Close
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The interaction between explicit knowledge and implicit motor -sequence learning following focal brain damage
(USC Thesis Other)
The interaction between explicit knowledge and implicit motor -sequence learning following focal brain damage
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE INTERACTION BETWEEN EXPLICIT KNOWLEDGE
AND IMPLICIT MOTOR-SEQUENCE LEARNING
FOLLOWING FOCAL BRAIN DAMAGE
Copyright 2001
by
Lara A. Boyd
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOKINESIOLOGY)
August 2001
Lara A. Boyd
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3093413
UMI
UMI Microform 3093413
Copyright 2003 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
L a ra A . Boyd
under the direction of hzx. Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR O F PHILO SO PHY
Dean of Graduate Studies
Date August _ ,^2001
DISSERTATION COMMITTEE
Chairperson
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ii
ACKNOWLEDGEMENTS
So many individuals have contributed to the work detailed in these
pages that it is hard to know where to begin. First, my gratitude to my family
and friends, who constantly supported this work and continue encourage my
ongoing quest to better understand the relationships between the brain and
behaviors. In particular, my thanks to my parents who taught me always to be
curious, and above all else to value education and learning.
Dr. Carolee Winstein has been an unfailing advisor, mentor, and friend.
A great scientist, critical thinker, and excellent clinician she blends all of the
characteristics elemental for behaviorally based research. I can only hope to
live up to her example and expectations.
Many thanks to my dissertation committee, Dr. Stanley Azen, Dr.
Lucinda Baker, Dr. Helena Chui, Dr. James Gordon, and Dr. Jack Turman
whose insights have guided my dissertation research as well as my growth as
an academician.
I could not have completed this work without the critical appraisal,
support, and friendship of my fellow graduate students (past and present) at
the University of Southern California; Katherine Sullivan, Beth Fisher, Dorian
Rose, Kathleen Ganley, Craig Newsam, and Chelle Prettyman. In particular,
thanks to Somporn Onla-or, my friend and role model, we still have much work
to complete together.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
From Rancho my thanks to Dr. Jacquelin Perry, Dr. Sara Mulroy, and
Dr. Rebecca Lewthwaite, I owe each of you a debt of gratitude for many years
of encouragement, support, and guidance.
Last, my unending gratitude to my partner in life and all else, Brenda
Wessel. You have been my support and sanity. I can only hope that someday
you let me return the favor.
Funding for this work was provided by the Neurology Section of the
American Physical Therapy Association (Patricia Leahy Scholarship),
Foundation for Physical Therapy (PODS Scholarships Levels I and II), and the
Department of Biokinesiology and Physical Therapy, University of Southern
California that generously contributed to the CT task apparatus and software,
as well as additional scholarship funding (Jacquelin Perry Scholarship).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iv
TABLE OF CONTENTS
Acknowledgements ii
List of Tables vi
List of Figures vii
Abstract xiii
Preface xv
Chapter 1 THE QUESTION: 1
INTRODUCTION AND OVERVIEW
Chapter 2 DECLARATIVE AND PROCEDURAL SYSTEMS: 5
NEUROANATOMY AND FUNCTION
Chapter 3 NEUROPSYCHOLOGY OF IMPLICIT MOTOR- 67
SEQUENCE LEARNING
Chapter 4 RESEARCH DESIGN AND METHODS 90
Chapter 5 THE IMPACT OF FOCAL BRAIN DAMAGE 114
ON IMPLICIT MOTOR-SEQUENCE LEARNING
Chapter 6 THE EFFECT OF PRIOR EXPLICIT 152
KNOWLEDGE ON IMPLICIT MOTOR-
SEQUENCE LEARNING FOLLOWING
FOCAL BRAIN DAMAGE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 7
Chapter 8
References
Appendix
THE INTERACTIONS BETWEEN EXPLICIT
KNOWLEDGE, TASK DEMAND, AND FOCAL
BRAIN DAMAGE ON IMPLICIT MOTOR-
SEQUENCE LEARNING
GENERAL DISCUSSION AND SUMMARY
v
2 0 0
221
257
279
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vi
TABLES
Chapter 2:
Table 1. Overview of Representative Animal Studies Using Motor 18
Learning Paradigms.
Table 2. Synopsis of Representative Imaging Studies in Humans 20
during Explicit and Implicit Learning and Memory Tasks.
Table 3. Summary of Relevant Human Patient Group Studies 24
Using Implicit Motor-Sequence Learning Tasks.
Table 4. Predicted impact of explicit knowledge condition and lesion 54
location on implicit motor-sequence learning.
Chapter 4:
Table 1. Experimental Design and Overview. 92
Table 2. Subject Characteristics. 96
Table 3. Subjective Questionnaire for Explicit Awareness of the 101
Serial Reaction Time and Continuous Tracking Tasks.
Chapter 5:
Table 1. Explicit Knowledge of the No-EK Groups. 139
Chapter 6:
Table 1. Explicit Knowledge of the EK Groups. 163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vii
FIGURES
Chapter 2:
Figure 1. Classification of declarative and procedural memory and 7
learning systems.
Figure 2. Schematic representation of the neural network underpinning 52
implicit learning.
Figure 3. Putative interrelationships between implicit and explicit neural 62
networks.
Figure 4. The interrelationships between the basal ganglia and 64
cerebellum during implicit motor-skill learning.
Chapter 3:
Figure 1. RT change across sequence block practice for ST-Unaware 72
group. At the conclusion of practice there was no difference between
RTs for the random and sequence practice conditions. Error bars are
the standard error of the mean (SEM). The zero line is the median RT
for the second random sequence block.
