSHIP BETWEEN CONTEXT-DEPENDENT LEARNING AND
SET-SHIFTING ABILITY ............................................................................................................25
Introduction ...............................................................................................................................25
Methods .....................................................................................................................................27
Results .......................................................................................................................................30
Discussion .................................................................................................................................31

CHAPTER 4: ROLE OF THE DORSOLATERAL PREFRONTAL CORTEX IN CONTEXT-
DEPENDENT LEARNING ...........................................................................................................35
vi

Introduction ...............................................................................................................................35
Methods .....................................................................................................................................37
Results .......................................................................................................................................42
Discussion .................................................................................................................................48
Conclusion .................................................................................................................................51

CHAPTER 5: SUMMARY AND GENERAL DISCUSSION ......................................................53
Summary of Main Results .........................................................................................................54
Role of the Frontostriatal Circuit in Context-Dependent Learning ...........................................55
Implications for Motor Learning ...............................................................................................62
Clinical Implications .................................................................................................................64
Limitations and Future Directions ..............................................................................................67
Conclusion .................................................................................................................................68

REFERENCES ..............................................................................................................................70

vii

LIST OF FIGURES

Figure 1.1 Illustration of dissertation hypothesis. The neuronal interaction between
the dorsolateral prefrontal cortex (DLPFC) and the dorsal striatum within
the frontostriatal circuit is hypothesized to play specific roles in context-
dependent learning.

4
Figure 1.2 Illustrations of the hypotheses for Study 1 and Study 3. It is hypothesized
that (A) people with PD who have striatum impairment would
demonstrate excessive context-dependent learning when compared to
control participants, while (B) perturbation to the DLPFC with low
frequency rTMS in non-disabled adults would lead to a decreased
context-dependent learning.

7
Figure 2.1 The finger sequence task. (A) The participants practiced three sequences
in a pseudorandom order. Each sequence was embedded within a specific
color and spatial location on the computer screen. (B) Participants
pressed the corresponding key button in the order of 1-2-3-4. (C)
Examples of SAME and SWITCH conditions during the second day
tests. The SAME condition was when the sequence and its associated
context remained the same as practice, while the SWITCH condition was
when the sequence and its associated context changed from that of
practice.

13
Figure 2.2 (A) Total time, (B) response time, and (C) movement time accuracy cost
of the PD and the Control groups throughout practice blocks and second
day testing conditions.
Abbreviation: SW: SWITCH condition, SA: SAME condition.

16
Figure 2.3 Percentage switch cost of the total time, response time, and movement
time accuracy costs of the PD and the Control groups.
Abbreviation: TT_AC: total time accuracy cost, RT_AC: response time
accuracy cost, MT_AC: movement time accuracy cost. * p < 0.05

18
Figure 2.4 Time differences between the SWITCH and SAME conditions of each
finger key press. When the sequence-context association was changed
under the SWITCH condition, the participants in the Control group spent
most of the extra time during the response time period (1st key press).
On the other hand, participants with PD demonstrated additional
planning time for each subsequent key press in order to carry out the
19
viii

numerical sequence accurately. * p < 0.05

Figure 3.1 Figure 3.1 Spearman correlation between result of the trail making test
(TMT) and the switch cost of total time accuracy cost (TT
AC
) in (A)
individuals with Parkinson’s disease, and (B) non-disabled adults. The
number next to each data point in the PD group represents the Hoehn and
Yahr Stage score for each individual participant.

31
Figure 4.1 Study procedure. All participants in the three groups practiced the finger
sequence task on the first day. On the second day, the participants were
tested under the SWITCH and SAME testing conditions (see Methods
section for detailed description). The participants in the DLPFC and
Vertex groups received low frequency repetitive transcranial magnetic
stimulation (rTMS) before the two testing conditions, while the
participants in the Control group did not receive rTMS.

41
Figure 4.2 (A) Total time, (B) response time, and (C) movement time accuracy cost
results of the Control group (open circle), the DLPFC group (closed
square), and the Vertex group (open square) throughout the first day
practice blocks. All three groups improved throughout practice and
achieved a similar performance level by the end of practice.

43
Figure 4.3 Results of the motor performance under the SWITCH and SAME testing
conditions in (A) total time, (B) response time, and (C) movement time
accuracy cost. End of practice (EOP) data was presented here for a better
visualization of the results demonstrating that the DLPFC group showed
less decrement under the SWITCH condition compared to the Control
and Vertex groups.

45
Figure 4.4 Percentage switch cost of the total time accuracy cost (TT_AC), response
time accuracy cost (RT_AC), and movement time accuracy cost
(MT_AC) of the Control, DLPFC, and Vertex groups. The DLPFC group
demonstrates a smaller switch cost than the other two groups. * p < 0.05

47
Figure 5.1 Combined total time accuracy cost (TT_AC) switch cost results for the
PD group (black), the Control group (white), the DLPFC group (dark
gray), and the Vertex group (light gray). Compared to the Control and the
Vertex groups, the PD group demonstrates a greater switch cost while the
DLPFC group demonstrates a smaller switch cost.

62
ix

LIST OF TABLES

Table 3.1 Mean values of the outcomes measures for the Parkinson’s disease (PD)
and Control groups.
Data is presented in mean ± standard deviation.
Abbreviations: TT
AC
: Total time accuracy cost, TMT-B: Part B of the
trail making test, TMT-A: Part A of the trail making test.

30

x

ABSTRACT

This dissertation is designed to investigate neural substrates associated with context-
dependent learning. Context-dependent learning is a phenomenon in which people demonstrate
superior performance in the environmental context where they originally learned a motor task
and conversely, do not perform as well if the task is carried out in a novel context. While
context-dependent learning has been mostly established in healthy young adults, it has not been
systematically investigated in people with neurological disorders or non-disabled older adults. In
addition, the neural substrates associated with context-dependent learning for motor skill
acquisition are not well understood. One neural network that could potentially be associated with
context-dependent learning is the frontostriatal circuit – recurrent neural connections between the
dorsolateral prefrontal cortex (DLPFC) and the dorsal striatum. Animal and computer simulation
studies have suggested that the frontostriatal circuit is important for selecting an appropriate
movement plan according to environmental stimuli. While the DLPFC encodes all contextual
information associated with a task, the dorsal striatum selects and filters task relevant
information in order to generate the most appropriate action plan. Based on this proposed
function of the frontostriatal circuit in processing contextual information, we hypothesized that
the neuronal interactions between the DLPFC and dorsal striatum within the frontostriatal circuit
could be important for mediating context-dependent learning. Therefore, three studies were
designed in this dissertation to test this hypothesis.
In the first study, we recruited individuals with Parkinson’s disease (PD), known to have
striatum impairments, to test the hypothesis that striatum is a potential neural substrate for
context-dependent learning. Ten individuals with PD and 10 age-matched non-disabled adults
xi

were recruited into the PD group and the Control group. The study was conducted over two
consecutive days approximately 24 hours apart. On the first day, participants practiced a finger
sequence task consisting of 3 numerical sequences. Unknown to the participants, each sequence
was embedded within a specific colored circle and a specific location on the computer screen. On
the second day, the participants were tested under two testing conditions: SWITCH and SAME
conditions. Under the SWITCH condition, the context associated with each sequence was
changed from that of practice; while under the SAME condition, the sequence-context
association remained the same as practice. The primary outcome measure was total time
accuracy cost (TT
AC
), which took both movement speed and accuracy into account. From the
second day testing conditions, switch cost was calculated to indicate context-dependent learning.
Switch cost was the TT
AC
performance difference between the SWITCH and SAME conditions
normalized by the SAME condition (100 % × [SWITCH − SAME] / SAME). A larger switch
cost value would indicate greater context-dependent learning. The results showed that
individuals with PD and non-disabled adults demonstrated comparable learning of the finger
sequence task when tested under the SAME condition. When tested under the SWITCH
condition, participants in both groups showed a decrement in motor performance. However,
individuals with PD demonstrated a significantly greater decrement in motor performance than
the control participants, leading to a higher TT
AC
switch cost. Additional analysis of the switch
cost showed that when the sequence-context association was switched, participants with PD
engaged in additional planning while executing subsequent finger presses.
Study 2 was designed to investigate whether the frontostriatal circuit is associated with
context-dependent learning. To indicate the integrity of the frontostriatal circuit, set-shifting
ability was tested. The participants in Study 1 also participated in this study. After completion of
xii

the finger sequence task on the second day, the participants were given the trail making test
(TMT) to assess their set-shifting ability. The result of the TMT was correlated with the TT
AC

switch cost obtained from Study 1. Findings from Study 2 showed that TT
AC
switch cost was
positively correlated with the result of the TMT in people with PD, suggesting that an individual
with PD who had greater difficulty performing set-shifting also demonstrated greater context-
dependency. However, this relationship was not observed in non-disabled adults. The results of
Study 2 suggested that context-dependent learning could be related to the integrity of the
frontostriatal circuit.
To test the hypothesis that the DLPFC plays a specific role in context-dependent learning,
30 non-disabled adults (age-matched to the participants with PD in Study 1) were recruited for
Study 3. The participants were recruited into the Control group, the rTMS DLPFC group and
rTMS Vertex group. The participants in the Control group were the same participants in Study 1.
Similar to the procedures of Study 1, all participants practiced the finger sequence task on the
first day. Before the SWITCH and SAME testing conditions on the second day, the participants
in the rTMS DLPFC and rTMS Vertex groups received 1 Hz rTMS over the left DLPFC or the
Vertex for 20 minutes. The rTMS DLPFC group demonstrated a reduced TT
AC
switch cost when
compared to the Control group or the rTMS Vertex group, suggesting that perturbation to the
DLPFC reduced context-dependent learning compared to the control conditions.
Overall, findings from these three studies suggest that the neuronal interaction between the
DLPFC and the striatum within the frontostriatal circuit have specific roles in context-dependent
learning. While impairment of the striatum as exists in PD leads to a heightened context-
dependent learning, decreased neuronal excitability of the DLPFC reduces context-dependency.
Given these results, it is reasonable to hypothesize that the DLPFC is relatively over-activated in
xiii

people with PD in order to compensate for the impaired striatum. This over-activation of the
DLPFC with excessive encoding may be the cause of greater context-dependency observed in
PD.
This dissertation work is clinically relevant. First, the findings from Study 1 suggest that
training environment is important for people with PD. When designing a rehabilitation program
for individuals with PD, clinicians should consider the training context for the specific motor
task. Second, the results of the third study suggest that context-dependent learning could be
reduced with low frequency rTMS applied over the DLPFC in non-disabled adults. One
interesting future study is to investigate the influence of low frequency rTMS over the DLPFC
on context-dependent learning in people with PD.

1

CHAPTER 1
BACKGROUND AND OVERVIEW

Incidental context is the environmental condition(s) under which a task is performed and
learned (Wright & Shea, 1991). Although not essential for motor performance, incidental context
often contains and provides information that is associated with a particular task (Chun & Jiang,
1998). It has often been observed that an individual demonstrates superior performance if the
testing context is the same as the practice context, while performance may worsen if the testing
and practice contexts are different (Bertsch & Sanders, 2005; Smith & Vela, 2001). Better
performance under the same practice context compared to poorer performance in a different
context is termed context-dependent learning. One well-known empirical example of context-
dependent learning is the “home field advantage,” in which athletes usually can perform best and
have a higher chance to win a game when the game is played on their own practice field
(Schmidt & Lee, 2005).
Context-dependent learning has been well established and investigated in the verbal
learning literature (Smith, 1982, 1986; Smith & Vela, 2001). Neuroimaging studies showed that
context-dependency in verbal learning is associated with the function of the hippocampus
(Bertsch & Sanders, 2005; Vakil, Raz, & Levy, 2010). Contrary to verbal learning studies,
context-dependent learning is less established as a feature of motor skill development. Wright
and Shea (1991) conducted the first study that used a systematic method to investigate context-
dependent learning for motor skill development. They observed that healthy young participants
demonstrated a significant decrement in motor performance when changes were made to the
associated contexts (symbol, color, and location) in which numerical sequences were originally
2

embedded (Wright & Shea, 1991). Keetch et al. also showed that skilled basketball players were
more accurate when performing set shots at the foul line compared to other locations on the court
(Keetch, Schmidt, Lee, & Young, 2005). It was concluded that the superior performance was
mainly due to the specific visual contextual information available at the foul line (Keetch, Lee, &
Schmidt, 2008). Although these studies demonstrated context-dependency during motor skill
learning in healthy young adults, the manifestation of this phenomenon in older adults or clinical
populations has not been investigated. Additionally, the neural substrates associated with
context-dependent motor learning are not known.
Clinically, it has often been observed that people with Parkinson’s disease (PD) appear to
have difficulty generalizing motor skills learned in the clinic to a different environmental context,
such as the home or community (Nieuwboer et al., 2001). In addition, it has been well
established that use of cueing strategies (e.g., visual and/or auditory cues) are effective in
improving motor function for people with PD (Baker, Rochester, & Nieuwboer, 2007). However,
once the cues are removed from the initial training environment, the motor performance may
dramatically decrease (Rochester et al., 2007). These observations suggest that individuals with
PD may over-rely on the available contextual cues to perform motor tasks and thus demonstrate
excessive context-dependent learning. Context-dependent learning has not yet been investigated
in people with PD. Therefore, the first objective of this dissertation was to use a systematic
method to test the hypothesis that whether people with PD demonstrate greater context-
dependency than non-disabled adults when learning a motor task.
The second objective of the dissertation was to determine the neural substrates associated
with this phenomenon. While the evidence is primarily anecdotal and based on clinical reports,
the striatum, known to be impaired in people with PD, could be an important neural substrate
3

associated with context-dependent learning. The striatum has numerous neural connections with
other cortical areas forming corticostriatal circuits. Among the five main corticostriatal circuits,
the frontostriatal circuit is functionally identified as subserving cognitive and executive functions,
such as motor learning, action selection, and set-shifting (Alexander, DeLong, & Strick, 1986;
Packard & Knowlton, 2002; Wise, Murray, & Gerfen, 1996). In particular, the frontostriatal
circuit has been proposed to play an important role in processing contextual information during
action selection (Dominey & Boussaoud, 1997; Nee & Brown, 2013). The frontostriatal circuit
comprises the recurrent neural connections between the dorsolateral prefrontal cortex (DLPFC)
and the dorsal striatum (Alexander et al., 1986). While the DLPFC receives and encodes all types
of sensory contextual information from the sensorimotor areas, the dorsal striatum mainly
functions as a gate or a modulator to select and filter the information that is behaviorally relevant
(Chakravarthy, Joseph, & Bapi, 2010; Dominey & Boussaoud, 1997; Frank, 2011; Moustafa &
Gluck, 2011; Wise et al., 1996). Neuronal interaction between the DLPFC and the dorsal
striatum seems to be important for selecting the appropriate contextual information in order to
generate the most optimal behavior (Nee & Brown, 2013). Based on this proposed role in
processing contextual information, we hypothesize that the frontostriatal circuit plays a role in
context-dependent learning (Figure 1.1).
4

Figure 1.1 Illustration of dissertation hypothesis. The neuronal interaction between the
dorsolateral prefrontal cortex (DLPFC) and the dorsal striatum within the frontostriatal circuit is
hypothesized to play specific roles in context-dependent learning.