Figure 2. RT change across 3 practice days for ST-Extended Practice 74
group. Each point represents median RT change score for that day’s
practice. The zero line is the median RT for the second random
condition from the last day of practice. Despite extended practice mean
median RT did not decrease for the sequence relative to the random
condition.
Figure 3. RT change across practice for ST-Explicit Knowledge group. 76
A significant decrease in RT for the last block of sequence practice as
compared to the random condition was noted. Error bars are SEM.
The zero line is the median RT for the second random sequence block.
Figure 4. Mean change in RT for ST-Unaware, ST-Extended Practice, 77
and ST- Explicit Knowledge groups. Positive numbers demonstrate a
decrease in sequence RT relative to the random condition and reflect
implicit learning. The ST-Explicit Knowledge group demonstrated a
significantly larger decrease in median RT when compared to both
the ST-Unaware and ST-Extended Practice groups.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
viii
Chapter 4:
Figure 1. Set-up for the serial reaction time task. Subjects were 98
seated in front of a computer screen with their hand resting on the
tabletop and four fingers (all except thumb) lightly touching the
“v”, “b”, “n”, and “m” keys. The keys were colored yellow, red,
blue, and green.
Figure 2. Sample waveforms from the continuous tracking task 105
with the repeating third highlighted. The middle third of every trial
was identical and repeated. The first and last thirds were generated
randomly and differed for every trial.
Figure 3. Set-up for the continuous tracking task. Subjects were 106
seated facing a dark computer screen with one arm resting on a
frictionless lever. Subjects moved the lever in an arc from 0 to 90°
of internal rotation as they tracked the target across the screen.
Internal rotation moved the cursor down while external rotation
moved it up on the screen.
Figure 4. Sample target and subject waveforms from the second 110
day of practice.
(A) Example of the repeated middle third of the waveform along
with a representative subject response. The correlation coefficient
(r) and the time lag of the subject’s response are also presented.
(B) For the time series analysis the subject’s response was “slid”
along the target, with correlation coefficients calculated serially;
for every interval the subject response was moved. When the
correlation coefficient reached a maximum, the two waveforms were
considered a best fit. The magnitude of the distance the subject’s
waveform was moved was converted to time (msec) and expressed
as tracking lag.
Chapter 5:
Figure 1. Change in response time (RT) for the repeated sequence 123
across acquisition on the SRT task for the No-EK groups. Error bars are
the SEM. The zero line represents random sequence RT. Data below
this line illustrate decreased RT and reflect implicit learning. All groups
(healthy control (HC), BG, CB, and SMC) decreased RT for the repeated
sequence relative to random with practice.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ix
Figure 2. RT for repeated versus random sequence at the retention 126
test on the SRT task. Error bars are SEM. A Main Effect of Sequence
showed that for all groups RT was faster for the repeated as compared
to that for random sequences. Statistically reliable differences
demonstrated that the repeated sequence RTs for the BG and
SMC groups were larger than that for the HC group (Main Effect
of Group).
Figure 3. Retention test learning scores for the SRT task. Error bars 127
are SEM. Positive numbers show a decrease in repeating sequence
RT relative to the random condition and reflect implicit learning. A
trend for between group differences was found. This was due to the
larger magnitude of change in repeating sequence RT shown by the
SMC group compared to that for the HC group. An effect size
calculation confirmed this difference (ES=0.82).
Figure 4. Change in tracking error across acquisition for CT task 129
No-EK groups. Error bars are the SEM. The zero line represents
random sequence tracking error. Data below this line show decreased
tracking error and reflect implicit learning. All groups significantly
decreased repeating segment tracking error relative to that seen
for the random segments across acquisition.
Figure 5. Tracking error for repeated and random sequences at the 131
retention test, CT task No-EK groups. Error bars are SEM. A Main
Effect of Sequence showed that for all groups less tracking errors were
made for the repeated as compared to that for random segments.
The SMC groups’ repeated segment tracking error was reliably larger
than that for the HC, CB, or BG groups (Main Effect of Group).
Figure 6. Retention test learning score for tracking error for the No-EK 132
groups. Positive numbers show a decrease in repeating sequence RT
relative to the random condition and reflect implicit learning. No between
group differences were identified for the learning score; all of the groups
demonstrated implicit motor-sequence learning of the tracking task.
Figure 7. Time lag of repeated segment tracking across acquisition. 133
Error bars are the SEM. The CB group failed to change their time lag
of tracking relative to that seen for random segment tracking. The
HC, BG and SMC groups significantly decreased their time lag
relative to the random condition.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
X
Figure 8. Tracking accuracy across practice. Error bars are SEM. 135
All groups significantly improved their tracking accuracy relative to
random.
Figure 9. Random sequence tracking accuracy and time lag. 137
Error bars are the SEM.
Figure 10. Time lag and tracking accuracy for the repeated segment 138
at retention test. Error bars are SEM. All three focal stroke groups
had significantly longer tracking time lag as compared to HC. In
addition, the SMC group was significantly less accurate than the
HC group at this time.
Chapter 6:
Figure 1. RT change for the HC EK and No-EK groups. Both 161
significantly decreased RT with practice (Main Effect of Block).
However, larger decreases in RT were found for the EK group
(Knowledge by Block interaction). There were no between group
differences for retention or transfer tests. Error bars are SEM.
Figure 2. Change in tracking error for HC groups. A significant Main 165
Effect of Block was found showing that both groups (EK and No-EK)
benefited from practice regardless of the presence of explicit knowledge.
At retention and transfer tests there was no effect of explicit knowledge.
Error bars are SEM.
Figure 3. Time series analysis demonstrated that both HC groups 167
decreased their time lag (Main Effect of Block) across acquisition.
The EK group was able to decreased time lag of tracking more than
the No-EK group. There was a trend for differences in time lag for
the two HC groups at the retention test. A large effect size confirmed
this effect of explicit knowledge for tracking time lag (ES=0.73).
Tracking accuracy did not differ between the groups across acquisition
or at retention. Error bars are SEM.
Figure 4. Across SRT task acquisition both cerebellar EK and No-EK 169
groups benefited from practice (Main Effect of Block). However, the
CB EK group decreased their RTs to a larger extent than the No-EK
group (Main Effect of Knowledge). No reliable between group
differences were noted for the retention or transfer tests. Error
bars are SEM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 5. Both cerebellar groups decreased tracking error 171
across acquisition (Main Effect of Block). The large difference
between the groups evident in the magnitude of this change
was confirmed by a Knowledge by Block interaction, which
showed an advantage of explicit knowledge. However, this
between group difference was not maintained at retention.
No group differences were seen for transfer. Error bars are SEM.
Figure 6. There was not effect of practice or knowledge on the 172
CB groups’ tracking time lag or tracking accuracy. Error bars
are SEM.
Figure 7. Both the BG EK and No-EK groups changed their 174
RT across acquisition (Main Effect of Block). The No-EK group
demonstrated more decrease in RT than the EK group
(Group by Knowledge interaction). This between group difference
disappeared by retention. No differences were evident for transfer.
Error bars are SEM.
Figure 8. Across practice both BG EK and No-EK showed 175
decreased tracking error (Main Effect of Block). Again, larger
changes were demonstrated by those in the No-EK group
relative to the EK (Knowledge by Block interaction). This
between group difference was temporary and not seen at retention.
Again, there were no differences between the groups for the
transfer tests. Error bars are SEM.
Figure 9. Both the BG EK and No-EK groups decreased tracking 177
time lag with practice (Main Effect of Block). There was no effect
of explicit knowledge and no between group differences at retention.
Tacking accuracy did not differ between the groups. Error bars are SEM.
Figure 10. RT for the repeated sequence decreased for both 179
groups across practice (Main Effect of Block). Providing explicit
knowledge slowed RT relative to No-EK (Knowledge by Block interaction).
At retention, the SMC No-EK group was significantly faster (Main
Effect of Knowledge). There were no between group differences
for the transfer test. Error bars are SEM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
xii
Figure 11. Both SMC groups decreased tracking error (Main 180
Effect of Block). Explicit knowledge decreased the amount of
change shown by the SMC EK group (Main Effect of Knowledge).
There were no between group differences for the retention or
transfer tests. Error bars are SEM.
Figure 12. Both SMC groups decreased tracking time lag 182
across acquisition (Main Effect of Block). There was no effect
of knowledge on this decrease. There were no practice or
knowledge effects on tracking accuracy across acquisition
or at retention. Error bars are SEM.
Chapter 7:
Figure 1. Slope of change for the SRT and CT tasks across 206
practice by day and explicit knowledge condition.
(A) Healthy control group, (B) Cerebellar group, (C) Basal
ganglia group, (D) Sensorimotor cortical area group.
Across practice and groups, larger changes in performance
were seen for the SRT task as compared to the CT task
(Group by Task by Block interaction). Interestingly, there
was a negative slope of change shown by the BG EK group
on day three for both tasks (indicating worsening performance).
Conversely, at the same time the BG EK group demonstrated
positive changes for both tasks.
Figure 2. Percent change in repeated sequence as compared 209
to random sequence performance for both RT (SRT) and RMSE
(CT) at retention. All groups (HC, CB, BG, SMC) demonstrated
more change in performance for the simple task (Main Effect of Task).
Larger changes in performance ability for the repeated sequence
relative to random is demonstrated by larger percent change.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
xiii
ABSTRACT
THE INTERACTION BETWEEN EXPLICIT KNOWLEDGE
AND IMPLICIT MOTOR-SEQUENCE LEARNING
FOLLOWING FOCAL BRAIN DAMAGE
To determine the interaction between explicit knowledge and implicit
motor-sequence learning following focal brain damage, 27 individuals with
unilateral stroke affecting the cerebellum (CB; n=7), basal ganglia (BG; n=10),
or sensorimotor cortical areas (SMC; n=10), and ten healthy controls (HC)
practiced two unilateral motor-sequencing tasks. All individuals with focal
stroke used their less involved arm; the HC group was matched for arm use.
During practice (50 trials / day / task 3 days, day 4 retention) there were
repeated and random sequences; implicit learning was measured as the
difference in performance between these two. Half of the participants were
not informed of the repeating sequences (No-EK), half received incremental
explicit knowledge (EK).
All No-EK stroke groups demonstrated implicit motor-sequence learning
(Sequence Effect, p=0.026) and comparable performance change (random -
repeated sequence) at retention relative to the HC group for both tasks (Group
Effect, p=0.313). Despite this, none of the focal stroke groups were able to
reduce response times or tracking errors to the same degree as the HC group,
and each demonstrated specific deficits.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
xiv
Explicit knowledge benefited acquisition performance for HC (SRT task,
Knowledge Effect, p=0.020) and CB groups (both tasks, Knowledge Effect,
p=0.007). However, the CB groups were not able to reduce tracking time lag
across practice. Explicit knowledge severely disrupted acquisition
performance of the BG group (Knowledge X Block interaction, p=0.023), and
SMC group for both tasks (Knowledge Effect, p=0.050). Further, the SMC
group was not able to improve their tracking accuracy with practice.