When part of the frontostriatal circuit is malfunctioning, such as the striatum impairment in
PD, the interaction between the DLPFC and dorsal striatum would be affected. This in turn may
affect the ability to select the most appropriate action plan according to the environmental
context (Fogelson, Shah, Scabini, & Knight, 2009; E. Y . Lee et al., 2010). In the first study of
this dissertation, we investigated people with PD (a human model of striatal impairment) to
determine whether the dorsal striatum plays a role in context-dependent learning. We
hypothesized that when the striatum is impaired, people with PD would demonstrate difficulty
filtering and selecting task relevant information leading to excessive context-dependency when
compared to non-disabled adults (Figure 1.2A).
In addition to its role in processing contextual information, the frontostriatal circuit plays
important roles in several other cognitive functions (Packard & Knowlton, 2002). One of the
most well-known cognitive functions mediated by the frontostriatal circuit is set-shifting
5

(Monchi et al., 2004; Monchi, Petrides, Petre, Worsley, & Dagher, 2001; Owen, Roberts, Polkey,
Sahakian, & Robbins, 1991). Set-shifting is the ability for an individual to quickly alter or
reorganize behavior when there is a change in goal or environmental circumstances (Flowers &
Robertson, 1985). Neuroimaging studies have shown activations in the DLPFC, dorsal striatum,
and thalamus while performing set-shifting tasks, suggesting that the frontostriatal circuit is the
neural network that mediates set-shifting ability (Monchi et al., 2004; Monchi et al., 2001). In
addition, individuals with PD demonstrate difficulty performing set-shifting tasks even in the
early stages of the disease, suggesting that malfunctioning of the frontostriatal circuit influences
the set-shifting ability (Cools, Barker, Sahakian, & Robbins, 2001; Monchi, Petrides, Mejia-
Constain, & Strafella, 2007; Owen et al., 1992; Owen et al., 1991). Therefore, to investigate the
association between the frontostriatal circuit and context-dependent learning, set-shifting ability
could be used to correlate with context-dependent learning. We hypothesized that an individual
with greater difficulty performing set-shifting would also demonstrate greater context-
dependency.
To understand the neuronal interaction between the DLPFC and the dorsal striatum within
the frontostriatal circuit, the third study of this dissertation work used low frequency repetitive
transcranial magnetic stimulation (rTMS) to perturb the neuronal processing of the DLPFC in
non-disabled adults. Low frequency rTMS is a non-invasive neuroimaging device that could
temporarily reduce neuronal excitability of the stimulated area (Hallett, 2007). Since the DLPFC
has been proposed to play a role in encoding contextual information and maintaining the
information in working memory (Dominey & Boussaoud, 1997), perturbation to the DLPFC with
rTMS would reduce the context encoding process. With the reduced encoding of contextual
information, we hypothesized that individuals who receive rTMS over the DLPFC would be less
6

affected by the change of context and thus demonstrate less context-dependency (Figure 1.2B).
The overall goal of this dissertation is to investigate whether the frontostriatal circuit is the
neural network associated with context-dependent learning for motor skill development. Three
specific aims are included in this dissertation work.
Specific aim 1: To investigate context-dependent learning in individuals with PD and non-
disabled adults.
Hypothesis: Individuals with PD will demonstrate greater context-dependency (as indexed
by greater switch cost) than non-disabled adults in a finger sequence task.
Specific aim 2: To determine the association between the set-shifting ability and context-
dependent learning.
Hypothesis: A measure of set-shifting will be positively correlated with a measure of
context-dependent learning. Specifically, the results of the Trail Making Test (TMT) will be
positively correlated with the percentage of switch cost in the finger sequence task.
Specific aim 3: to determine the role of the DLPFC in context-dependent learning.
Hypothesis: Perturbing the neuronal processing of the DLPFC in non-disabled adults with
low frequency rTMS will decrease context-dependency (as indexed by smaller switch cost)
when compared to the control participants.

7

Figure 1.2 Illustrations of the hypotheses for Study 1 and Study 3. It is hypothesized that (A)
people with PD who have striatum impairment would demonstrate excessive context-dependent
learning when compared to control participants, while (B) perturbation to the DLPFC with low
frequency rTMS in non-disabled adults would lead to a decreased context-dependent learning.

Overview
This dissertation is organized into three separate but related studies in order to address the
three specific aims. Each study (Chapter 2 to 4) is presented separately with its own introduction,
purpose, methods, results, and discussion. The first study (Chapter 2, Specific aim 1) was
designed to test the hypothesis that individuals with PD would demonstrate greater context-
dependent learning than non-disabled adults. One group of participants diagnosed with idiopathic
PD and one age- and gender-matched Control group were recruited for this study. A finger motor
sequence task was specifically designed to determine context-dependent learning for the
participants. To address Specific aim 2 (Chapter 3), we asked the question of whether there is an
association between context-dependent learning and set-shifting ability, which is a behavioral
indicator of the integrity of the frontostriatal circuit. Same groups of participants from Study 1
also completed Study 2. The participants were given an addition trail making test (TMT) to
8

assess their set-shifting ability. The relationship between context-dependent learning and set-
shifting ability in participants with PD and non-disabled adults was determined. To test the
hypothesis that the DLPFC plays a specific role in context-dependent learning and perturbing the
neuronal processing of the DLPFC in non-disabled adults would reduce context-dependent
learning (Chapter 4, Specific aim 3), we recruited two additional groups of non-disabled adults
age-matched to the participants with PD in the first study. These two groups of participants
received low frequency rTMS either over the DLPFC or the vertex before testing context-
dependency. The data obtained from these two rTMS groups was compared to the results of the
Control group from the first study to determine the role of the DLPFC in context-dependent
learning. Chapter 2, 3, 4 are in manuscript format in preparation for submission for publication.
The final chapter (Chapter 5) consists of a summary, general discussion, the limitations, clinical
implications, and future directions.

9

CHAPTER 2
CONTEXT-DEPENDENT LEARNING IN PEOPLE WITH PARKINSON’S DISEASE

Introduction
Parkinson’s disease (PD) is a progressive neurodegenerative disease characterized by motor
and cognitive dysfunction (Fritsch et al., 2012). In addition to pharmacological treatment, motor
rehabilitation has been considered an important method to improve functional ability and quality
of life for people with PD (Abbruzzese, Trompetto, & Marinelli, 2009). One proposal for an
effective rehabilitation program for people with PD includes three components: the individual,
the task, and environmental factors (Kelly, Eusterbrock, & Shumway-Cook, 2012). Numerous
studies have investigated the relationship between the individual and task factors, such as
determining the effects of treadmill training (task) on gait and balance ability (individual) (Miyai
et al., 2000; Yang, Lee, Cheng, & Wang, 2010). However, to date, no study has specifically
investigated the influence of context (environmental factor) on motor learning and performance
in PD (Kelly et al., 2012).
There is empirical evidence that practice context affects motor performance in people with
PD. Clinicians often observe that when people with PD learn a motor task (e.g., walking) in the
clinic, they appear to have difficulty transferring that performance to an alternate environment
such as their home or community. Nieuwboer et al. (2001) examined the effect of a home
physical therapy exercise program for individuals with PD. Interestingly, they observed that the
home exercise training resulted in better performance when the participants with PD were tested
in their home compared to a hospital setting (Nieuwboer et al., 2001). Finally, Onla-or and
Winstein (2008) designed a study to determine the optimal challenge point for motor learning in
10

people with PD. The results showed that individuals with PD were as capable as non-disabled
adults of learning a relatively fast and complex arm trajectory matching task. However, this
capacity was only evident when the testing context was the same as the practice context. A post
hoc explanation offered the possibility that compared to non-disabled adults, people with PD
over-rely on the practice context to carry out learned motor tasks (Onla-or & Winstein, 2008).
Context, the environmental condition(s) under which a task is performed and learned, plays
a subtle but important role in human performance (C. H. Shea & Wright, 1995). Although
context may be incidental to task performance, it has been consistently observed that an
individual can demonstrate superior performance if the testing context is the same as the practice
context, while performance may worsen if the testing and practice contexts are different (Bertsch
& Sanders, 2005; Smith & Vela, 2001). This phenomenon of better performance in the original
practice context is termed context-dependent learning. One of the most well-known empirical
examples observed in daily life is the “home field advantage,” in which athletes usually have a
better chance to win a game when it is played on the team’s own practice field (Schmidt & Lee,
2005). While reinstatement of the practice context can benefit performance (Smith & Vela, 2001;
Wright & Shea, 1991), it is also crucial for an individual to have the capability to carry out
learned motor tasks under a variety of environmental contexts (Armbruster, Ueltzhoffer, Basten,
& Fiebach, 2012), such as the opposing team’s athletic field. Consequently, over-reliance on the
environmental context for learned behaviors may impede generalization of a well-learned skill to
different contexts.
It may well be the case that this difficulty generalizing to various contexts is what has been
observed clinically for people with PD, as described earlier. Since the evidence provided so far is
either anecdotal or indirect, this study aimed to use a systematic method to investigate whether
11

individuals with PD are more context-dependent than age-matched controls for sequence
learning. We hypothesized that individuals with PD would demonstrate greater context-
dependency when compared to age-matched non-disabled participants. Determining context-
dependent learning in people with PD is clinically important because it may facilitate the
development of more efficient training programs that are independent of the environmental
context.

Methods
Participants
Ten individuals diagnosed with idiopathic PD (6 M/4 F, mean age: 64.6 ± 10.1) and ten
age- and gender-matched non-disabled adults (6 M/4 F, mean age: 64.0 ± 12.7) were recruited
into the PD group and the Control group for this study. Participants were excluded if they had
difficulty following the instructions for the task, severe hand action tremor, other neurological
disorders, or hand orthopedic problems that prohibited them from performing the task.
Characteristics of the PD group were mean Hoehn & Yahr stage 2.3 ± 0.5 during “on”
medication cycle, and mean disease duration of 5.8 ± 4.4 years. Before the study, the participants
signed an informed consent approved by the Institutional Review Board of University of
Southern California.

Finger Sequence Task
A finger sequence key pressing task was used to evaluate context-dependent learning
(Wright & Shea, 1991). Participants sat in a chair in front of a computer screen and were
instructed to put the index and middle fingers of both hands on a designated location of an
12

enlarged key-board. The finger sequence task consisted of three numerical, four element-
sequences presented on a computer screen; each sequence was embedded within different
colored circles and was in one of three different locations on the computer screen (Figure 2.1A).
Participants were instructed to press the keys in an ascending order, that is 1-2-3-4, as accurately
and as fast as possible, but to prioritize accuracy. To illustrate the task, the blue sequence
depicted in Figure 2.1A would be performed with the following order of key presses: right
index – left index – right middle – left middle fingers (Figure 2.1B). The participants were not
explicitly informed that each sequence was associated with a specific context (color and location
on the computer screen).
On day 1, all participants practiced the three sequences in a pseudorandom order for a total
of 324 practice trials (108 trials per sequence). Twenty-four hours after practice, each participant
returned to the laboratory for testing. All participants were tested on the same three numerical
sequences under two conditions: SWITCH and SAME conditions. Under the SWITCH condition,
the original sequence-context association was changed from that of practice, while the sequence-
context association remained the same as practice under the SAME condition (Figure 2.1C).
Thus, the only difference between the SWITCH and SAME conditions was the sequence-context
association. The participants were asked to perform 9 trials (3 trials of each sequence) under the
SWITCH condition first then the SAME condition.
After completion of the finger sequence task on the second day, all participants were given
a recognition test questionnaire to determine whether they had any explicit awareness of the
sequence-context associations.

13

Figure 2.1 The finger sequence task. (A) The participants practiced three sequences in a
pseudorandom order. Each sequence was embedded within a specific color and spatial location
on the computer screen. (B) Participants pressed the corresponding key button in the order of 1-
2-3-4. (C) Examples of SAME and SWITCH conditions during the second day tests. The SAME
condition was when the sequence and its associated context remained the same as practice, while
the SWITCH condition was when the sequence and its associated context changed from that of
practice.

Outcome measures and statistics
The main outcome measure was total time accuracy cost (TT
AC
), which took both speed
and accuracy into account (Aizenstein et al., 2004; Bruyer & Brysbaert, 2011). Total time (TT)
for each trial was calculated from the time of stimulus presentation to the last finger key press.
For the purpose of analysis, the data were grouped into blocks. During the acquisition phase,
14

each block consisted of 27 practice trials (9 trials of each sequence in the order of their
presentation). Mean TT during acquisition was determined as the average of 27 trials within a
block. For testing, mean TT included one block of 9 trials for the SWITCH condition and one
block of 9 trials for the SAME condition. Accurate trials were those in which all four elements of
the sequence were performed correctly. Performance accuracy was calculated as the percentage
of accurate trials per block such that 1 indicated 100% accuracy and 0 indicated no accurately
performed trials. Error rate was then the percentage of inaccurate trials per block. The TT
AC
for
each block was the mean TT normalized by performance accuracy (mean TT/ performance
accuracy). TT
AC
was calculated as a measure of performance that could be used to account for
differences in error rate between the groups. The TT
AC
was then decomposed into response time
accuracy cost (RT
AC
) and movement time accuracy cost (MT
AC
) in order to separate the planning
(RT) and execution (MT) components attributable to each group. RT was the time between the
stimulus presentation and the first key press, while MT was calculated from the first key press to
the last key press. The TT
AC
, RT
AC
, and MT
AC
during practice were analyzed separately using a 2
group (PD vs. Control) × 12 practice block repeated measures analysis of variance (ANOV A).
To evaluate TT
AC
, RT
AC
, and MT
AC
acquired during the testing phase on day 2, a 2 group
(PD vs. Control) × 3 condition (end of practice (EOP = last block of practice phase) vs. SWITCH
vs. SAME) repeated measures ANOVA was performed. The EOP data was included in this
analysis in order to compare with the SAME condition and determine whether the participants
learned the finger sequences or not.
Switch cost was calculated as an indicator of context-dependent learning. Switch cost was
the performance (TT
AC
, RT
AC
, and MT
AC
) difference between the SWITCH condition and the
SAME condition normalized by the SAME condition (100 % × [SWITCH − SAME] / SAME).
15

This normalization was used to account for expected “baseline” performance differences (i.e.,
the SAME condition) between the PD and Control groups. A larger switch cost value would
indicate greater context dependency. To evaluate group differences in switch cost, a 2 group (PD
vs. Control) × 3 outcome measure (TT
AC
vs. RT
AC
vs. MT
AC
) repeated measures ANOVA was
performed. Post hoc analyses were performed if main effects and/or interactions were observed.
All statistical analysis was performed using PASW Statistics 18.0 software (SPSS Inc., Chicago,
IL), and alpha level was set at 0.05.

Results
There were no statistically significant differences between the PD and Control groups in
error rates during practice or retention. However, it was observed that the PD group on average
made more errors than the Control group.

Practice Phase
Throughout practice, both groups improved significantly in TT
AC
(F
11,198
= 11.451, p =
0.001), RT
AC
(F
11,198
= 7.706, p = 0.008), and MT
AC
(F
11,198
= 14.404, p < 0.001). There were no
group by practice block interactions for any of the outcome measures. Both the groups seemed to
reach a plateau of performance after the 5
th
block of practice (after 135 trials). The performance
improvement for both groups can be seen in Figure 2.2. The PD group was generally slower than
the Control group throughout practice. However, in part due to the large variability between
participants, no statistically significant group main effect was found (TT
AC
: p = 0.087, RT
AC
: p =
0.132, MT
AC
: p = 0.067).

16

Figure 2.2 (A) Total time, (B) response time, and (C) movement time accuracy cost of the PD
and the Control groups throughout practice blocks and second day testing conditions.
Abbreviation: SW: SWITCH condition, SA: SAME condition.
17

Testing phase
The 2 group (PD vs. Control) by 3 condition (EOP, SWITCH, and SAME) repeated
measures ANOV A for TT
AC
, RT
AC
, and MT
AC
resulted in a significant condition main effect for
all three outcome measures (TT
AC
: F
2,36
= 9.486, p < 0.001; RT
AC
: F
2,36
= 10.264, p < 0.001;
MT
AC
: F
2,36
= 7.381, p = 0.002) with no significant group by condition interaction (TT
AC
: p =
0.150; RT
AC
: p = 0.325; MT
AC
: p = 0.095). Post hoc analysis revealed that for both groups,
performance under the SWITCH condition was significantly poorer than performance at EOP
(TT
AC
: p = 0.008; RT
AC
: p = 0.006; MT
AC
: p = 0.021) and the SAME condition (TT
AC
: p < 0.001;
RT
AC
: p = 0.001; MT
AC
: p < 0.001) with no significant difference between EOP and the SAME
condition. Additionally, there was a significant group main effect for TT
AC
(F
1,18
= 6.465, p =
0.020), RT
AC
(F
1,18
= 6.060, p = 0.024), and MT
AC
(F
1,18
= 5.805, p = 0.027) demonstrating that
the PD group performed worse than the Control group in EOP, SWITCH, and SAME conditions.

Switch Cost (Figure 2.3)
Individuals with PD demonstrated greater context-dependency than the control participants.
Once the sequence-context association was switched, the PD group showed a significantly
greater switch cost than the Control group. A 2 group (PD vs. Control) by 3 outcome measure
(TT
AC
, RT
AC
, and MT
AC
) repeated measures ANOV A revealed a significant group by outcome
measure interaction (F
2,36
= 6.361, p = 0.020) with a significant group main effect (p = 0.036) but
no outcome measure effect (p = 0.089). Post hoc analysis demonstrated that the interaction effect
was due to a significant group difference in MT
AC
switch cost (p = 0.022), but not RT
AC
switch
cost (p = 0.185).
18

Figure 2.3 Percentage switch cost of the total time, response time, and movement time accuracy
costs of the PD and the Control groups.
Abbreviation: TT_AC: total time accuracy cost, RT_AC: response time accuracy cost, MT_AC:
movement time accuracy cost.
* p < 0.05

Time Differences between the SWITCH and SAME Conditions for Each Key Press (Figure 2.4)
The performance decrement in MT
AC
but not RT
AC
invoked a hypothesis that participants
with PD required greater movement time during the SWITCH condition as a result of additional
online planning while conducting subsequent key presses. To test this hypothesis, we calculated
the amount of time spent moving between the 1
st
and 2
nd
, 2
nd
and 3
rd
, and 3
rd
and 4
th
keys, and
compared the time differences between the SWITCH and SAME conditions for both the PD and
Control groups. For this calculation, the raw time data from accurate trials were analyzed.
When compared to the Control group, the PD group spent significantly longer time under
the SWITCH condition for the 1
st
to 2
nd
key press (p = 0.019), 2
nd
to 3
rd
key press (p = 0.012),
and 3
rd
to 4
th
key press (p = 0.025). The extra time that the PD group spent for each of these
19

subsequent key presses ranged from 75-132ms, and was significantly longer than the time the
Control group spent (range 16-43ms) (Figure 2.4). This result supported our hypothesis that the
increased MT
AC
switch cost exhibited for the PD group was the result of additional online
planning for each finger movement under the SWITCH condition.