The ability of the CB groups to benefit from EK was likely due to the
preserved neural network interconnecting prefrontal cortex, basal ganglia,
premotor cortex (PMC), thalamus, and critically, the undamaged cerebellar
hemisphere. However, it appears that bilateral CB activation is necessary to
alter tracking timing. Unilateral BG damage severely disrupted the use of EK,
demonstrating that bilateral BG function is important for the incorporation of
explicit information into the motor plan, as are regions within the sensorimotor
cortex (e.g. PMC). Performance deficits following unilateral stroke, even when
invoking the undamaged hemisphere, demonstrate the importance of bilateral
neural function for implicit-motor-sequence learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PREFACE
“Every action is an idea before it is an action, and perhaps a feeling before it is
an idea, and every idea rests upon other ideas that have preceded it in time. ”
-Wallace Stegner
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
CHAPTER 1
Introduction and Overview
The Question
Since Scoville and Milner’s (1957) original description of the well-known
amnesic patient H.M., which identified severely disrupted declarative memory
following bilateral medial temporal lobe ablation, there has been acute interest
in the neuroanatomic systems that subserve and functionally express learning
and memory. Further study of H.M. potentiated a greater understanding of the
functional and neuroanatomic distinctions between the declarative and non
declarative memory systems. Over the past 45 years, numerous paradigms
have been developed that successfully dissociate the declarative and non
declarative, or procedural, memory systems. From these, it is clear that
declarative memory is distinctly subserved neuroanatomically by discrete
regions within the medial temporal lobe.
The identification of the particular neuroanatomic structure(s) that
mediate the non-declarative memory system has been less successful. No
single focal lesion completely eliminates the capability to develop procedural
memories, although many may severely disrupt these processes. Combining
the results of imaging studies (e.g. functional magnetic resonance imaging
(fMRI) and positron emission tomography (PET)) with investigations into the
impairments demonstrated by various patient groups (e.g. Parkinson’s
disease, cerebellar ataxia, focal stroke), and work with animal models has
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
identified the putative neural network supporting procedural learning and
memory. This neural network is highly distributed and likely includes the
cerebellum, basal ganglia, thalamus, and regions within the sensorimotor
cortical areas (e.g. primary motor cortex, premotor area, supplementary motor
area, pre-supplementary motor area). Despite these advances, the particular
roles of each neural region within the procedural learning network, and their
relative contributions during the practice and learning of different tasks, remain
unclear.
It is likely that in an intact and normally functioning neural system, the
declarative and procedural systems function together, complementing one
another during motor learning. How these two neural and functional systems
relate to one another continues to be unclear; there have been few systematic
investigations into the manner in which the declarative and procedural
systems interact with and influence one another (both positively and
negatively) during learning. During early learning when it is available and
salient, information from the declarative system (e.g. instructions, feedback)
may be used to guide procedural performance during motor skill learning.
However, many questions remain concerning the influence of declarative
knowledge on procedural learning including (to name a few) the timing and
method of delivery of information, the content of instructions, and the relative
impact of declarative knowledge following brain damage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
Finally, of those studies focused on the procedural system, none have
systematically evaluated the effect of task type on skill learning. Intuitively it
would appear that as task complexity increases procedural learning would
slow, necessitating more practice, and using more neural resources.
However, these assumptions have yet to be verified.
Therefore, this dissertation was centered on three distinct yet
interrelated purposes (specific aims):
1) To investigate how three brain regions (cerebellum, basal ganglia, and
sensorimotor cortical areas) distinctly contribute to and coordinate
procedural motor-sequence learning;
2) To determine the relative impact of declarative knowledge on
procedural motor-sequence learning following focal, unilateral brain
damage to the cerebellum, basal ganglia, or sensorimotor cortical
areas, and;
3) To determine if there is an interaction between procedural motor-
sequence learning and task type.
Overview
This dissertation begins with a discussion regarding the constructs of
declarative and procedural memory and learning. Next a comprehensive
review of the neural and functional characteristics of the declarative and
procedural memory and learning systems is presented; this includes a
discussion of the existing data that describe the interactions between these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
two memory systems (Chapter 2). In Chapter 3, neuropsychological data that
demonstrate procedural learning are discussed with respect to motor learning
following stroke and the principles of motor learning. Following these reviews,
a chapter describing the general methods used in this dissertation is
presented (Chapter 4). Next, there are three chapters that describe and
discuss results. The first concerns the effect of unilateral focal stroke in the
cerebellum, basal ganglia, and sensorimotor cortical areas on procedural
motor-sequence learning (Chapter 5). Next, data that evaluates the impact of
declarative knowledge on procedural motor-sequence learning is presented
(Chapter 6). Experimental results are discussed that describe and quantify the
interactions between task type, declarative knowledge, and stroke location
during procedural motor-sequence learning (Chapter 7). Last, a general
discussion and summary is presented (Chapter 8).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
CHAPTER 2
Declarative and Procedural Systems:
Neuroanatomy and Function
Introduction
The purpose of this chapter is to review the current literature relevant to
the neuroanatomic and functional divisions between the declarative and
procedural memory systems. First, distinctions will be made between the
definitions of and constructs for the declarative and procedural memory and
learning systems. Next a review of the critical neural substrates subserving
declarative (medial temporal lobe and pre-frontal areas) and procedural
learning (cerebellum, basal ganglia, sensorimotor cortical areas) as well as
their inter-relationships will be discussed. Last, behavioral and neuroanatomic
data that describe the possible interactions during learning between the
declarative and procedural systems will be considered. At the conclusion of
each section the hypotheses that form the basis of this dissertation will be
presented.
Declarative and Procedural Memory
Divisions among Learning and Memory Systems
Learning and memory are not singular processes but are composed of
many separate abilities. The broad category of long-term memory can be sub
divided into two main types - declarative and non-declarative or procedural
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
(Squire, 1987). For the purpose of this discussion the terms declarative and
procedural will be used to discuss hypothetical memory and learning systems,
while the terms explicit and implicit refer to a distinction between tasks. In
general, it is assumed that explicit tasks draw on the declarative, whereas
implicit tasks engage the procedural memory system (Willingham, 1998).
Explicit learning may be assessed directly via memory tests that tap
subjective, factual knowledge of the task as well as recognition and recall.
However, implicit learning is measured indirectly, as a subject’s responses are
altered or facilitated by the acquisition of knowledge about the structural
properties of the task itself. At this level task becomes critical as it must be
able to guide the subject’s responses even if s/he has had no prior experience
or exposure (Buchner & Wippich, 1998; Frensch, 1998; Willingham, 1998).
Another set of constructs critical to this discussion concern the
distinctions between learning and memory. Learning refers to acquisition of
knowledge about a task with practice or exposure. During explicit learning,
factual knowledge is typically acquired. The knowledge that is gained during
implicit learning concerns regularities for a task or relationships within a
sequence of objects or events that are not intuitive or obvious. In contrast,
memory refers to the accumulated effects of past experiences with a set of
facts, objects or events. Explicit memories are assessed in much the same
fashion, as is explicit learning, via evaluation of recognition subjects’ and / or
recall of factual information. Specific to implicit learning, however, the effect of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
prior experiences or practice may be observed for a task via some
performance measure (e.g. decreased reaction times, fewer errors) despite
the fact that subjects are not specifically asked to relate their current
performance with a previous episode (Buchner & Wippich, 1998; Frensch,
1998).
Long-Term
Memory
Declarative
Procedural
Facts Events Priming Skills Associative Nonassociative
Habits Learning Learning
Adapted from Squire, 1987
Figure 1. Classification of declarative and procedural memory and learning
systems.
Declarative and procedural memory differ fundamentally in the “type” of
the information each stores and uses. Consequently, each is mediated by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
separate neural systems. Declarative knowledge is the conscious memory of
facts, events, and episodes. It may be formed very quickly (even in one
exposure) and is directly accessible to conscious recollection (Squire, 1987;
Figure 1). Declarative memory guides high level cognition where decisions
are based on complex rules and information. Multiple studies have
demonstrated that declarative memory is severely impaired by damage to the
medial temporal lobe (hippocampus, and adjacent cortex including entorhinal,
perirhinal, and parahippocampal structures; Scoville & Milner, 1957; Milner,
Corkin, & Teuber, 1968; Squire & Zola-Morgan, 1991).
Procedural learning is the capacity to acquire skill through physical
practice and is not directly accessible to conscious recollection as facts or
data. The development of procedural knowledge occurs incrementally, with
practice, over a period of time and exposure. Strong evidence for the
dissociation of declarative and procedural memory comes from the finding that
individuals with medial temporal lobe damage suffer profound declarative
deficits while retaining procedural memory capabilities. Scoville and Milner
(1957) first described impairments in declarative memory following bilateral
medial temporal lobe damage in one investigation of the preserved abilities of
the amnesic patient H.M.. In an attempt to ameliorate intractable seizures,
H.M. underwent bilateral surgical removal of the medial temporal lobes.
Following this procedure H.M. completely lost all declarative memory function
despite retaining normal IQ, short-term memory, and memory for early life
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
events. Despite his overt declarative memory deficits it was noted that with
practice H.M. was able to learn and retain skill for various motor tasks. For
example, over the course of several days of practice H.M. was able to
significantly decrease the number of errors made in a mirror-tracing task,
despite a total lack of conscious awareness of previously having performed
this motor task (Corkin, 1968; Milner et al., 1968). Further H.M.’s ability to
reduce errors over practice was similar to that seen for healthy control
subjects. This finding of preserved procedural learning and memory in the
absence of declarative memory function has been confirmed by numerous
reports of procedural learning in individuals with amnesia (for examples see
Cohen & Squire, 1981; Nissan & Bullemer, 1987; Reber & Squire, 1998).
Procedural learning may be further sub-divided into motor skills and
habits (e.g. sequence learning, mirror tracing), priming (e.g. word-completion),
associative learning (e.g. classic and operant conditioning) and non-
associative learning (e.g. habituation, sensitization; Squire, 1987; Doyon
1997a; Figure 1). Each of these types of procedural learning and memory are
preserved following amnesia (Squire, 1987). Additional criteria defining
procedural learning include that 1) knowledge gained is not accessible to
conscious awareness, 2) learning is an incidental consequence of information
processing during practice and does not involve any conscious hypothesis
testing, and 3) learned information is more complex than just a simple
stimulus-response association (Segar, 1994; Reber 1976, 1989). The focus of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
this dissertation will be on a sub-set of the procedural system - motor-skill (i.e.
sequence) learning.
Robust procedural learning in individuals without neurologic pathology
has been demonstrated using multiple paradigms (Nissan & Bullemer, 1987;
Reber 1989; Maylor & Rabbitt, 1995; Lewicki, Hill, & Bizot, 1988; Knowlton,
Mangles, & Squire, 1996; Knowlton & Squire, 1995). Nissan and Bullemer
(1987) who developed the serial reaction time (SRT) task first separated
implicit motor-sequence learning from explicit learning. The SRT task was the
first paradigm that experimentally isolated the implicit from the explicit learning
and memory system (even in individuals without neural pathology). In the
SRT task, subjects repeatedly are cued to respond to lights when they are lit
by pushing corresponding buttons. Unknown to the subject, s/he is following a
repeating sequence of cues or stimuli varying from 5 to 15 elements in length.