Figure 2.4 Time differences between the SWITCH and SAME conditions of each finger key
press. When the sequence-context association was changed under the SWITCH condition, the
participants in the Control group spent most of the extra time during the response time period
(1st key press). On the other hand, participants with PD demonstrated additional planning time
for each subsequent key press in order to carry out the numerical sequence accurately.
* p < 0.05

Recognition Test Questionnaire
Post study recognition test revealed that none of the participants showed explicit awareness
of the sequence-context association. These results provided support for the notion that the greater
20

context-dependency exhibited by the PD group was not due to any explicit conscious processing
of the contextual information for sequence execution.

Discussion
The goal of this study was to determine whether, compared to age-matched non-disabled
adults, individuals with PD would demonstrate greater context-dependency when learning a
finger sequence task. Our finding showed that individuals with PD exhibited a greater decrement
in performance than non-disabled adults when the sequence-context combinations established in
practice were changed in the SWITCH testing condition one day later. This novel finding
supports the idea that people with PD are more context-dependent than non-disabled adults when
learning a finger sequence task.
To our knowledge, this is the first study to use a systematic method to investigate context-
dependent learning in individuals with PD. Clinicians and family members often observe that
people with PD can skillfully perform a motor task in the same environmental context in which
the task was originally practiced, while motor performance decreases if the task is carried out in
a novel context. This clinical observation was initially thought a motor learning deficit in people
with PD (Nieuwboer, Rochester, Muncks, & Swinnen, 2009). Consistent with other studies
(Abbruzzese et al., 2009; Nieuwboer et al., 2009; Onla-or & Winstein, 2008), we demonstrated
that people with PD had the capability to learn a finger sequence task to the same degree as non-
disabled adults after extended practice. The finding of no differences between the end of practice
(EOP) performance on the first day and the SAME condition on the second day supports an
intact motor learning capability for the PD group. However, the PD group demonstrated greater
difficulty generalizing the learned motor task from the practice context to a different context,
21

even though the numerical sequences remained the same and presumably context is not essential
for task performance. This study supports the context-dependent learning observed by Onla-or
and Winstein (2008). They also showed that people with PD could demonstrate comparable
learning as non-disabled adults in a relatively complex motor task, but only when the testing
condition was the same as the original practice condition (Onla-or & Winstein, 2008). To take
into account the potential “baseline” motor capability differences between the PD and Control
groups, we calculated switch cost. Compared with the Control group, the higher switch cost for
the PD group is not the result of differences in motor capability (i.e., bradykinesia), but rather
due to a greater reliance on contextual information for execution of learned sequence.
Interestingly, this obligatory reliance on context appeared to be mediated by an implicit process,
particularly given that they were not aware of any sequence-context association during practice
or testing.
Our additional analysis of switch cost showed that when the sequence-context association
was changed, individuals with PD did not plan the whole sequence during the RT period but
engaged in additional planning while executing the sequence. This extra planning is manifested
by greater MT
AC
switch cost, in contrast to that for RT
AC
switch cost. A distinct pattern of time
allocation between the Control and the PD group was found (Figure 2.4). For the Control group,
the participants spent more time during the RT period under the SWITCH compared to the
SAME condition with not much time difference between the two testing conditions for
subsequent key presses. This result indicated that when encountering a new sequence-context
association, non-disabled adults spent extra processing time during the RT period to reselect and
reconstruct the motor sequence, and then execute the entire sequence without extra planning
(Immink & Wright, 1998). On the other hand, the PD group demonstrated a different strategy to
22

account for the new sequence-context associations. The participants with PD required additional
planning time for each finger key press under the SWITCH testing condition in order to carry out
the whole sequence correctly. This type of movement strategy is often observed when an
individual encounters a new sequence (Bo & Seidler, 2010; Immink & Wright, 2001; Verwey &
Dronkert, 1996). Therefore, it is possible that individuals with PD respond to the numerical
sequences under the SWITCH condition as if they were different sequences.
Context-dependent behavior observed in the PD group could be the result of an impaired
striatum, which functions as a gate or modulator to select and filter task relevant information
(Dominey & Boussaoud, 1997; Wise et al., 1996). It is possible that as a result of an impaired
“filter,” all the information associated with a motor task is considered to be essential for
movement execution (Zgaljardic, Borod, Foldi, & Mattis, 2003). Thus, if even non-essential
information is missing, people with PD may demonstrate a decrement in motor performance. An
alternative explanation is that people with PD are more susceptible to interference of “irrelevant”
contextual information (van Eimeren, Monchi, Ballanger, & Strafella, 2009). Therefore,
changing the context associated with the numerical sequence, as done here, may interfere with
the execution of finger movements leading to a greater switch cost.
It is important to note that the results of this study may not be applicable to all patients with
PD. Only 10 individuals with PD and with varied disease severity, disease duration, and
medication usage, participated. Despite the heterogeneity of the participants, we still observed a
context effect, particularly in MT
AC
, suggesting that over-reliance on context may be a problem
inherent in PD. Future studies can aim to recruit a larger sample of participants and stratify by
disease severity to establish the relationship between context-dependent learning and disease
progression.
23

Clinical Implications/ Conclusions
This study, we believe, has important clinical implications for PD rehabilitation. First, our
findings point to the importance of the training/learning environment for people with PD (Kelly
et al., 2012; Nieuwboer et al., 2009). Although context may be incidental to task performance,
individuals with PD may implicitly utilize the information as a “cue” for performance of the
learned motor task. Therefore, when designing a rehabilitation program for PD, clinicians should
consider that the environment plays a meaningful part in task performance (Nieuwboer et al.,
2009). For example, if the patient’s goal is to complete activities of daily living at home, then
training the patient in his/her home environment is likely the optimal plan. On the other hand, for
an individual with PD to function at a high level regardless of environment, clinicians maybe
should challenge and train the person with the same motor tasks in a variety of settings (Smith,
1982, 1984).
Another important issue to consider is the commonly used “cueing strategies” for training
individuals with PD. Visual and auditory cues are very effective for improving motor
performance, such as gait and balance (Baker et al., 2007; Lim et al., 2010). However, once the
cues are removed, motor performance significantly decreases (Nieuwboer et al., 2009; Rochester
et al., 2007). This observation, along with the findings from our study, implies that people with
PD may over-rely on the visual/auditory cues and thus have difficulty performing a motor task
once the cues are no longer available. Therefore, when using external cues to train patients with
PD, clinicians could incorporate a faded feedback training strategy to decrease over-reliance on
the cues (Winstein, 1991). For example, when training walking with the use of visual strips
placed on the floor, the number and frequency of strips could be gradually tapered off to help the
24

patient learn walking while reducing reliance on the visual cues.
In conclusion, this study demonstrated that individuals with PD are more context-
dependent than age-matched controls when learning a finger sequence task. As stated above,
future studies are needed to determine context-dependent learning across disease severity in
people with PD. In addition, future studies are needed to understand whether context-dependent
learning of a finger sequence task generalizes to other more complex functional tasks. Finally,
unraveling the potential neural mechanisms associated with context-dependent behavior in PD
may facilitate clinicians to design more optimal training strategies for this patient population in
the future.

25

CHAPTER 3
RELATIONSHIP BETWEEN CONTEXT-DEPENDENT LEARNING AND SET-
SHIFTING ABILITY

Introduction
When driving on a familiar route, such as driving from home to the work place, an
individual can listen to the news or sing along with the music while maintaining the criterion
driving speed. However, if the individual is required to drive in a new city or take an unfamiliar
route to work, the driving speed may decrease and he/she might disregard the information from
the radio. This is a daily example of context-dependent learning, a phenomenon in which people
demonstrate superior performance in the context where they originally learned a motor task and
conversely, do not perform as well if the task is carried out in a novel context (Smith & Vela,
2001; Wright & Shea, 1991). Although motor performance may decrease slightly in an
unfamiliar environment, an individual has the ability to carry out a learned motor task in various
contexts; for instance, an individual can still drive in a new city. It could become problematic if
an individual over-relies on the contextual information acquired during practice such that his/her
performance dramatically decreases when the familiar context is no longer available. This over-
reliance on the practice context would limit the capability to adapt a learned motor task to a new
context.
Difficulty in generalizing a learned motor task to various contexts has been observed in the
clinical setting in people with Parkinson’s disease (PD). In our first study, individuals with PD
demonstrated greater context-dependency while learning a motor sequence task when compared
to non-disabled adults. Interestingly, individuals with PD appeared to implicitly utilize the
26

contextual information to guide their movement (Chapter 2). The functional neural mechanism
associated with this observed behavior is not well understood. Therefore, the goal of this study
was to indirectly investigate the neural network potentially associated with context-dependent
behavior observed in PD.
Our first study supported the hypothesis that the impairment of the dorsal striatum in
people with PD may result in context-dependent behavior (Chapter 2). The striatum has
numerous neuronal connections to other cortical areas forming the cortico-striatal circuits. One
of these circuits that may plausibly mediate context-dependent learning is the frontostriatal
circuit, which is the recurrent neural connection between the dorsolateral prefrontal cortex
(DLPFC) and the dorsal striatum (Alexander et al., 1986). Single cell recording in animals and
computational models demonstrated neuronal activations in the DLPFC and the dorsal striatum
during context encoding and action selection (Dominey & Boussaoud, 1997; Wise et al., 1996).
Electroencephalograph (EEG) recordings showed that different from non-disabled adults,
patients with lateral prefrontal lesions failed to generate a robust P3b signal to a predictive
contextual cue, indicating that patients with prefrontal lesions were unable to utilize the
contextual information to the same extent as non-disabled adults (Fogelson et al., 2009). On the
other hand, the amplitude of P3b signal was found to be larger in people with PD when
compared to non-disabled adults while processing contextual information (e.g. predictive visual
stimuli), suggesting that people with PD may need to allocate more attention to the contextual
information due to the impaired striatum (Fogelson, Fernandez-del-Olmo, & Santos-Garcia,
2011). The above evidence provides beginning evidence that context-dependent behavior may be
related to the integrity of the frontostriatal circuit.
In addition to the potential role in context-dependent learning, the frontostriatal circuit has
27

been functionally identified as subserving numerous cognitive and executive functions
(Alexander et al., 1986; Doyon, Penhune, & Ungerleider, 2003), such as motor learning, action
selection, and set-shifting (Monchi et al., 2007; Monchi et al., 2001; Packard & Knowlton, 2002).
Disruption of the frontostriatal circuit (e.g., PD) may lead to impairment of these cognitive and
executive functions. One of the most commonly observed cognitive impairments in people with
PD is a set-shifting deficit (Monchi et al., 2007; Owen et al., 1992; Packard & Knowlton, 2002).
Set-shifting is the ability of an individual to shift attention and switch between movements based
on changes in movement goals and/or environmental stimuli (e.g., shifting from focusing on the
color to the shape of an object) (Flowers & Robertson, 1985). Neuroimaging studies have
consistently demonstrated that set-shifting ability is mediated by the co-activation of the DLPFC
and the dorsal striatum (Monchi et al., 2004; Monchi et al., 2001; Zakzanis, Mraz, & Graham,
2005). With impairments in the dorsal striatum, people with PD demonstrate set-shifting deficits
even in the early stages of the disease. Moreover, the set-shifting deficit appears to be associated
with disease severity (Owen et al., 1992), suggesting that set-shifting ability could be used as a
behavioral indicator of the frontostriatal circuit integrity.
The goal of this study was to determine whether the frontostriatal circuit is a neural
network associated with context-dependent learning. Since it is well known that set-shifting
ability is mediated by the frontostriatal circuit, we investigated the relationship between set-
shifting ability and context-dependent learning in both people with PD and non-disabled adults.
We expected that a measure of context-dependent learning would be positively correlated with a
measure of set-shifting ability in people with PD.

Methods
Participants
28

Seven individuals with PD and 9 age- and gender-matched non-disabled adults completed
this study. Those participants were the same participants from Study 1 (Chapter 2). Before the
study, the participants signed an informed consent approved by the Institutional Review Board of
University of Southern California. Participants were excluded if they had difficulty following the
instructions, severe hand action tremor, or hand orthopedic problems that prohibited them from
performing the task.

Tasks
Finger motor sequence task
As described in Study 1 (Chapter 2), a modified finger sequence key pressing task was
used to measure context-dependent learning (Wright & Shea, 1991). The participants used the
index and middle fingers of both hands to perform three numerical sequences on an enlarged
keyboard. The participants were instructed to press the keys in the order of 1-2-3-4 according to
the corresponding location of their fingers (Figure 2.1B). In addition, the participants were asked
to perform the key presses as accurately and as fast as possible, while prioritizing accuracy.
Unknown to the participants, each numerical sequence was embedded within a colored circle and
a specific location on the computer screen throughout practice (Figure 2.1A). These 3 sequences
were presented in a pseudorandom order on the first day of practice. Approximately twenty-four
hours after practice, the participants returned to the laboratory for the SWITCH and the SAME
testing conditions. In the SWITCH condition, the original sequence-context associations were
changed from those in practice, while the sequence-context association remained the same as that
of practice under the SAME condition (Figure 2.1C).

29

Set-shifting task: the trail making test (TMT)
After completion of the finger sequence task, set-shifting ability was tested on all
participants using the trail making test (TMT). The TMT consisted of two parts: Part A (TMT-A)
and Part B (TMT-B). In Part A, participants were presented with an array of numbered circles
and were required to draw a line connecting consecutive numbers (i.e., 1-2-3-4-5…, etc.). In Part
B, the participants were required to connect the circles with alternating numbers and letters in a
sequential order (i.e., 1-A-2-B-3-C…, etc.) (Higginson, Lanni, Sigvardt, & Disbrow, 2013). The
participants were asked to connect the circles as fast as possible, and the time to complete each
part was recorded. The time ratio between Part A and Part B (TMT-B/ TMT-A) is a quantitative
indicator of set-shifting ability (Arbuthnott & Frank, 2000).

Outcome measures and statistics
The main outcome measure for the finger sequence task was total time accuracy cost
(TT
AC
). Total time (TT) was the time from stimulus presentation to the last finger key press, and
an averaged TT value was calculated for each block (27 practice trials per block). Performance
accuracy was the percentage of accurate trials of each block (number of correct trials/ 27 trials
per block). The TT
AC
was calculated as the averaged TT divided by performance accuracy. To
evaluate context-dependent learning, the switch cost was calculated. Switch cost was the TT
AC

performance difference between the SWITCH condition and the SAME condition normalized by
the SAME condition (100% × [SWITCH - SAME] / SAME). A larger switch cost value indicates
greater context-dependency.
For the TMT, the time for an individual to complete Part A and Part B of the test was
recorded, and the time ratio (time to complete TMT-B/ time to complete TMT-A) was calculated
30

to indicate the set-shifting ability (Arbuthnott & Frank, 2000). Due to the small sample size in
each group, we performed Spearman correlation using the PASW 18.0 statistics software (SPSS
Inc., Chicago, IL) to examine the relationship between the TMT-B/TMT-A ratio and the TT
AC

switch cost.

Results
The mean values of TT
AC
switch cost and TMT-B/TMT-A ratio for both the PD and Control
groups are presented in Table 3.1. Spearman correlation results showed that the set-shifting
ability (as demonstrated by the TMT-B/TMT-A ratio) was strongly correlated with TT
AC
switch
cost (r = 0.86, p = 0.014) in people with PD (Figure 3.1). Contrary to the results for the PD group,
the Spearman correlation results for the non-disabled participants showed no relationship
between the TMT-B/TMT-A ratio and the TT
AC
switch cost (r = 0.07, p = 0.865).

Table 3.1 Mean values of the outcomes measures for the Parkinson’s disease (PD) and Control
groups.
PD Group Control Group
TT
AC
Switch Cost (%) 41.38 ± 38.99 21.32 ± 9.47
TMT-B/ TMT-A Ratio 2.16 ± 0.60 1.62 ± 0.41
Data is presented in mean ± standard deviation.
Abbreviations: TT
AC
: Total time accuracy cost, TMT-B: Part B of the trail making test, TMT-A:
Part A of the trail making test.

31

Figure 3.1 Spearman correlation between result of the trail making test (TMT) and the switch
cost of total time accuracy cost (TT
AC
) in (A) individuals with Parkinson’s disease (PD), and (B)
non-disabled adults. The number next to each data point in the PD group represents the Hoehn
and Yahr Stage score for each individual participant.

Discussion
The main finding of this study was that set-shifting ability, as measured by the Trail
Making Test (TMT) was highly correlated with the TT
AC
switch cost in people with PD.
However, no correlation was found between the set-shifting ability and the TT
AC
switch cost for
the non-disabled adults (Figure 3.1).
Set-shifting is the ability for an individual to quickly shift attention from one dimension to
another based on the changes of environmental stimuli (Flowers & Robertson, 1985; Owen et al.,
1991). The ability to shift sets requires an individual to inhibit the current movement and quickly
gather related information to carry out the next movement; it has been well established that this
32

ability is mediated by the frontostriatal circuit (Monchi et al., 2007; Owen et al., 1992). In
healthy young adults, neuroimaging studies have shown increased cortical activity in bilateral
DLPFC, caudate nucleus, and the dorsal thalamus while performing the Wisconsin Card Sorting
Task, a test of set-shifting ability (Monchi, 2001). Compared to non-disabled adults, people with
PD were found to demonstrate altered cortical activation while performing the same set-shifting
task (Monchi et al., 2007). When the set-shifting ability was measured behaviorally using either
the Wisconsin Card Sorting Task or the Trail Making Test, individuals with PD were slower and
made more errors than non-disabled adults (Cools et al., 2001; Kierzynka, Kazmierski, &
Kozubski, 2011; Monchi et al., 2007; Owen et al., 1992). Consistent with the results from the
above referenced studies, the findings from this study also showed a higher TMT-B/TMT-A ratio
in the PD group compared to the Control group (Table 3.1), indicating that participants with PD
demonstrated greater difficulty performing the set-shifting task.
The results showed a positive relationship between the results of the TMT (TMT-B/TMT-A
ratio) and the TT
AC
switch cost in people with PD. As the TMT reflects set-shifting ability and is
known to be mediated by the frontostriatal circuit (Higginson et al., 2013; Monchi et al., 2004;
Monchi et al., 2001; Zakzanis et al., 2005), it stands to reason that the positive relationship we
observed between set-shifting and context-dependent learning in people with PD suggests that
context-dependent behavior may also be mediated by the frontostriatal circuit. It has been found
that the set-shifting deficit is associated with disease severity (Owen, 1992). Therefore, it is also
possible that individuals with PD at more severe stages of the disease would also demonstrate
greater context-dependency. Due to the small sample size of this study, we are unable to
determine the relationship between the degree of context-dependent learning and disease severity.
In contrast to the results of the PD group, there was no relationship between the TMT-
33

B/TMT-A ratio and the TT
AC
switch cost in the Control group (Figure 3.1B), indicating that set-
shifting is not associated with context-dependent behavior in non-disabled adults. This distinct
relationship observed in the PD and Control groups could be related to the integrity of the
frontostriatal circuit. The frontostriatal circuit is presumably intact in non-disabled adults; thus,
they have both a functional DLPFC and dorsal striatum to perform tasks that require the ability
to shift between sets or to process contextual information. Since the non-disabled participants
have sufficient neural resources to perform the cognitive tasks in this study, the observed
individual performance differences in the Control group could be due to other confounding
factors that are not specifically related to the function of the frontostriatal circuit (Keys & White,
2000). Therefore, no correlation was found between the result of the TMT and the TT
AC
switch
cost in non-disabled adults. In individuals with PD, due to impairment of the striatum, tasks that
require shifting between sets or processing contextual information may greatly depend on the
function of the DLPFC. Thus, individuals with PD who demonstrate greater difficulty
performing the set-shifting task also demonstrate greater context-dependency during motor
learning. The findings from individuals with PD and non-disabled adults altogether suggest that
excessive context-dependent learning is associated with the disrupted frontostriatal circuit.
A major limitation of this study was the small number of participants in each group. In
addition, we did not obtain other behavioral tests that are also mediated by the frontostriatal
circuit, such as the Stroop test or working memory test, to assess their relationship with context-
dependent behavior. Future studies should include a larger number of participants and also obtain
other cognitive and executive function tests to assess this relationship.
In conclusion, this study demonstrated a high correlation between set-shifting ability and
context-dependent behavior in individuals with PD, but not in non-disabled adults. These results
34

suggest that excessive context-dependent learning in people with PD is related to the impaired
integrity of the frontostriatal circuit. To further test the hypothesis that excessive context-
dependent learning in people with PD is a result of impaired frontostriatal circuit, one logical
next step is to use repetitive transcranial magnetic stimulation to perturb the neuronal processing
of the DLPFC, another important neural substrate within the frontostriatal circuit, in non-
disabled adults and investigate its impact on context-dependent behavior.
35

CHAPTER 4
ROLE OF THE DORSOLATERAL PREFRONTAL CORTEX IN CONTEXT-
DEPENDENT LEARNING

Introduction
Incidental context, the environmental condition(s) in which a task is performed and learned,
contains information that is associated with a task and can later facilitate memory retrieval (Chun
& Jiang, 1998). It has often been observed that people demonstrate superior motor performance
when the incidental context is the same as the original practice context; conversely, performance
may decrease if the task is carried out in a different context (Smith & Vela, 2001; Wright & Shea,
1991). One well-known empirical example is the “home field advantage,” in which athletes often
perform best when competing on their own practice field (Schmidt & Lee, 2005). Better
performance under the same practice context compared to poorer performance in a different
context is termed context-dependent learning.
While context-dependent learning during motor skill acquisition has been widely
demonstrated in healthy young adults (Keetch et al., 2008; Ruitenberg, De Kleine, Van der
Lubbe, Verwey, & Abrahamse, 2012; Wright & Shea, 1991), the neural substrates that mediate
this phenomenon are not well understood. Based on its known role in processing contextual
information, the frontostriatal circuit is a plausible candidate to mediate context-dependent
learning (Dominey & Boussaoud, 1997; Nee & Brown, 2013). The frontostriatal circuit consists
of the recurrent neural connections between the dorsolateral prefrontal cortex (DLPFC) and the
dorsal striatum (Alexander et al., 1986). Both the DLPFC and the dorsal striatum play specific
roles in processing contextual information in order to select the most appropriate movement
action (Dominey & Jeannerod, 1997; Wise et al., 1996). The DLPFC is especially important for
36

encoding and updating contextual information received from other sensorimotor cortical areas,
and maintaining the information in working memory (Dominey & Boussaoud, 1997; Reynolds,
O'Reilly, Cohen, & Braver, 2012; Wise et al., 1996). The dorsal striatum mainly functions as a
gate or a modulator to filter and integrate task relevant information in order to select the most
appropriate movement responses according to the environmental stimuli (Atallah, Frank, &
O'Reilly, 2004; Disbrow et al., 2013; Dominey & Boussaoud, 1997; Frank, 2011; Gluck,
Mercado, & Myers, 2008). While overly simplified, the following real life example may
illustrate the roles of the DLPFC and the striatum. When driving a car on a street, the DLPFC
receives and encodes various sensory stimuli, such as the traffic lights and signs, stores along the
streets, pedestrians walking on the sidewalk, and etc. The striatum selects the relevant
information that is important for driving (e.g., the traffic lights and signs) and filters out the
incidental stimuli (e.g., stores and pedestrians along the street). Therefore, a neuronal interaction
between the DLPFC and the dorsal striatum is necessary to maintain a stable and flexible goal-
directed behavior (Dominey & Boussaoud, 1997; Fogelson et al., 2011; Nee & Brown, 2013).
We hypothesized that when one of the critical nodes of the frontostriatal circuit is impaired,
such as the striatum impairment in Parkinson’s disease (PD), the neuronal interaction between
the DLPFC and the dorsal striatum is disrupted; thus, impacts the ability to select the most
appropriate action according to environmental context (Fogelson et al., 2009; E. Y . Lee et al.,
2010). In our previous study, we investigated context-dependent learning in people with PD and
age-matched non-disabled adults. The participants practiced three finger sequences embedded
within specific incidental context (color and location on the computer screen). At testing, it was
found that compared to non-disabled adults, individuals with PD demonstrated a greater
decrement in motor performance compared with the Control group when the incidental context
37

associated with the sequence was different from the original practice context (Study 1, Chapter
2). This finding suggested that context-dependent learning is mediated to some extent by the
dorsal striatum. To better understand the neuronal interaction between the two putative nodes of
the frontostriatal circuit, the role of the DLPFC in context-dependent learning is yet to be
determined.
The aim of this study was to investigate the role of the DLPFC in context-dependent
learning. Based on the proposed function of the DLPFC in encoding context and maintaining the
information in working memory for movement execution (Nee & Brown, 2013), we
hypothesized that a temporary perturbation to the DLPFC will reduce context encoding. This in
turn suggests that the individual would be less affected by the change of context and demonstrate
a decreased context-dependent learning. The neuronal processing of the DLPFC was temporarily
perturbed in non-disabled adults using low frequency repetitive transcranial magnetic stimulation
(rTMS). Previous studies have applied low frequency rTMS over primary motor cortex and
found a decrease in neuronal excitability as measured by a reduction in the size of motor evoked
potential (Hallett, 2007; S. Kantak, Fisher, Sullivan, & Winstein, 2010; Siebner & Rothwell,
2003). Therefore, one assumption of this study was that low frequency rTMS would reduce
neuronal excitability when stimulated over the DLPFC (S. S. Kantak, Sullivan, Fisher, Knowlton,
& Winstein, 2010; Pascual-Leone, Wassermann, Grafman, & Hallett, 1996).

Methods
Participants
A total of 30 non-disabled adults (age-matched to the participants with PD from our
previous study) participated in this study. Twenty participants were recruited into the rTMS
38

DLPFC group (n = 10) and the rTMS Vertex group (n = 10). Ten participants in the Control
group were the same participants from a previous study comparing context-dependent learning in
individuals with PD and non-disabled adults (Study 1, Chapter 2). Before the study, the
participants signed an informed consent approved by the Institutional Review Board of
University of Southern California. Participants were excluded if they had hand orthopedic
problems that prohibited them from performing the task, a history of epilepsy, a history of
neurological disorder, or metal implanted in their body.

Finger Sequence Task
The finger sequence task was used to evaluate context-dependent learning (Study 1,
Chapter 2). The task consisted of three numerical sequences presented on a computer screen.
Each sequence was associated with a specific color and a location on the computer screen
(Figure 2.1A); however, this specific sequence-context association was not explicitly told to the
participants. The participants were instructed, using their index and middle fingers of both hands,
to make finger key presses in the order of 1-2-3-4 as accurately and as fast as possible (Figure
2.1B). The participants practiced the three finger sequences in a pseudorandom order for a total
of 324 trials on the first day. Twenty-four hours after practice, the participants returned to the lab
and performed the finger sequence task under 2 testing conditions: the SWITCH and the SAME
conditions. In the SWITCH condition, the original sequence-context association was changed
from that of practice; while in the SAME condition, the sequence-context association remained
the same as that of practice (Figure 2.1C). There were 9 trials (3 trials of each sequence) under
each testing condition.

39

Repetitive transcranial magnetic stimulation (rTMS) procedure
On the second day before the finger sequence testing conditions, the DLPFC group and the
Vertex group underwent the rTMS procedure. Repetitive TMS was delivered through a 70-mm
diameter figure-of-eight coil attached to a Rapid
2
Magstim stimulator (Rapid
2
stimulator;
Magstim, Whitland, UK). For all participants who received rTMS, the cortical representation
area (“hotspot”) of the right first dorsal interosseous (FDI) and its motor threshold were
determined over the left hemisphere. The hotspot of FDI was identified as the spot in which the
largest and most consistent motor evoked potential (MEP) amplitudes could be elicited, while
motor threshold was defined as the lowest TMS intensity that is required to elicit 50μV MEP
amplitude at least 5 times out of 10 trials (Siebner & Rothwell, 2003). We were able to obtain
T1-weighted structural brain scan for four participants in the DLPFC group. With these
participants, the location of left DLPFC was determined as the middle third of the middle frontal
gyrus by using the stereotaxic frameless neuronavigation system (Brainsight, Rogue Research,
Montreal, Canada) based on each individual’s brain scan (Cohen & Robertson, 2011). When
determined with a cap grid, the averaged stimulation locations of these 4 participants were 5.5
cm anterior to the FDI hotspot. For the other six participants in the DLPFC group who did not
have an individual brain scan, rTMS was applied 5.5 cm anterior to the FDI hotspot (Bradfield,
Reutens, Chen, & Wood, 2012; Knoch, Gianotti, et al., 2006; Michael et al., 2003). To stimulate
the DLPFC with rTMS, the coil was held tangential to the scalp and the coil-handle oriented 45
degrees from the mid-sagittal plane (Cohen & Robertson, 2011; S. S. Kantak et al., 2010). The
rTMS Vertex group was recruited to serve as a TMS control group. Vertex was identified as the
intersection between the midpoint of the left and right tragi and the midpoint of the nasion and
inion (Beam, Borckardt, Reeves, & George, 2009). The coil was held tangential to the scalp and
40

parallel to the mid-sagittal plane. Applying rTMS over vertex was to determine that (1) the
DLPFC is a specific site related to context-dependent learning, and (2) the behavioral changes
observed in the DLPFC group is not due to random noise introduced by TMS. The stimulation
protocol for rTMS was at a frequency of 1Hz and an intensity of 90% resting motor threshold for
20 minutes (a total of 1200 pulses) (Bruckner, Kiefer, & Kammer, 2013; S. Kantak et al., 2010;
van den Heuvel, Van Gorsel, Veltman, & Van Der Werf, 2013).

Experiment Design
The study was conducted over two consecutive days approximately 24 hours apart. On the
first day, after signing the informed consent, participants practiced the three finger sequences for
a total of 324 trials (108 trials/ per sequence). On the second day, the DLPFC group and the
Vertex group first underwent the rTMS procedure. Immediately after rTMS, the two groups were
tested under the SWITCH and SAME testing conditions. Previous studies suggested that
contextual information mainly affects motor performance either during the encoding or the
retrieval phase of learning (Kimbrough, Wright, & Shea, 2001; Parnell, Grasby, & Talk, 2012).
Since the DLPFC also plays an important role in acquiring the numerical sequences during the
early phase of sequence learning (Poldrack et al., 2005), such as the first day of practice, we
chose not to perturb this encoding period. Therefore, the rTMS procedure was applied
immediately before the two testing conditions on the second day to affect the retrieval phase. The
participants in the Control group did not receive rTMS, so they were tested with the two testing
conditions immediately on the second day (Figure 4.1).
41

Figure 4.1 Study procedure. All participants in the three groups practiced the finger sequence
task on the first day. On the second day, the participants were tested under the SWITCH and
SAME testing conditions (see Methods section for detailed description). The participants in the
DLPFC and Vertex groups received low frequency repetitive transcranial magnetic stimulation
(rTMS) before the two testing conditions, while the participants in the Control group did not
receive rTMS.

Outcome measures and statistics
The primary outcome measure was the time to complete the finger sequence task. In order
to account for any accuracy differences among groups, total time accuracy cost (TT
AC
) was
obtained (Aizenstein et al., 2004; Bruyer & Brysbaert, 2011). Total time (TT) of each trial was
the time from stimulus presentation to the subject’s completion of last finger key press. For data
analysis purpose, mean TT of every 27 practice trials (9 trials per sequence) was calculated as
one practice block. Performance accuracy was calculated as the percentage of accurate trials
within a block with a number of 1 as indicative of 100% accuracy. As such, TT
AC
was the mean
TT of each practice block normalized by the performance accuracy. The TT
AC
was further
decomposed into response time accuracy cost (RT
AC
) and movement time accuracy cost (MT
AC
)
in order to determine the processing differences during the planning and execution period,
42

respectively. The RT was calculated from the time of stimulus presentation to the first key press,
while the MT was the time between the first and the last key press. The TT
AC
, RT
AC
, and MT
AC

during practice were analyzed using a 3 group (Control vs. DLPFC vs. Vertex) × 12 practice
block repeated measures analysis of variance (ANOV A).
To test the hypothesis that the DLPFC group is less context-dependent than the other two
control groups, a 3 group (Control vs. DLPFC vs. Vertex) × 2 testing condition (SWITCH vs.
SAME) repeated measures ANOV A was performed. In addition, switch cost was calculated to
indicate context-dependent learning. Switch cost was the performance (TT
AC
, RT
AC
, and MT
AC
)
difference between the SWITCH condition and the SAME condition normalized by the SAME
condition (100% × [SWITCH-SAME] / SAME). A larger switch cost value would indicate
greater context dependency. The group differences in switch cost were analyzed with one way
ANOV A for TT
AC
, RT
AC
, and MT
AC
. Tukey or Games-Howell post-hoc analysis was performed
if a group main effect was found. All data was analyzed using PASW 18.0 statistic software
(SPSS Inc., Chicago, IL), and the significance level was set at 0.05.