With practice, subjects’ responses during the sequence become increasingly
faster (as demonstrated by either decreased reaction times (RT) or response
times, compared to a random condition without a repeating sequence).
Nissan and Bullemer (1987) developed this paradigm using a 10-element
sequence to demonstrate that healthy individuals and individuals with amnesia
were able to significantly decrease RT over practice (80 trials of responses).
During SRT learning, healthy individuals may gain explicit awareness of
the sequence in the latter phases of practice. Occasionally, as implicit motor-
sequence learning of the SRT task becomes robust, individuals may even
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
become explicitly aware of the specific composition of the sequence. Implicit
learning of the SRT task has also been shown for sequences whose
composition are governed by probabilities (Lewicki, et al., 1988; Stadler,
1989), deterministic noisy finite-state grammars (Cleermans & McClelland,
1991; Jimenez, Mendez, & Cleermans, 1996), and fixed repeating locations
(Nissan & Bullemer, 1987; Wulf & Schmidt, 1997). Serial reaction time
sequence composition may be manipulated to facilitate faster implicit (and
perhaps explicit) learning. For example, when sequence composition is
probabilistic or rule governed (e.g. red always precedes blue), implicit SRT
learning occurs with less practice than when sequences are ambiguous or
random (Pascal-Leone, Grafman, Clark, et al., 1993; Curran & Keele, 1993).
Additionally, increased sequence length, complexity (Curran & Keele, 1993;
Wulf & Schmidt, 1997) and distraction (Stadler & Roediger, 1998; Curran &
Keele, 1993; Jimenez Mendez, & Cleermans, 1996; Lewicki, Czyzewska, &
Hoffman, 1987; Nissan & Bullemer, 1987; Stadler, 1995) delay the acquisition
of explicit knowledge of the sequence for healthy individuals. Across the
procedural learning literature, however, few studies have employed techniques
(i.e. retention tests; Salmoni, Schmidt, & Walter, 1984; Schmidt & Lee, 1999)
to separate immediate performance effects from relatively permanent changes
in responding (i.e. learning) (for exceptions see Frensch, Lin & Buckner, 1998;
Frensch, Wenke & Runger, 1999; Wulf & Schmidt, 1997; Boyd & Winstein,
2001). Therefore, it remains unclear how much actual learning occurs
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
following implicit motor-sequence practice. One goal of this dissertation is to
employ a retention test design to more accurately gauge the relatively
permanent changes in responding following implicit motor-sequence practice.
The neural substrates subserving the declarative and procedural
learning and memory systems are discussed next.
Neural Substrates Subserving Declarative Learning and Memory
Evidence from the study of H.M., along with numerous other patient
populations, imaging and animal studies, demonstrate that declarative and
procedural learning are functionally and neuroanatomically distinct. Given that
declarative memory is independent from procedural memory what are the
neural substrates that subserve each of these systems? First, a review of the
neural regions for declarative memory will be presented. These areas most
critically include the medial temporal lobe (hippocampus, and adjacent cortex
including entorhinal, perirhinal, parahippocampal structures; Scoville & Milner,
1957; Milner, Corkin, &Teuber, 1968; Squire & Zola-Morgan, 1991), as well as
regions in, and associated with, the frontal cortex (Grafton, Hazeltine, & Ivry,
1995; Hazeltine, Grafton, & Ivry, 1997).
Medial Temporal Lobe
A critical feature of the declarative memory system is the capability to
transfer new, short-term memories into long-term ones. Damage to the neural
regions that support declarative learning and memory may disrupt the ability to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
form new long-term memories. This condition is known as amnesia, and may
be a side effect of surgery (as in the case of H.M.) and also is a common
consequence of encephalitis. It has been repeatedly demonstrated that
damage to the limbic association areas in the medial temporal lobe
(hippocampal formation, perirhinal and parahippocampal cortices) severely
disrupts the declarative memory system (for examples see Nissan & Bullemer,
1987; Reber & Squire, 1998). In one such demonstration, Knowlton and
Squire (1996) reported that amnesic patients were able to invoke the
procedural memory system through implicit learning of a probabilistic
classification task without demonstrating any explicit memory for the training
session. Other research has also reported that damage to the medial
temporal lobe disrupts only declarative memories for learning without affecting
procedural motor-sequence learning. In one such study, Reber and Squire
(1998) actually found a procedural learning benefit for individuals with amnesia
over age-matched healthy controls for implicit learning of the SRT task.
Experimental evidence from animal models has been able to more
precisely identify the regions within the medial temporal lobe that support the
declarative learning and memory system. In a series of experiments (using a
monkey model) it has been shown that it is isolated lesions in the
hippocampus, perirhinal, and parahippocampal cortices that severely disrupt
declarative (spatial) memory, while lesions in the amygdala alone or in
combination with hippocampal damage do not (Suzuki, Zola-Morgan, Squire,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
& Amaral, 1993; Zola-Morgan, Squire, Clower, & Rempel, 1993; Zola-Morgan,
Squire, & Ramus, 1995). Similarly, rats with lesions in the hippocampus have
been shown to be impaired in tasks that require declarative (spatial and
relational) memory (Packard & McGaugh, 1992).
Recent imaging studies also demonstrate focal activity within the medial
temporal lobe during specific declarative memory tasks. Activity in the left
medial temporal lobe has been shown for the encoding of words (Wagner,
Gabrieli, & Verfallie, 1997), while encoding of pictures invokes bilateral medial
temporal lobe activity (Gabrieli, Brewer, Desmond, & Glover, 1997). Retrieval
of semantic information also activates medial temporal lobe. Stark and Squire
(2000) imaged (fMRI) 21 healthy individuals as they identified words and
pictures as “known” or novel. In this task bilateral hippocampal activity was
increased during the recognition memory task. Thus, it is apparent that the
medial temporal lobe is essential for both encoding and retrieval of declarative
information.