Results
Practice Phase
The Control, DLPFC, and Vertex groups achieved a similar performance level by the end of
practice (Figure 4.2). Repeated measures ANOV A showed that during the acquisition phase, all
three groups improved significantly in TT
AC
(F
11,297
= 9.770, p = 0.003) with no significant
group by practice block interaction (F
22,297
= 1.312, p = 0.286) or group main effect (F
2,27
=
0.462, p = 0.635). Similarly, all groups improved significantly throughout practice in RT
AC

(F
11,297
= 6.390, p = 0.015) and MT
AC
(F
11,297
= 14.186, p < 0.001) with no significant group by
43

practice block interactions (RT
AC
: F
22,297
= 1.215, p = 0.199, MT
AC
: F
22,297
= 1.239, p = 0.214) or
group main effect (RT
AC
: F
2,27
= 0.583, p = 0.565, MT
AC
: F
2,27
= 0.339, p = 0.716).

44

Figure 4.2 (A) Total time, (B) response time, and (C) movement time accuracy cost results of the
Control group (open circle), the DLPFC group (closed square), and the Vertex group (open
square) throughout the first day practice blocks. All three groups improved throughout practice
and achieved a similar performance level by the end of practice.

Testing Phase
The DLPFC group showed less decrement in motor performance than the Control and
Vertex groups under the SWITCH condition. The 3 group (Control, DLPFC, vs. Vertex) by 2
testing condition (SWITCH vs. SAME) repeated measures ANOV A of TT
AC
revealed a
significant condition main effect (F
1,27
= 72.01, p < 0.001) and a significant group by condition
interaction (F
2,27
= 3.705, p = 0.038) with no significant group main effect (F
2,27
= 0.028, p =
0.972). The condition main effect was that all three groups performed worse in the SWITCH
testing condition. The group by condition interaction was attributed to less decrement in TT
AC

performance under the SWITCH condition for the DLPFC group compared to the Control and
the Vertex groups (Figure 4.3A). Similar to the results of TT
AC
, a significant condition main
effect was found for both RT
AC
(F
1,27
= 31.445, p < 0.001) and MT
AC
(F
1,27
= 31.266, p < 0.001).
However, no significant group main effect or group by condition interaction was found with
these two outcome measures (Figure 4.3B and 4.3C).
45

Figure 4.3 Results of the motor performance under the SWITCH and SAME testing conditions in
(A) total time, (B) response time, and (C) movement time accuracy cost. End of practice (EOP)
data was presented here for a better visualization of the results demonstrating that the DLPFC
group showed less decrement under the SWITCH condition compared to the Control and Vertex
groups.
46

Switch Cost
A smaller switch cost was found for the DLPFC group compared to the Control and the
Vertex groups. One way ANOVA revealed a significant group differences in TT
AC
switch cost
(F
2,27
= 4.945, p = 0.023) and RT
AC
switch cost (F
2,27
= 3.866, p = 0.033) with no significant
difference in MT
AC
switch cost (F
2,27
= 2.499, p = 0.101) (Figure 4.4). Post hoc analysis revealed
that compared to the Control group, the DLPFC group showed a significantly smaller TT
AC

switch cost (p = 0.023) and RT
AC
switch cost (p = 0.036). Although the average values were
smaller, the difference was not reliable between the DLPFC group and the Vertex group in TT
AC
switch cost (p = 0.077) or RT
AC
switch cost (p = 0.109). There was also no statistical significance
in any of the switch cost values between the Control group and the Vertex group.

Recognition Test Questionnaire
Following the two testing conditions, all participants were given a recognition test
questionnaire to determine their explicit knowledge of the specific sequence-context associations
throughout practice and testing. The questionnaire revealed that none of the participants were
explicitly aware of the sequence-context association, which implied that the sequence-context
associations were acquired and utilized through implicit processes.

47

Figure 4.4 Percentage switch cost of the total time accuracy cost (TT_AC), response time
accuracy cost (RT_AC), and movement time accuracy cost (MT_AC) of the Control, DLPFC,
and Vertex groups. The DLPFC group demonstrates a smaller switch cost than the other two
groups. * p < 0.05
48

Discussion
The aim of this study was to determine the role of the DLPFC in context-dependent
learning. We used low frequency rTMS to temporarily perturb the neuronal processing of the
DLPFC in non-disabled adults and examine its impact on context-dependent behavior. The
results showed that perturbation to the DLPFC with rTMS in non-disabled adults reduced
context-dependent learning when compared to the no rTMS Control group.
This finding supports the function of the DLPFC in contextual information processing
proposed by previous studies. Before movement execution, the frontostriatal circuit is thought to
process contextual information to facilitate the selection of the most appropriate motor responses
based on the environmental context (Dominey & Boussaoud, 1997; Fogelson & Fernandez-Del-
Olmo, 2013). While the DLPFC encodes contextual information associated with a task and
maintains that information in working memory, the dorsal striatum filters and selects task
relevant information (Dominey & Boussaoud, 1997; Fogelson et al., 2013; Nee & Brown, 2013;
Wise et al., 1996). Therefore, neuronal interaction between the DLPFC and the dorsal striatum is
crucial for mediating context-dependent behavior. The results of this study showed that
decreased activation of the DLPFC led to smaller switch costs than the Control and the Vertex
groups, indicating that the participants in the DLPFC group were less context-dependent. This
finding supports that the DLPFC, with its known function in encoding and maintaining new
contextual information (Dominey & Boussaoud, 1997; Nee & Brown, 2013), plays a role in
context-dependent learning. When the DLPFC was perturbed with rTMS, the new context may
be “overlooked,” inducing less interference with the motor performance of the numerical
sequences.
This account corresponds with the “context-dependent filtering” hypothesis proposed by
49

Ruitenberg et al. (2012) to explain context-dependent learning. They proposed that during the
early phase of learning, incidental context may interfere with optimal movement performance.
With extended practice, people learn to filter out the non-essential incidental context in order to
deal effectively with the interference. However, when context is changed, the “filter” may no
longer be effective and an individual’s motor performance may decrease due to the interference
of the new context (Ruitenberg et al., 2012). Based on this hypothesis, it is possible that when
the new sequence-context association was introduced under the SWITCH condition in this study,
the new context interfered with the original sequence-context association performance leading to
a decrement in motor performance. Since the DLPFC plays an important role in encoding
contextual information (Dominey & Boussaoud, 1997; Reynolds et al., 2012; Wise et al., 1996),
the interference from the new context is probably detected by the DLPFC (Fuster, 2008). When
the DLPFC was perturbed with rTMS, the encoding processes for the new context may be
disrupted. As a consequence, the participants who received rTMS over the DLPFC were less
affected by the changes of context when tested under the SWITCH condition leading to a smaller
switch cost (Figure 4.4).
The findings of this study enable us to hypothesize the potential neural mechanism
associated with excessive context-dependent learning observed in PD (Study 1, Chapter 2). The
current study demonstrated that low frequency rTMS over the DLPFC in non-disabled adults
reduced context-dependency. In combination with the findings from our previous PD study, we
believe that the DLPFC may be over-activated in individuals with PD in order to compensate for
the disrupted dorsal striatum. Neuroimaging studies have shown that the frontostriatal circuit,
including both the DLPFC and the striatum, is significantly activated when learning a finger
sequence task (Fogelson & Fernandez-Del-Olmo, 2013; Galea, Albert, Ditye, & Miall, 2010).
50

Functional magnetic resonance imaging (fMRI) studies conducted in individuals with PD
learning a finger sequence task demonstrated an increase in neural activation of the DLPFC. The
authors suggested that when the striatum is disrupted, as seen in PD, a compensatory increase in
neural activation of the network specialized for sequence learning may be necessary in order to
achieve a similar behavioral performance level as non-disabled adults (Mallol et al., 2007;
Mentis et al., 2003). It follows that this over-activation of the DLPFC could lead to excessive
context-dependency in PD. Without the modulation of the dorsal striatum, the DLPFC may
encode and consider all contextual information to be essential for motor learning and
performance (Dominey & Boussaoud, 1997; Nee & Brown, 2013). When a piece of information
is no longer available or changed, the DLPFC may perceive the task as different and try to
reprogram the sequence (Fuster, 2008). Reprogramming the sequence requires longer processing
time leading to a greater RT
AC
and MT
AC
switch costs as observed in the PD group from our
previous study (Study 1, Chapter 2).
One interesting observation of this study is that there were no differences in any of the
outcome measures between the Control and the Vertex groups. Inclusion of the Vertex group
confirmed that the DLPFC is a specific region associated with context-dependent learning, and
the changes observed in the DLPFC group were not due to a random TMS effect. The magnetic
field generated by the rTMS introduces “noise” and perturbs the neuronal processing of the
stimulated area (Sack & Linden, 2003). Therefore, it is important to choose a control stimulation
site, such as the vertex, to demonstrate that the observed behavioral changes of the DLPFC group
were not due to random noise introduced by rTMS. On the other hand, random noise may have
led to greater variability in motor performance for the participants in the Vertex group leading to
no statistical significance in the outcome measures between the DLPFC group and the Vertex
51

group. Nevertheless, a clear difference can be observed in the mean TT
AC
, RT
AC
, and MT
AC

switch cost values between the DLPFC and the Vertex groups, while no differences were
observed between the Vertex and the Control groups (Figure 4.4).
One limitation of this study is that the results were interpreted based on the assumption that
low frequency rTMS down-regulates neural excitability of the stimulated area. It is well
established that low frequency rTMS can reduce neural excitability of the primary motor cortex
(Hallett, 2007; S. Kantak et al., 2010). Researchers have applied similar parameters over DLPFC
and have assumed that it also induced an inhibitory effect (S. S. Kantak et al., 2010; Pascual-
Leone et al., 1996; van den Heuvel et al., 2013). Studies that have used fMRI or Positron
emission tomography (PET) to investigate the neurophysiological changes after low frequency
rTMS found an increased BOLD signal over DLPFC suggesting increased activation (Knoch,
Treyer, et al., 2006; Li et al., 2004). However, a brain region with increased metabolic response
could be a result of changes either in the excitatory or inhibitory neural networks (Sack & Linden,
2003). Therefore, while low frequency rTMS clearly influenced the processing of the DLPFC in
our study, the exact neurophysiologic mechanism is not known. In addition, since the DLPFC
has numerous neuronal connections to other cortical and subcortical areas, it is not possible to
rule out the possibility that the behavioral changes observed in the DLPFC group were due to
changes in neuronal activation of other brain areas (Ko et al., 2008; Strafella, Paus, Barrett, &
Dagher, 2001).

Conclusion
This study demonstrated that low frequency rTMS over the DLPFC in non-disabled adults
reduced context-dependency during motor sequence learning when compared to the individuals
52

who did not receive rTMS (the Control group) or those who received rTMS over a control site
(the Vertex group). This finding, in combination with the known function of the DLPFC in
encoding and processing contextual information for movement execution, supports the role of
the DLPFC in context-dependent learning. Furthermore, the results of this study suggest that the
DLPFC may be over-activated in individuals with PD as a compensation for the impaired dorsal
striatum during motor learning and performance. Future studies using rTMS over the DLPFC in
people with PD could confirm the over-activation hypothesis by demonstrating reduction in
context-dependent behavior.

53

CHAPTER 5
SUMMARY AND GENERAL DISCUSSION

The overall goal of this dissertation is to investigate the neural substrates associated with
context-dependent learning during motor skill acquisition. Although context-dependent learning
has been observed and investigated in healthy young adults (Keetch et al., 2008; Ruitenberg et al.,
2012; Wright & Shea, 1991), the mechanisms and neural substrates associated with this
phenomenon are not known. In Chapter 1, we hypothesized that context-dependent learning was
likely mediated by the frontostriatal circuit – recurrent neural connections between the
dorsolateral prefrontal cortex (DLPFC) and the dorsal striatum (Alexander et al., 1986). Three
studies were conducted in this dissertation to elucidate the role of the frontostriatal circuit in
context-dependent learning.
Parkinson’s disease (PD) is a neurodegenerative disorder resulting from dopamine
depletion with known dorsal striatum impairments (Packard & Knowlton, 2002). Therefore, the
first study of this dissertation compared context-dependent learning in people with PD and age-
matched non-disabled adults (Chapter 2). To determine whether the frontostriatal circuit is
associated with context-dependent learning, the second study investigated the relationship
between set-shifting ability and context-dependent learning (Chapter 3). In the third study, we
utilized repetitive transcranial magnetic stimulation (rTMS) as a perturbation tool to temporarily
disrupt neuronal processing of the DLPFC in non-disabled adults to investigate the role of the
DLPFC in context-dependent learning (Chapter 4).
This chapter begins by summarizing the specific aims and main findings of each study.
Based on the previous literature and the findings from this dissertation, the role of the dorsal
54

striatum and the DLPFC in context-dependent learning are then discussed. The contribution of
this dissertation work to motor learning research and the clinical implications will be discussed.
Finally, the limitations of this study as well as future directions will be addressed.

Summary of Main Results
The first study of this dissertation investigated context-dependent learning in people with
PD and age-matched non-disabled adults (Chapter 2). Although anecdotal evidence suggests that
people with PD show context-dependent behavior, to our knowledge, this is the first study that
used a systematic method to demonstrate that people with PD are more context-dependent than
non-disabled adults during motor learning and performance. The switch cost results revealed that
greater context-dependency in people with PD was mainly attributed to the movement time (MT),
but not the response time (RT) period. Additional analysis of the movement time accuracy cost
(MT
AC
) switch cost data showed that people with PD conducted extra on-line planning while
executing the numerical sequences under the SWITCH condition. The additional on-line
planning suggests that while the numerical sequence remained the same as practice, people with
PD may consider the sequences under the SWITCH condition as a different sequence (Bo &
Seidler, 2010; Immink & Wright, 2001; Verwey & Dronkert, 1996).
We then hypothesized that context-dependent behavior observed in people with PD could
be related to the integrity of the frontostriatal circuit. In order to establish this relationship, the
second study (Chapter 3) investigated the correlation between context-dependent learning and
set-shifting ability, which is known to be mediated by the frontostriatal circuit (Monchi et al.,
2001; Zakzanis et al., 2005). We used the trail making test (TMT) to determine the set-shifting
ability in people with and without PD, and correlated the results of the TMT to the switch cost
55

value obtained from Study 1. The results showed that switch cost was positively correlated with
the TMT in people with PD, indicating that individuals with PD who had greater difficulty with
set-shifting also demonstrated greater switch cost. However, this relationship was not observed
in non-disabled adults. The findings suggest that the context-dependent behavior observed in
people with PD is related to the disruption of the frontostriatal circuit. Since neuronal processing
of the frontostriatal circuit is intact in non-disabled participants, no relationship was observed
between the switch cost and the result of TMT.
The purpose of the third study was to further determine the role of another important neural
substrate within the frontostriatal circuit, the DLPFC, in context-dependent learning. The study
utilized low frequency rTMS to temporarily perturb neuronal processing of the DLPFC in non-
disabled adults (Chapter 4). The results showed that perturbation to the DLPFC reduced context-
dependency in non-disabled adults. This result, in combination with a known function of the
DLPFC in encoding and processing contextual information (Nee & Brown, 2013), supports the
role of the DLPFC in context-dependent learning. Furthermore, the findings in non-disabled
adults suggest that, during motor learning and performance, the DLPFC may be relatively over-
activated in people with PD in order to compensate for the impaired dorsal striatum.

Role of the Frontostriatal Circuit in Context-Dependent Learning
This dissertation was designed to investigate the role of the frontostriatal circuit in context-
dependent learning. One important function of the frontostriatal circuit is to process contextual
information and flexibly adjust or maintain an action plan according to the environmental
context (Fogelson et al., 2011; Nee & Brown, 2013). Within the frontostriatal circuit, the DLPFC
encodes contextual information and maintains the information in working memory, while the
56

striatum acts as a modulator to filter and select task relevant information (Dominey & Boussaoud,
1997; Nee & Brown, 2013). Therefore, the neuronal interaction between the DLPFC and the
dorsal striatum is critical for updating context and guiding flexible behavior (Nee & Brown,
2013). When the neuronal processing of the frontostriatal circuit is disrupted, the ability to
update contextual information may be affected, and context-dependent behavior may be altered.
In the following two sections, the role of the dorsal striatum and the DLPFC in context-
dependent learning will be discussed separately.

Role of the Striatum in Context-Dependent Learning
As stated above, the striatum functions as a gate or a modulator to filter and select
contextual information from the DLPFC in order to generate the most appropriate movement
(Atallah et al., 2004; Chakravarthy et al., 2010; Frank, 2011; Moustafa & Gluck, 2011). The
striatal neurons respond to information that is behaviorally relevant, and filters information that
is irrelevant to task performance (Gluck et al., 2008; Wise et al., 1996). Da Cuhan et al. proposed
a ‘mosaic of broken mirrors model’ emphasizing the role of the striatum in informational
integration and action selection (Da Cunha et al., 2009). This model suggests that different
cortical inputs (sensory, motor, or emotional) are like image fragments, and the striatum
functions as a processor that puts these fragments together and constructs a mosaic. The mosaic
is thus an optimal response output after integrating all valid information (Da Cunha et al., 2009).
Without the modulation from the striatum, all incoming contextual information encoded by the
DLPFC may be considered important for movement execution. This could potentially lead to
high movement interference by trying to execute all responses considered by the DLPFC
(Atallah et al., 2004).
57

Lesions in the dorsal striatum impair the ability to update contextual information and
modify motor behavior accordingly (McDonald & White, 1994; Rogers, Phillips, Bradshaw,
Iansek, & Jones, 1998). In an animal study using the Morris water maze, rats were trained to use
contextual cues (balloon or lights) to find a platform placed in a certain location in murky water.
At testing, the platform was moved to a new location in the tank, but the relative location of the
contextual cues remained the same. It was found that rats without brain lesion had no difficulty
swimming to the new location and find the platform, while rats with dorsal striatum lesions were
unable to transfer the contextual information to the new platform location. Instead, they first
swam to the old platform location before they were aware that the platform was no longer there
(McDonald & White, 1994). This animal study provides direct evidence of the importance of the
dorsal striatum in updating contextual information for motor performance.
Human studies investigating individuals with PD also provide support for the importance
of the dorsal striatum in filtering and integrating task relevant information for optimal action
selection. It has been found that external cues can help individuals with PD initiate movement
(Lim et al., 2005; Lim et al., 2010). However, if too many external cues are provided or the cued
information is too complicated to process, people with PD demonstrate difficulty using the
information to carry out movement (Rogers et al., 1998). Using electroencephalography (EEG),
it was found that individuals with PD did not show the P3b latency shift (which occurred in non-
disabled adults) while processing contextual information, indicating that people with PD had
impairment utilizing contextual information effectively and efficiently (Fogelson et al., 2011).
The “context” used in the studies described above was information relevant to the task goal
that could be used to guide movement execution. In this dissertation, we used “context” as
information that is incidental to the task goal and should have a lesser impact on task
58

performance. We were especially interested in understanding how the incidental context affects
motor learning and performance in people who have lesions in the striatum, such as PD. The
findings of our first study (Chapter 2) revealed that even though context was not essential for
task performance, people with PD demonstrated an over-reliance on the incidental context (color
and location on the computer screen) in which the numerical sequence was embedded. The over-
reliance on incidental context during practice led to greater context-dependency such that motor
performance significantly decreased when the familiar practice context was no longer available.
Additional analysis of the results suggested that when the associated incidental context was
changed, people with PD might view the numerical sequence as different requiring the need to
reprogram the sequence as if they had not practiced it before. Due to the impaired dorsal striatum,
people with PD may have difficulty filtering out non-essential incidental context and may
consider all information to be essential for task performance during practice. Therefore, when the
incidental context associated with the numerical sequence was changed at testing, additional
planning was needed for individuals with PD to accurately carry out the new sequence-context
combination.

Role of the Dorsolateral Prefrontal Cortex (DLPFC) in Context-Dependent Learning
The DLPFC plays numerous important roles in executive function, such as working
memory, planning, set-shifting, and action selection (Fuster, 2008). When people select and
execute a movement, the DLPFC is especially important for encoding and processing context
necessary for generating the appropriate behavior according to the task goal (S. H. Lee, Kravitz,
& Baker, 2013; Nee & Brown, 2013). For every movement, the incoming context needs to be
updated and maintained in the DLPFC for movement preparation (Reynolds et al., 2012). In
59

addition, the DLPFC evaluates the incoming stimuli with what has already been stored in
memory, and further compares the candidate sequence with the goal sequence (Roberts, Robbins,
& Weiskrantz, 1998). If the DLPFC detects a discrepancy between the incoming context and the
information already stored in memory, all information is then transmitted to the striatum for
refined filtering and selection in order to generate the most appropriate action plan (Fuster, 2008).
When the DLPFC is disrupted, the ability to utilize contextual information appropriately is
also impaired. Animal studies have shown that dysfunction of the prefrontal cortex impairs the
encoding of contextual cues (Parnell et al., 2012), while human studies have shown that patients
with lateral prefrontal cortex lesions were impaired in processing context for a particular
movement goal (Fogelson et al., 2013). Fogelson et al. (2009) found that patients with prefrontal
lesions did not show differences in P3b signal between targets with and without contextual cues,
which was observed in non-disabled adults. This finding indicated that patients with prefrontal
lesions were less able to use contextual information to speed up cognitive processing (Fogelson
et al., 2009).
The third study (Chapter 4) of this dissertation was designed to determine the role of
DLPFC in context-dependent learning. Our results showed that those participants who received
low frequency rTMS over DLPFC, compared to control participants, demonstrated less
decrement in motor performance under the SWITCH condition. The exact neurophysiological
mechanism of this observed result is not clear. Based on the known function of the DLPFC, it is
possible that rTMS over DLPFC reduced the interference from the new context under the
SWITCH testing condition. As mentioned above, the DLPFC encodes new coming context for
each movement and compares the stimuli with previously stored memory (Armbruster et al.,
2012; Roberts et al., 1998). When the DLPFC detects a mismatch between the new and old
60

information, the new context may interfere with the performance of the original sequence-
context association. Although counterintuitive, we believe that low frequency rTMS over the
DLPFC reduced the encoding of new context; thus, less interference from the new context led to
a better performance under the SWITCH testing condition (Figure 4.4).
Ruitenberg et al. (2012) proposed a “context-dependent filtering” hypothesis to explain the
phenomenon of context-dependent learning. The hypothesis stated that the motor performance
under the SAME condition is superior to that of the SWITCH condition because the individual
learned to deal effectively with the interference from the incidental context throughout practice
(Ruitenberg et al., 2012). When a different context is introduced, the DLPFC would detect the
interference from the new context, and longer response time is necessary to reselect the
appropriate action plan. In Study 3 (Chapter 4), we believe that low frequency rTMS applied
over the DLPFC may perturb the encoding process of the new context and thus reduce the
interference (Figure 4.4). When we decomposed the TT
AC
switch cost, the results showed a
greater reduction in RT
AC
switch cost than MT
AC
switch cost. This finding further suggests that
the decreased amount of time necessary for reselecting or reprogramming the sequence may be
due to a reduced interference from new context.
The combined findings from Study 3 and Study 1 (Chapter 2 and 4) suggest that the
DLPFC may be over-activated in individuals with PD in order to compensate for their disrupted
dorsal striatum (Nakamura et al., 2001; Wu & Hallett, 2005). This hypothesis is supported by the
evidence that individuals with PD demonstrated a larger switch cost than non-disabled adults,
while rTMS applied over the DLPFC in non-disabled adults resulted in a smaller switch cost as
compared to those who did not receive rTMS or received rTMS over the Vertex (Figure 5.1).
Based on the assumption that low frequency rTMS reduces neuronal excitability of the
61

stimulated area (S. Kantak et al., 2010; van den Heuvel et al., 2013), we therefore suggest that
the DLPFC may be over-activated in people with PD while performing the finger sequence task.
Another indirect support for our hypothesis of DLPFC over-activation in people with PD is the
analysis of individual finger movement during the movement execution period (Chapter 2,
Figure 2.4). The results revealed that people with PD engaged in additional online planning
during movement execution under the SWITCH condition. This observation supports the
findings from an animal single cell recording study showing that the DLPFC neurons would
activate not only for the final behavioral goal but also for the intermediate goals in sequential
order to execute a movement sequence (Tanji, Shima, & Mushiake, 2007). Therefore, the
element by element additional planning during movement execution under the SWITCH
condition for people with PD could potentially be a result of DLPFC over-activation.

62

Figure 5.1 Combined total time accuracy cost (TT_AC) switch cost results for the PD group
(black), the Control group (white), the DLPFC group (dark gray), and the Vertex group (light
gray). Compared to the Control and the Vertex groups, the PD group demonstrates a greater
switch cost while the DLPFC group demonstrates a smaller switch cost.

Implications for Motor Learning
The effect of incidental context on learning and performance, although widely investigated
in the verbal learning literature (Markopoulos, Rutherford, Cairns, & Green, 2010; Smith & Vela,
2001; Wagner, Desmond, Glover, & Gabrieli, 1998), has been less studied in the field of motor
learning. Empirical evidence points to the influence of incidental context in motor learning and
performance. The “home-field advantage” and accurate set-shots performed by professional
basketball players only at the foul line are two examples in which context influences outcome
(Keetch et al., 2008; Keetch et al., 2005; Schmidt & Lee, 2005). These observations suggest that
63

incidental context, though not essential for task performance, affects our motor behavior in a
subtle but robust way. Wright and Shea (1991) were the first group of authors that used a
systematic method to investigate the influence of incidental context on motor learning in healthy
young adults (Wright & Shea, 1991). However, the neural substrates associated with context-
dependent learning are not known.
This dissertation investigated the neural substrates associated with context-dependent
learning during motor skill acquisition. We used PD as a human model of disruption of the dorsal
striatum, and applied low frequency rTMS to perturb neuronal processing of the DLPFC in non-
disabled adults to determine the role of the frontostriatal circuit in context-dependent learning.
The findings of this dissertation suggest that the integrity of the frontostriatal circuit is important
for context-dependent learning. We demonstrated that when the frontostriatal circuit is disrupted,
context-dependent behavior is also affected. Interestingly, the disruption of the frontostriatal
circuit does not affect the performance of a well-learned motor sequence if the sequence is tested
in the same environmental context as practice. Therefore, in this dissertation, the motor
performance of the second day SAME condition was not different from the performance at the
end of practice on the first day. Compared to the SAME condition, the motor performance under
the SWITCH condition was more affected. The SWITCH condition can be viewed as a “transfer”
condition from a motor learning perspective (Schmidt & Lee, 2005). The SWITCH condition
examines an individual’s capability to carry out a learned motor sequence in a different
environmental context. Damage to the dorsal striatum resulted in a decreased ability to transfer a
well-learned motor task to another context, while perturbation of the DLPFC with rTMS
facilitated this transfer ability. Altogether, these results suggest that a balanced neuronal
interaction between the dorsal striatum and the DLPFC plays an important role in context-
64

dependent learning.
In addition, this dissertation also points to the importance of incidental context during
motor learning, especially for people with PD. While previous motor learning studies have
placed an emphasis on the contextual interference (CI) effect comparing the benefits of random
versus blocked practice schedule (Lin, Fisher, Winstein, Wu, & Gordon, 2008; Schmidt & Lee,
2005; J. B. Shea & Morgan, 1979), this dissertation further suggests that the training
environment (incidental context) should also be considered during motor learning. If a motor
task is only going to be tested in a certain context (e.g., a basketball free throw shot is only
performed at the foul line), then the task should be trained in that particular environment. On the
other hand, if a motor task is carried out under various contexts (e.g., driving a car on various
road conditions), then the motor task cannot be trained in only one context. This could be
especially important for training patients with neurological disorders, such as PD, because they
may over-rely on the training context to the extent that they have difficulty generalizing a learned
task to various contexts. The clinical implications of this dissertation for individuals with PD will
be discussed in the next section.

Clinical Implications
The findings of this dissertation have several important clinical implications, especially for
Parkinson’s disease (PD) rehabilitation. First, Study 1 revealed the significance of the training/
learning environment for people with PD (Chapter 2). It has been suggested that an effective PD
rehabilitation program should consider the individual, the task, and environmental factors (Kelly
et al., 2012). Most of the previous intervention studies have focused on investigating the
individual and the task factors, such as investigating the effect of a treadmill training program on
65

gait and balance in people with PD (Miyai et al., 2000; Yang et al., 2010). Some studies have
begun to suggest that clinicians should also consider the “environment” in which a motor task is
trained for people with PD (Kelly et al., 2012; Nieuwboer et al., 2009). However, little is known
regarding how environmental context affects motor learning and performance in PD. The
findings from our first study (Chapter 2) revealed that although incidental context may not be
essential for task performance, individuals with PD subconsciously over-rely on the contextual
information to perform a learned motor task. If the testing context is different from the practice
context, the motor performance decreases. Therefore, choosing an appropriate training
environment is critical when training people with PD (Nieuwboer et al., 2009). If the goal of a
patient with PD is to complete motor tasks at home, then clinicians should probably consider a
home physical therapy exercise to start with. However, if the goal of the patient is to be able to
carry out a motor task under various conditions (e.g., walk effectively at home, in a mall, and in a
park), then training the motor task under various environmental settings is necessary so the
patient is able to carry out the same task regardless of the context.
The second study (Chapter 3), which showed that switch cost was positively correlated
with the results of TMT in people with PD, also has potential clinical importance. The findings
of Study 2 suggest that an individual with PD who has difficulty shifting between sets may also
demonstrate greater context-dependency. In the clinic, it might not be as easy or feasible to
implement a motor task to test context-dependent learning in PD. The trail making test (TMT),
on the other hand, is a very fast and easy to use clinical tool that can be administered within a
few minutes. Since a positive correlation was found between the result of the TMT and switch
cost in Study 2, the TMT could potentially be used in the clinic to provide some insight regarding
whether a patient with PD may demonstrate context-dependent learning.
66

Study 3 (Chapter 4) in conjunction with the behavioral findings from Study 1 (Chapter 2)
has important clinical implications that could affect our treatment strategies. The participants
with PD in Study 1 demonstrated excessive context-dependency when compared to the non-
disabled adults; contrarily, rTMS over the DLPFC in non-disabled adults reduced context-
dependency as compared to participants who did not receive rTMS or received rTMS over the
vertex (Figure 5.1). These findings altogether suggest that the DLPFC may be over-activated in
individuals with PD, and the over-activation of the DLPFC may lead to excessive context-
dependent behavior. Neuroimaging studies have found an increased activation of the DLPFC in
individuals with PD, when compared to non-disabled adults, while learning a motor sequence
task. This over-activation of the DLPFC has been suggested to be a compensatory strategy for
the impaired striatum (Mallol et al., 2007; Mentis et al., 2003; Nakamura et al., 2001; Wu &
Hallett, 2005). With compensation made by the DLPFC as well as other cortical areas, people
with PD may be able to learn a motor task to a similar behavioral level as non-disabled adults.
However, we found that this compensatory strategy may be “aberrant” because the comparable
learning observed in PD is limited to the same environmental context in which the task was
originally practiced. Over-activation of the DLPFC may limit the ability to generalize a learned
task from one context to another. Since low frequency rTMS reduced context-dependency in
non-disabled adults and since people with PD demonstrated enhanced context-dependency, one
method that could be considered in future studies is to apply rTMS over the DLPFC in people
with PD and determine whether context-dependent learning in this patient population could be
reduced.
Another important issue to consider is the “cueing” strategies commonly used in the clinic.
Since 1987, cueing strategies, such as visual or auditory cues, have been widely used to train
67

walking and balance ability in people with PD (Baker et al., 2007; Dunne, Hankey, & Edis, 1987;
Lim et al., 2010). However, it has been found that improvement in motor function after cueing
training does not have a long-lasting effect once the cues are removed (Nieuwboer et al., 2009;
Rochester et al., 2007). These observations, in conjunction with the results from this dissertation,
imply that people with PD may over-rely on the visual/auditory cues to perform a motor task.
The external cues might be beneficial for movement production via increased activation of the
prefrontal areas (Disbrow et al., 2013), but may not be beneficial for generalizing a learned
motor task to various contexts. Here, we are not opposing the use of cueing strategies, but
proposing that clinicians should take into account the possibility that individuals with PD may
over-rely on the cues, resulting in a difficulty performing a motor task once the cues are removed.
Therefore, when using external cues to train patients with PD, clinicians could incorporate other
motor learning principles, such as a faded feedback method (Winstein, 1991), to decrease the
over-reliance on the cues. For example, when training walking with the use of visual strips
placed on the floor, the number and frequency of strips could be gradually tapered off to help the
patient learn walking while reducing reliance on the visual cues.