Prefrontal Cortex
The prefrontal cortex has been shown to be critical for the storage of
long-term declarative information. In particular, prefrontal areas support
working memory - the temporary storage of information necessary to guide
and modify future actions (Saper, Iverson, & Frackowiak, 2000; Pochon, Levy,
Poline, Crozier, Leheicy, et al., 2001). Neurons within the prefrontal cortex
increase their activity as soon as a visual stimulus is provided. Importantly,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
these neurons will continue firing throughout a delay period. Thus, prefrontal
neurons likely are holding information about a task during periods of delay.
This action is critical for working memory and disrupting or halting prefrontal
cortical activity during a delay period essentially causes forgetting of
information (visual). Additionally, Goldman-Rakic (1992) demonstrated that
neurons in the prefrontal cortex are not only used to direct attention during and
following a delay, but that the activity of certain neurons may also be location
specific.
Various imaging studies have shown activity in the dorsolateral
prefrontal cortex (DLPFC) when implicit motor-sequence learning is being
guided by conscious hypothesis testing (i.e. explicit) such as occurs during
trial and error learning (Toni, Krams, Turner, & Passingham, 1998; Sakai,
Hikosaka, Miyachi, Tajino, Sasakai, & Puta, 1998), as well as when explicit
knowledge is either gained (Grafton et al., 1995) or provided (Honda, Dieber,
Ibanez, Pascal-Leone, Zhaung, & Hallett, 1998). In each of these studies it
appeared that factual information or declarative knowledge was being held in
working memory and subsequently integrated with other visual-spatial
information in order to guide motor performance. Sakai et al., (1998) scanned
(fMRI) seven normal subjects over four periods of time and demonstrated that
the DLPFC was more active during the early acquisition stage of a practice
session of a key-press sequence (learning by trial and error). Later in
practice, after the subjects had demonstrated via behavioral criteria (e.g.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
decreased response times and errors) that they had learned the sequence,
neural activity shifted to parietal regions, suggesting that the parietal cortex is
more involved in retrieval of known motor-sequences. In addition, Pochon et
al. (2001) reported significant DLPFC activity when subjects were required to
mentally prepare a sequential action.
Therefore, it is most likely that the formation of declarative memory
during sequence learning is a multifaceted process supported by several
neural structures. Declarative information is probably first processed by the
association areas (pre-frontal for visual, limbic for auditory, and parieto
occipital for somatic information). The parahippocampal, perirhinal, and
entorhinal cortices then process short-term memories. The entorhinal cortex
is the primary input to the hippocampus (projecting to the dentate gyrus). It is
also the output pathway of the hippocampus and mediates the transfer of
information as it is conveyed back to the association cortices (Kandel,
Kupfermann, & Iverson, 2000). As patients with hippocampal and
parahippocampal damage do not lose their long-term memories (Milner, 1966;
Glisky & Schacter, 1987), long-term memory storage likely occurs elsewhere.
It appears that specific association areas store particular memories in a
modality-specific fashion. For example, when learning a motor task (the SRT
task), even if explicit knowledge is available, neural activity increases in
regions associated with movement skills (i.e. prefrontal, premotor, and parietal
cortices). When attention is divided (e.g. during dual task practice) cognitive
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
resources are reduced leaving little processing capability to devote to the
acquisition of explicit knowledge and increased activation in the prefrontal
motor associated regions is not seen (Grafton et al., 1995; Hazeltine et al.,
1997). Most critical to the formation of declarative memories, however, are the
medial temporal lobe structures (hippocampus and adjacent cortex) as
damage to these areas eliminates both the ability to encode and retrieve newly
formed declarative knowledge (Table 1).
As there are a large number of representative studies concerning motor
skill learning in animal models, as well as in humans using both imaging
techniques and individuals with specific neural pathology three summary
tables (1, 2, and 3) have been provided. These contain information regarding
various study designs, experimental models and summarized pertinent
findings, and will be referred to throughout this chapter.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1. Overview of Representative Animal Studies Using Motor Learning Paradigms
Authors Species Task Conditions Results Significance
Nixon &
Passingham
2000
(exp. 3)
6 cyno-
mologus
monkeys
SRT task
(4-
element)
1) Following bilat.CB
lesion (n=3) trained
on sequence
(matched to non
lesion group, n=3)
1) Monkeys with CB lesions
demonstrate some implicit motor
sequence learning, however, this
learning was not at same rate or to
same degree as non-lesion monkey
group
CB plays a role in
automation of motor
sequence learning.
Matsumoto
et al., 1999
3 macaque
monkeys
Sequential
push
button
motor task
(2X5)
Depleted dopamine
unilaterally using
MPTP, tested
animals using arm
contra- and
ipsilateral arms
(n=2). Trained first
then dopamine
depleted (n=1)
1) With arm ipsilateral to dopamine
depletion monkeys show motor
learning
2) With arm contralateral to depletion
monkeys do not show learning.
Monkey also unable to switch between
arms.
3) Monkey that was trained first able to
eventually relearn sequence.
Striatum function
during initial learning
of sequences
allowing them to run
as a single motor
plan. Striatum is also
somewhat involved in
retrieval of program.
Lu et al.,
1998
2 macaque
monkeys
Sequential
push
button task
(2X5)
Pre-learned task.
Muscimol injections
to CB nuclei.
Learning deficits only when ipsilateral
dorsal and central dentate inactivated
Dentate important for
storage and retrieval
of sequences.