Limitations and Future Directions
As discussed in each chapter, one limitation of this dissertation is the small sample size in
each group. We recruited 10 participants in each group, and only 10 individuals with PD
participated in this study. Therefore, future studies are needed to recruit a larger sample of
participants and stratify by disease severity to determine the relationship between context-
dependent learning and disease progression.
Another potential limitation of this dissertation is that we interpreted the results of Study 3
68

based on the assumption that low frequency rTMS induces an inhibitory effect on neuronal
activation of the stimulated area (Hallett, 2007; S. Kantak et al., 2010). Low frequency rTMS has
been long used to down-regulate the neuronal excitability of the primary motor cortex, and past
literature has used low frequency rTMS to reduce neuronal excitability of other cortical areas,
such as the premotor cortex or the DLPFC (Goh, Lee, & Fisher, 2013; S. S. Kantak et al., 2010;
Knoch, Gianotti, et al., 2006; Pascual-Leone et al., 1996). The neurophysiologic effect of low
frequency rTMS over the DLPFC is not well understood. Neuroimaging studies have found an
increased BOLD signal in the DLPFC and its connected areas after low frequency rTMS (Knoch,
Treyer, et al., 2006; Li et al., 2004). Due to the limitations of functional neuroimaging devices, it
is yet to be determined whether the increase in BOLD signal is indicative of increased activation
of the excitatory or inhibitory neural circuits (Sack & Linden, 2003).
Along with the previous limitation, the third limitation is the lack of other neuroimaging
techniques, such as fMRI or PET, to support the findings of this dissertation. Based on the
findings from Study 1 and 3, we hypothesize that the DLPFC may be over-activated in
individuals with PD and this over-activation may be the cause of context-dependent behavior.
Therefore, more neuroimaging studies are needed to support this hypothesis. In addition, future
studies could apply low frequency rTMS over the DLPFC in people with PD and determine
whether the excessive context-dependent learning could be reduced. If a decreased context-
dependency is observed after low frequency rTMS over DLPFC in people with PD, the finding
would also support the over-activation hypothesis proposed by this dissertation.

Conclusion
To our knowledge, the three studies of this dissertation are the first series of studies
69

investigating the neural substrates associated with context-dependency during motor learning.
This dissertation has increased our knowledge of the role of the frontostriatal circuit in context-
dependent learning, and has proposed a potential neural mechanism associated with context-
dependent behavior often observed in people with PD. The findings of this dissertation could be
an essential step for developing more efficient rehabilitation strategies for people with PD.

70

REFERENCES

Abbruzzese, G., Trompetto, C., & Marinelli, L. (2009). The rationale for motor learning in
Parkinson's disease. Eur J Phys Rehabil Med, 45(2), 209-214. doi: R33092106 [pii]
Aizenstein, H. J., Stenger, V . A., Cochran, J., Clark, K., Johnson, M., Nebes, R. D., & Carter, C.
S. (2004). Regional brain activation during concurrent implicit and explicit sequence
learning. Cereb Cortex, 14(2), 199-208.
Alexander, G. E., DeLong, M. R., & Strick, P. L. (1986). Parallel organization of functionally
segregated circuits linking basal ganglia and cortex. Annu Rev Neurosci, 9, 357-381. doi:
10.1146/annurev.ne.09.030186.002041
Arbuthnott, K., & Frank, J. (2000). Trail making test, part B as a measure of executive control:
validation using a set-switching paradigm. J Clin Exp Neuropsychol, 22(4), 518-528. doi:
10.1076/1380-3395(200008)22:4;1-0;FT518
Armbruster, D. J., Ueltzhoffer, K., Basten, U., & Fiebach, C. J. (2012). Prefrontal cortical
mechanisms underlying individual differences in cognitive flexibility and stability. J
Cogn Neurosci, 24(12), 2385-2399. doi: 10.1162/jocn_a_00286
Atallah, H. E., Frank, M. J., & O'Reilly, R. C. (2004). Hippocampus, cortex, and basal ganglia:
insights from computational models of complementary learning systems. Neurobiol
Learn Mem, 82(3), 253-267. doi: S1074742704000693 [pii]
10.1016/j.nlm.2004.06.004
Baker, K., Rochester, L., & Nieuwboer, A. (2007). The immediate effect of attentional, auditory,
and a combined cue strategy on gait during single and dual tasks in Parkinson's disease.
Arch Phys Med Rehabil, 88(12), 1593-1600. doi: S0003-9993(07)01454-2 [pii]
71

10.1016/j.apmr.2007.07.026
Beam, W., Borckardt, J. J., Reeves, S. T., & George, M. S. (2009). An efficient and accurate new
method for locating the F3 position for prefrontal TMS applications. Brain Stimul, 2(1),
50-54. doi: 10.1016/j.brs.2008.09.006
Bertsch, S., & Sanders, R. E. (2005). Age differences in context-cue forgetting. Psychol Rep,
96(3 Pt 2), 879-888.
Bo, J., & Seidler, R. D. (2010). Spatial and symbolic implicit sequence learning in young and
older adults. Exp Brain Res, 201(4), 837-851. doi: 10.1007/s00221-009-2098-5
Bradfield, N. I., Reutens, D. C., Chen, J., & Wood, A. G. (2012). Stereotaxic localisation of the
dorsolateral prefrontal cortex for transcranial magnetic stimulation is superior to the
standard reference position. Aust N Z J Psychiatry, 46(3), 232-239. doi:
10.1177/0004867411430602
Bruckner, S., Kiefer, M., & Kammer, T. (2013). Comparing the after-effects of continuous theta
burst stimulation and conventional 1 Hz rTMS on semantic processing. Neuroscience,
233, 64-71. doi: 10.1016/j.neuroscience.2012.12.033
Bruyer, R., & Brysbaert, M. (2011). Combining speed and accuracy in cognitive psychology: is
the inverse efficiency score (IES) a better dependent variable than the mean reaction time
(RT) and the percentage of errors (PE)? Psychologica Belgica, 51(1), 5-13.
Chakravarthy, V . S., Joseph, D., & Bapi, R. S. (2010). What do the basal ganglia do? A modeling
perspective. Biol Cybern, 103(3), 237-253. doi: 10.1007/s00422-010-0401-y
Chun, M. M., & Jiang, Y . (1998). Contextual cueing: implicit learning and memory of visual
context guides spatial attention. Cogn Psychol, 36(1), 28-71. doi: S0010-0285(98)90681-
8 [pii]
72

10.1006/cogp.1998.0681
Cohen, D. A., & Robertson, E. M. (2011). Preventing interference between different memory
tasks. Nat Neurosci, 14(8), 953-955. doi: 10.1038/nn.2840
Cools, R., Barker, R. A., Sahakian, B. J., & Robbins, T. W. (2001). Mechanisms of cognitive set
flexibility in Parkinson's disease. Brain, 124(Pt 12), 2503-2512.
Da Cunha, C., Wietzikoski, E. C., Dombrowski, P., Bortolanza, M., Santos, L. M., Boschen, S. L.,
& Miyoshi, E. (2009). Learning processing in the basal ganglia: a mosaic of broken
mirrors. Behav Brain Res, 199(1), 157-170. doi: S0166-4328(08)00554-8 [pii]
10.1016/j.bbr.2008.10.001
Disbrow, E. A., Sigvardt, K. A., Franz, E. A., Turner, R. S., Russo, K. A., Hinkley, L. B., . . .
Malhado-Chang, N. (2013). Movement activation and inhibition in Parkinson's disease: a
functional imaging study. J Parkinsons Dis, 3(2), 181-192. doi: 10.3233/JPD-130181
Dominey, P. F., & Boussaoud, D. (1997). Encoding behavioral context in recurrent networks of
the fronto-striatal system: a simulation study. Brain Res Cogn Brain Res, 6(1), 53-65. doi:
S0926-6410(97)00015-3 [pii]
Dominey, P. F., & Jeannerod, M. (1997). Contribution of frontostriatal function to sequence
learning in Parkinson's disease: evidence for dissociable systems. Neuroreport, 8(5), iii-ix.
Doyon, J., Penhune, V ., & Ungerleider, L. G. (2003). Distinct contribution of the cortico-striatal
and cortico-cerebellar systems to motor skill learning. Neuropsychologia, 41(3), 252-262.
doi: S0028393202001586 [pii]
Dunne, J. W., Hankey, G. J., & Edis, R. H. (1987). Parkinsonism: upturned walking stick as an
aid to locomotion. Arch Phys Med Rehabil, 68(6), 380-381.
Flowers, K. A., & Robertson, C. (1985). The effect of Parkinson's disease on the ability to
73

maintain a mental set. J Neurol Neurosurg Psychiatry, 48(6), 517-529.
Fogelson, N., & Fernandez-Del-Olmo, M. (2013). Implicit versus explicit local contextual
processing. PLoS One, 8(6), e65914. doi: 10.1371/journal.pone.0065914
Fogelson, N., Fernandez-del-Olmo, M., & Santos-Garcia, D. (2011). Contextual processing
deficits in Parkinson's disease: the role of the frontostriatal system. Clin Neurophysiol,
122(3), 539-545. doi: 10.1016/j.clinph.2010.07.017
Fogelson, N., Li, L., Li, Y ., Fernandez-Del-Olmo, M., Santos-Garcia, D., & Peled, A. (2013).
Functional connectivity abnormalities during contextual processing in schizophrenia and
in Parkinson's disease. Brain Cogn, 82(3), 243-253. doi: 10.1016/j.bandc.2013.05.001
Fogelson, N., Shah, M., Scabini, D., & Knight, R. T. (2009). Prefrontal cortex is critical for
contextual processing: evidence from brain lesions. Brain, 132(Pt 11), 3002-3010. doi:
10.1093/brain/awp230
Frank, M. J. (2011). Computational models of motivated action selection in corticostriatal
circuits. Curr Opin Neurobiol, 21, 1-6. doi: S0959-4388(11)00040-7 [pii]
10.1016/j.conb.2011.02.013
Fritsch, T., Smyth, K. A., Wallendal, M. S., Hyde, T., Leo, G., & Geldmacher, D. S. (2012).
Parkinson disease: research update and clinical management. South Med J, 105(12), 650-
656. doi: 10.1097/SMJ.0b013e318273a60d
Fuster, Joaquin M. (2008). The prefrontal cortex (4th ed.). Amsterdam ; Boston: Academic
Press/Elsevier.
Galea, J. M., Albert, N. B., Ditye, T., & Miall, R. C. (2010). Disruption of the dorsolateral
prefrontal cortex facilitates the consolidation of procedural skills. J Cogn Neurosci, 22(6),
1158-1164. doi: 10.1162/jocn.2009.21259
74

Gluck, Mark A., Mercado, Eduardo, & Myers, Catherine E. (2008). Learning and memory : from
brain to behavior. New York: Worth Publishers.
Goh, H. T., Lee, Y . Y ., & Fisher, B. E. (2013). Neural correlates of dual-task practice benefit on
motor learning: a repetitive transcranial magnetic stimulation study. Eur J Neurosci,
37(11), 1823-1829. doi: 10.1111/ejn.12192
Hallett, M. (2007). Transcranial magnetic stimulation: a primer. Neuron, 55(2), 187-199. doi:
10.1016/j.neuron.2007.06.026
Higginson, C. I., Lanni, K., Sigvardt, K. A., & Disbrow, E. A. (2013). The contribution of trail
making to the prediction of performance-based instrumental activities of daily living in
Parkinson's disease without dementia. J Clin Exp Neuropsychol, 35(5), 530-539. doi:
10.1080/13803395.2013.798397
Immink, M. A., & Wright, D. L. (1998). Contextual Interference: A Response Planning Account.
The Quarterly Journal of Experimental Psychology, 51, 735-754.
Immink, M. A., & Wright, D. L. (2001). Motor programming during practice conditions high and
low in contextual interference. J Exp Psychol Hum Percept Perform, 27(2), 423-437.
Kantak, S., Fisher, B. E., Sullivan, K. J., & Winstein, C. (2010). Effects of different doses of low
frequency rTMS on motor corticospinal excitability. Journal of Neurology and
Neurophysiology, 1(1), 6.
Kantak, S. S., Sullivan, K. J., Fisher, B. E., Knowlton, B. J., & Winstein, C. J. (2010). Neural
substrates of motor memory consolidation depend on practice structure. Nat Neurosci,
13(8), 923-925. doi: nn.2596 [pii]
10.1038/nn.2596
Keetch, K. M., Lee, T. D., & Schmidt, R. A. (2008). Especial skills: specificity embedded within
75

generality. J Sport Exerc Psychol, 30(6), 723-736.
Keetch, K. M., Schmidt, R. A., Lee, T. D., & Young, D. E. (2005). Especial skills: their
emergence with massive amounts of practice. J Exp Psychol Hum Percept Perform, 31(5),
970-978. doi: 2005-13471-011 [pii]
10.1037/0096-1523.31.5.970
Kelly, V . E., Eusterbrock, A. J., & Shumway-Cook, A. (2012). A review of dual-task walking
deficits in people with Parkinson's disease: motor and cognitive contributions,
mechanisms, and clinical implications. Parkinsons Dis, 2012, 918719. doi:
10.1155/2012/918719
Keys, B. A., & White, D. A. (2000). Exploring the relationship between age, executive abilities,
and psychomotor speed. J Int Neuropsychol Soc, 6(1), 76-82.
Kierzynka, A., Kazmierski, R., & Kozubski, W. (2011). Educational level and cognitive
impairment in patients with Parkinson disease. Neurol Neurochir Pol, 45(1), 24-31.
Kimbrough, S. K., Wright, D. L., & Shea, C. H. (2001). Reducing the saliency of intentional
stimuli results in greater contextual-dependent performance. Memory, 9(2), 133-143.
Knoch, D., Gianotti, L. R., Pascual-Leone, A., Treyer, V ., Regard, M., Hohmann, M., & Brugger,
P. (2006). Disruption of right prefrontal cortex by low-frequency repetitive transcranial
magnetic stimulation induces risk-taking behavior. J Neurosci, 26(24), 6469-6472. doi:
10.1523/JNEUROSCI.0804-06.2006
Knoch, D., Treyer, V ., Regard, M., Muri, R. M., Buck, A., & Weber, B. (2006). Lateralized and
frequency-dependent effects of prefrontal rTMS on regional cerebral blood flow.
Neuroimage, 31(2), 641-648. doi: 10.1016/j.neuroimage.2005.12.025
Ko, J. H., Monchi, O., Ptito, A., Bloomfield, P., Houle, S., & Strafella, A. P. (2008). Theta burst
76

stimulation-induced inhibition of dorsolateral prefrontal cortex reveals hemispheric
asymmetry in striatal dopamine release during a set-shifting task: a TMS-
[(11)C]raclopride PET study. Eur J Neurosci, 28(10), 2147-2155. doi: 10.1111/j.1460-
9568.2008.06501.x
Lee, E. Y ., Cowan, N., Vogel, E. K., Rolan, T., Valle-Inclan, F., & Hackley, S. A. (2010). Visual
working memory deficits in patients with Parkinson's disease are due to both reduced
storage capacity and impaired ability to filter out irrelevant information. Brain, 133(9),
2677-2689. doi: awq197 [pii]
10.1093/brain/awq197
Lee, S. H., Kravitz, D. J., & Baker, C. I. (2013). Goal-dependent dissociation of visual and
prefrontal cortices during working memory. Nat Neurosci, 16(8), 997-999. doi:
10.1038/nn.3452
Li, X., Nahas, Z., Kozel, F. A., Anderson, B., Bohning, D. E., & George, M. S. (2004). Acute left
prefrontal transcranial magnetic stimulation in depressed patients is associated with
immediately increased activity in prefrontal cortical as well as subcortical regions. Biol
Psychiatry, 55(9), 882-890. doi: 10.1016/j.biopsych.2004.01.017
Lim, I., van Wegen, E., de Goede, C., Deutekom, M., Nieuwboer, A., Willems, A., . . . Kwakkel,
G. (2005). Effects of external rhythmical cueing on gait in patients with Parkinson's
disease: a systematic review. Clin Rehabil, 19(7), 695-713.
Lim, I., van Wegen, E., Jones, D., Rochester, L., Nieuwboer, A., Willems, A. M., . . . Kwakkel, G.
(2010). Does cueing training improve physical activity in patients with Parkinson's
disease? Neurorehabil Neural Repair, 24(5), 469-477. doi: 1545968309356294 [pii]
10.1177/1545968309356294
77

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. doi: 464293K72374524V [pii]
10.3200/JMBR.40.6.578-586
Mallol, R., Barros-Loscertales, A., Lopez, M., Belloch, V ., Parcet, M. A., & Avila, C. (2007).
Compensatory cortical mechanisms in Parkinson's disease evidenced with fMRI during
the performance of pre-learned sequential movements. Brain Res, 1147, 265-271. doi:
S0006-8993(07)00329-0 [pii]
10.1016/j.brainres.2007.02.046
Markopoulos, G., Rutherford, A., Cairns, C., & Green, J. (2010). Encoding instructions and
stimulus presentation in local environmental context-dependent memory studies. Memory,
18(6), 610-624. doi: 924425934 [pii]
10.1080/09658211.2010.497764
McDonald, R. J., & White, N. M. (1994). Parallel information processing in the water maze:
evidence for independent memory systems involving dorsal striatum and hippocampus.
Behav Neural Biol, 61(3), 260-270.
Mentis, M. J., Dhawan, V ., Feigin, A., Delalot, D., Zgaljardic, D., Edwards, C., & Eidelberg, D.
(2003). Early stage Parkinson's disease patients and normal volunteers: comparative
mechanisms of sequence learning. Hum Brain Mapp, 20(4), 246-258. doi:
10.1002/hbm.10142
Michael, N., Gosling, M., Reutemann, M., Kersting, A., Heindel, W., Arolt, V ., & Pfleiderer, B.
(2003). Metabolic changes after repetitive transcranial magnetic stimulation (rTMS) of
78

the left prefrontal cortex: a sham-controlled proton magnetic resonance spectroscopy (1H
MRS) study of healthy brain. Eur J Neurosci, 17(11), 2462-2468.
Miyai, I., Fujimoto, Y ., Ueda, Y ., Yamamoto, H., Nozaki, S., Saito, T., & Kang, J. (2000).
Treadmill training with body weight support: its effect on Parkinson's disease. Arch Phys
Med Rehabil, 81(7), 849-852. doi: 10.1053/apmr.2000.4439
Monchi, O., Petrides, M., Doyon, J., Postuma, R. B., Worsley, K., & Dagher, A. (2004). Neural
bases of set-shifting deficits in Parkinson's disease. J Neurosci, 24(3), 702-710. doi:
10.1523/JNEUROSCI.4860-03.2004
24/3/702 [pii]
Monchi, O., Petrides, M., Mejia-Constain, B., & Strafella, A. P. (2007). Cortical activity in
Parkinson's disease during executive processing depends on striatal involvement. Brain,
130(Pt 1), 233-244. doi: awl326 [pii]
10.1093/brain/awl326
Monchi, O., Petrides, M., Petre, V ., Worsley, K., & Dagher, A. (2001). Wisconsin Card Sorting
revisited: distinct neural circuits participating in different stages of the task identified by
event-related functional magnetic resonance imaging. J Neurosci, 21(19), 7733-7741. doi:
21/19/7733 [pii]
Moustafa, A. A., & Gluck, M. A. (2011). A neurocomputational model of dopamine and
prefrontal-striatal interactions during multicue category learning by Parkinson patients. J
Cogn Neurosci, 23(1), 151-167. doi: 10.1162/jocn.2010.21420
Nakamura, T., Ghilardi, M. F., Mentis, M., Dhawan, V ., Fukuda, M., Hacking, A., . . . Eidelberg,
D. (2001). Functional networks in motor sequence learning: abnormal topographies in
Parkinson's disease. Hum Brain Mapp, 12(1), 42-60.
79

Nee, D. E., & Brown, J. W. (2013). Dissociable frontal-striatal and frontal-parietal networks
involved in updating hierarchical contexts in working memory. Cereb Cortex, 23(9),
2146-2158. doi: 10.1093/cercor/bhs194
Nieuwboer, A., De Weerdt, W., Dom, R., Truyen, M., Janssens, L., & Kamsma, Y . (2001). The
effect of a home physiotherapy program for persons with Parkinson's disease. J Rehabil
Med, 33(6), 266-272.
Nieuwboer, A., Rochester, L., Muncks, L., & Swinnen, S. P. (2009). Motor learning in
Parkinson's disease: limitations and potential for rehabilitation. Parkinsonism Relat
Disord, 15 Suppl 3, S53-58. doi: S1353-8020(09)70781-3 [pii]
10.1016/S1353-8020(09)70781-3
Onla-or, S., & Winstein, C. J. (2008). Determining the optimal challenge point for motor skill
learning in adults with moderately severe Parkinson's disease. Neurorehabil Neural
Repair, 22(4), 385-395. doi: 1545968307313508 [pii]
10.1177/1545968307313508
Owen, A. M., James, M., Leigh, P. N., Summers, B. A., Marsden, C. D., Quinn, N. P., . . .
Robbins, T. W. (1992). Fronto-striatal cognitive deficits at different stages of Parkinson's
disease. Brain, 115 ( Pt 6), 1727-1751.
Owen, A. M., Roberts, A. C., Polkey, C. E., Sahakian, B. J., & Robbins, T. W. (1991). Extra-
dimensional versus intra-dimensional set shifting performance following frontal lobe
excisions, temporal lobe excisions or amygdalo-hippocampectomy in man.
Neuropsychologia, 29(10), 993-1006. doi: 0028-3932(91)90063-E [pii]
Packard, M. G., & Knowlton, B. J. (2002). Learning and memory functions of the Basal Ganglia.
Annu Rev Neurosci, 25, 563-593. doi: 10.1146/annurev.neuro.25.112701.142937
80

112701.142937 [pii]
Parnell, R., Grasby, K., & Talk, A. (2012). The prefrontal cortex is required for incidental
encoding but not recollection of source information in rodents. Behav Brain Res, 232(1),
77-83. doi: 10.1016/j.bbr.2012.03.040
Pascual-Leone, A., Wassermann, E. M., Grafman, J., & Hallett, M. (1996). The role of the
dorsolateral prefrontal cortex in implicit procedural learning. Exp Brain Res, 107(3), 479-
485.
Poldrack, R. A., Sabb, F. W., Foerde, K., Tom, S. M., Asarnow, R. F., Bookheimer, S. Y ., &
Knowlton, B. J. (2005). The neural correlates of motor skill automaticity. J Neurosci,
25(22), 5356-5364. doi: 25/22/5356 [pii]
10.1523/JNEUROSCI.3880-04.2005
Reynolds, J. R., O'Reilly, R. C., Cohen, J. D., & Braver, T. S. (2012). The function and
organization of lateral prefrontal cortex: a test of competing hypotheses. PLoS One, 7(2),
e30284. doi: 10.1371/journal.pone.0030284
Roberts, A. C., Robbins, Trevor W., & Weiskrantz, Lawrence. (1998). The prefrontal cortex :
executive and cognitive functions. Oxford ; New York: Oxford University Press.
Rochester, L., Nieuwboer, A., Baker, K., Hetherington, V ., Willems, A. M., Chavret, F., . . . Jones,
D. (2007). The attentional cost of external rhythmical cues and their impact on gait in
Parkinson's disease: effect of cue modality and task complexity. J Neural Transm,
114(10), 1243-1248. doi: 10.1007/s00702-007-0756-y
Rogers, M. A., Phillips, J. G., Bradshaw, J. L., Iansek, R., & Jones, D. (1998). Provision of
external cues and movement sequencing in Parkinson's disease. Motor Control, 2(2), 125-
132.
81

Ruitenberg, M. F., De Kleine, E., Van der Lubbe, R. H., Verwey, W. B., & Abrahamse, E. L.
(2012). Context-dependent motor skill and the role of practice. Psychol Res, 76(6), 812-
820. doi: 10.1007/s00426-011-0388-6
Sack, A. T., & Linden, D. E. (2003). Combining transcranial magnetic stimulation and functional
imaging in cognitive brain research: possibilities and limitations. Brain Res Brain Res
Rev, 43(1), 41-56.
Schmidt, Richard A., & Lee, Timothy D. (2005). Motor control and learning : a behavioral
emphasis (Fourth ed.). Champaign, Ill. :: Human Kinetics Publishers.
Shea, C.H., & Wright, D.L. (1995). Contextual dependencies: influence on response latency.
Memory, 3(1), 81-95.
Shea, J.B., & Morgan, R.L. (1979). Contextual interference effects on the acquisition, retention,
and transfer of a motor skill. Journal of Experimental Psychology: Human Learning and
Memory, 5(2), 179-187.
Siebner, H. R., & Rothwell, J. (2003). Transcranial magnetic stimulation: new insights into
representational cortical plasticity. Exp Brain Res, 148(1), 1-16. doi: 10.1007/s00221-
002-1234-2
Smith, S. M. (1982). Enhancement of recall using multiple environmental contexts during
learning. Mem Cognit, 10(5), 405-412.
Smith, S. M. (1984). A comparison of two techniques for reducing context-dependent forgetting.
Mem Cognit, 12(5), 477-482.
Smith, S. M. (1986). Environmental context-dependent recognition memory using a short-term
memory task for input. Mem Cognit, 14(4), 347-354.
Smith, S. M., & Vela, E. (2001). Environmental context-dependent memory: a review and meta-
82

analysis. Psychon Bull Rev, 8(2), 203-220.
Strafella, A. P., Paus, T., Barrett, J., & Dagher, A. (2001). Repetitive transcranial magnetic
stimulation of the human prefrontal cortex induces dopamine release in the caudate
nucleus. J Neurosci, 21(15), RC157.
Tanji, J., Shima, K., & Mushiake, H. (2007). Concept-based behavioral planning and the lateral
prefrontal cortex. Trends Cogn Sci, 11(12), 528-534. doi: 10.1016/j.tics.2007.09.007
Vakil, E., Raz, T., & Levy, D. A. (2010). Probing the brain substrates of cognitive processes
responsible for context effects on recognition memory. Neuropsychol Dev Cogn B Aging
Neuropsychol Cogn, 17(5), 519-544. doi: 921897052 [pii]
10.1080/13825581003690182
van den Heuvel, O. A., Van Gorsel, H. C., Veltman, D. J., & Van Der Werf, Y . D. (2013).
Impairment of executive performance after transcranial magnetic modulation of the left
dorsal frontal-striatal circuit. Hum Brain Mapp, 34(2), 347-355. doi: 10.1002/hbm.21443
van Eimeren, T., Monchi, O., Ballanger, B., & Strafella, A. P. (2009). Dysfunction of the default
mode network in Parkinson disease: a functional magnetic resonance imaging study. Arch
Neurol, 66(7), 877-883. doi: 10.1001/archneurol.2009.97
Verwey, W. B., & Dronkert, Y . (1996). Practicing a Structured Continuous Key-Pressing Task:
Motor Chunking or Rhythm Consolidation? J Mot Behav, 28(1), 71-79. doi:
10.1080/00222895.1996.9941735
Wagner, A. D., Desmond, J. E., Glover, G. H., & Gabrieli, J. D. (1998). Prefrontal cortex and
recognition memory. Functional-MRI evidence for context-dependent retrieval processes.
Brain, 121 ( Pt 10), 1985-2002.
Winstein, C. J. (1991). Knowledge of results and motor learning--implications for physical
83

therapy. Phys Ther, 71(2), 140-149.
Wise, S. P., Murray, E. A., & Gerfen, C. R. (1996). The frontal cortex-basal ganglia system in
primates. Crit Rev Neurobiol, 10(3-4), 317-356.
Wright, D. L., & Shea, C. H. (1991). Contextual dependencies in motor skills. Mem Cognit,
19(4), 361-370.
Wu, T., & Hallett, M. (2005). A functional MRI study of automatic movements in patients with
Parkinson's disease. Brain, 128(Pt 10), 2250-2259. doi: awh569 [pii]
10.1093/brain/awh569
Yang, Y . R., Lee, Y . Y ., Cheng, S. J., & Wang, R. Y. (2010). Downhill walking training in
individuals with Parkinson's disease: a randomized controlled trial. Am J Phys Med
Rehabil, 89(9), 706-714. doi: 10.1097/PHM.0b013e3181e721c5
Zakzanis, K. K., Mraz, R., & Graham, S. J. (2005). An fMRI study of the Trail Making Test.
Neuropsychologia, 43(13), 1878-1886. doi: 10.1016/j.neuropsychologia.2005.03.013
Zgaljardic, D. J., Borod, J. C., Foldi, N. S., & Mattis, P. (2003). A review of the cognitive and
behavioral sequelae of Parkinson's disease: relationship to frontostriatal circuitry. Cogn
Behav Neurol, 16(4), 193-210.

Abstract (if available)

Abstract This dissertation is designed to investigate neural substrates associated with context-dependent learning. Context-dependent learning is a phenomenon in which people demonstrate superior performance in the environmental context where they originally learned a motor task and conversely, do not perform as well if the task is carried out in a novel context. While context-dependent learning has been mostly established in healthy young adults, it has not been systematically investigated in people with neurological disorders or non-disabled older adults. In addition, the neural substrates associated with context-dependent learning for motor skill acquisition are not well understood. One neural network that could potentially be associated with context-dependent learning is the frontostriatal circuit – recurrent neural connections between the dorsolateral prefrontal cortex (DLPFC) and the dorsal striatum. Animal and computer simulation studies have suggested that the frontostriatal circuit is important for selecting an appropriate movement plan according to environmental stimuli. While the DLPFC encodes all contextual information associated with a task, the dorsal striatum selects and filters task relevant information in order to generate the most appropriate action plan. Based on this proposed function of the frontostriatal circuit in processing contextual information, we hypothesized that the neuronal interactions between the DLPFC and dorsal striatum within the frontostriatal circuit could be important for mediating context-dependent learning. Therefore, three studies were designed in this dissertation to test this hypothesis. ❧ In the first study, we recruited individuals with Parkinson’s disease (PD), known to have striatum impairments, to test the hypothesis that striatum is a potential neural substrate for context-dependent learning. Ten individuals with PD and 10 age-matched non-disabled adults were recruited into the PD group and the Control group. The study was conducted over two consecutive days approximately 24 hours apart. On the first day, participants practiced a finger sequence task consisting of 3 numerical sequences. Unknown to the participants, each sequence was embedded within a specific colored circle and a specific location on the computer screen. On the second day, the participants were tested under two testing conditions: SWITCH and SAME conditions. Under the SWITCH condition, the context associated with each sequence was changed from that of practice

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Neural substrates of motor memory consolidation: a double dissociation of primary motor cortex and dorsolateral prefrontal cortex induced by practice structure

PDF

Impact of enhanced efficacy expectation on motor learning in individuals with Parkinson’s disease

PDF

Locomotor skill learning in virtual reality in healthy adults and people with Parkinson disease

PDF

Relationship between interhemispheric inhibition and bimanual coordination in musicians

PDF

Behavioral and neurophysiological studies of hand dexterity in health and Parkinson's disease

PDF

Dynamic postural control during simple and complex locomotor tasks in persons with early stage Parkinson’s disease

PDF

Exploring the gut-brain axis in Parkinson’s disease and the influence of physical fitness to restore gut health

PDF

Contextual interference in motor skill learning: an investigation of the practice schedule effect using transcranial magnetic stimulation (TMS)

PDF

Experience-dependent neuroplasticity of the dorsal striatum and prefrontal cortex in the MPTP-lesioned mouse model of Parkinson’s disease

PDF

Corticomotor excitability of gluteus maximus: influence on hip extensor strength and hip mechanics

PDF

Age-related differences in amygdala subregion functional connectivity with Ventromedial Prefrontal Cortex and Dorsolateral Prefrontal Cortex

PDF

Mechanisms of stress effects on learning and decision making in younger and older adults

PDF

Brain and behavior correlates of intrinsic motivation and skill learning

PDF

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

PDF

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

PDF

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

PDF

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

PDF

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

PDF

Manipulating cognitive intensity during aerobic exercise: clinical proof of concept

Asset Metadata

Creator Lee, Ya-Yun (Alice) (author)

Core Title Neural substrates associated with context-dependent learning

School School of Dentistry

Degree Doctor of Philosophy

Degree Program Biokinesiology and Physical Therapy

Publication Date 11/01/2013

Defense Date 10/11/2013

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

Tag context-dependent learning,dorsolateral prefrontal cortex,OAI-PMH Harvest,Parkinson's disease,transcranial magnetic stimulation

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Fisher, Beth E. (committee chair), Gordon, James G. (committee member), Petzinger, Giselle M. (committee member), Winstein, Carolee J. (committee member), Zelinski, Elizabeth M. (committee member)

Creator Email s19006026@yahoo.com.tw,yayunlee@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-342152

Unique identifier UC11296575

Identifier etd-LeeYaYunAl-2126.pdf (filename),usctheses-c3-342152 (legacy record id)

Legacy Identifier etd-LeeYaYunAl-2126.pdf

Dmrecord 342152

Document Type Dissertation

Format application/pdf (imt)

Rights Lee, Ya-Yun (Alice)

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

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