Nakamura et
al., 1998
2 macaque
monkeys
Sequential
push
button task
(2X5)
Cellular recordings
during new hyperset
learning and during
performance of well
learned hyperset
sequences.
1) More Pre-SMA cells active during
acquisition
2) More SMA cells active in executing
learned sequences
Pre-SMA more
important for learning
of new sequences
Miyachi et
al., 1997
2 macaque
monkeys
Sequential
push
button task
(2X5)
Local inactivation of
anterior and
posterior striatum
(muscimol injection).
1) Inactivation of anterior striatum
(caudate and putamen) disrupted new
sequence learning 2) Inactivation of
posterior striatum (middle posterior
putamen) impaired execution of
previously learned sequences
Anterior striatum important
for new learning (via
DLPFC) / posterior striatum
important for retrieval of
known sequences (via
SMA)
0 0
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Authors Species Task Conditions Results Significance
Mushiake &
Strick, 1995
2 macaque
monkeys
Sequential
arm
movement
s
Movements made
under tracked and
remembered
conditions.
Intracellular
recordings of globus
pallidus (GP)
neurons.
2/3 of GP neurons found to be task
dependent (active during hyperset
sequence) and 65% of these neurons
more active during the remembered
condition than the tracking condition.
GP neurons encoding
detailed spatio-temporal
information that
characterizes the sequence
being learned. May play a
role in serial ordering of
responses
Tanji &
Shima, 1994
2 macaque
monkeys
Sequential
arm
movement
s
Recorded from
caudal SMA and M1
During sequence execution found
sequence specific neurons in SMA but
not in M1
SMA neurons providing
information about the order
of upcoming movements -
used to plan ahead
Kermandi et
al., 1993
1 rhesus
monkey
Sequential
push
button task
with visual
fixation
Intracellular
recordings made in
caudate nucleus
Specific caudate cells active for
particular sequences
Caudate may be delivering
instructions for next
movement
Berridge &
Whitshaw,
1992
40 Long-
Evans rats
Innate
Grooming
chain
activity
Removed:
1 )Striatum &
neocortex
2)M1 & SMA
3)SMA
4)Neocortex
5)CB
Removal of striatum produced
irreversible damage in coordination
among elements of grooming chain.
However, all individual movements
could still be performed. Cortical and
cerebellar removal resulted in
temporary deficits in grooming
sequences.
Striatum has a unique role
in coordinating among the
elements of the sequence.
Note: Studies are listed in chronological order with most recent first. (M1=primary motor area, SMC=sensorimotor cortex,
SMA=supplementary motor area, PMC=premotor cortex, SSA=supplementary sensory area, CB=cerebellum, R=right, L=left).
C D
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 2. Synopsis of Representative Imaging Studies in Humans during Explicit and Implicit Learning and Memory Tasks
Authors Subjects Scan
Type
Task Timing /
Conditions of
Scan
Results - Activated Brain Regions Comments
Stark &
Squire, 2000
22 HC fMRI Semantic
word
retrieval
task
During word
recognition
Bilateral hippocampus identified
during recognition
Hippocampus both
encodes and
recognizes
semantic
information
Boeckner et
al., 1998
7 HC PET Sequence
execution
Pre-learned simple
and complex
sequences
Rostral SMA increased activity with
more complex sequence, caudal SMA
active for motor execution
Function of SMA
may be segregated
and dependent on
complexity
Honda et al.,
1998
21 HC PET SRT task
using R-
hand
1) Explicit Learning
Condition
2) Implicit Learning
Condition
1) R dorsolateral prefrontal ctx., L
SMA, L thalamus, Bilat PMC
2) R SMC
1) Correlated brain
increasing rCBF
with decreasing RT
2) No view of CB
Rauch et al.,
1998
10 HC fMRI SRT task 1) Early Learning
2) Late Learning
1) Early learning - significant
increase in R caudate and putamen
with deactivation in thalamus
2) Late learning - increase in
activation of thalamus
Appeared that as
BG activation
increased, thalamic
activation
decreased
Sakai et al.,
1998
7 HC fMRI 10-element
sequence
by trial and
error
1) Early learning
2) Intermediate
3) Late learning
1) In early learn DLPFC and pre-SMA
active
2) Frontal regions less, parietal
regions more active
3) Precuneus and intraparietal sulcus
active in late learning
Acquisition of
sequence learning
under explicit
conditions uses
frontal areas
t o
o
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Authors Subjects Scan
Type
Task Timing /
Conditions of
Scan
Results - Activated Brain Regions Comments
Toni et al.,
1998
3 HC fMRI 8-element
sequence
by trial and
error
Scanned
intermittently for 40
minutes in 1
session. Bilat. key
press task
1)DLPFC, ant. cingulate cortex active
early, but not once the task was over
learned
2)PMC active early then decreased,
while SMA not active early but
increased. No change in motor cortex
3)L Caudate active early. Putamen
active bilat. early and continued
throughout session
4)CB active bilat. early
Trial and error
learning invokes
explicit memory
system. Transition
from PMC early to
SMA late reflects
shift from explicit to
implicit strategy
vanMier et 32 HC PET Maze Traced with both R- 1) For both R- & L-hand performance Due to scan
al., 1998 R-Handed Tracing & L-hands see increased activation in SMA,
while activation decreased in R PMC,
R inf. and superior parietal cortex, L
lateral CB
2) M1 and Anterior CB switch with
performance hand
resolution could not
assess laterality of
SMA activation
Hazeltine,
Grafton, Ivry,
1997
11 HC PET SRT task
(R-hand)&
tone count
or SRT task
alone
1) Dual Task
(explicit awareness
prevented)
2) Single Task
1) L M1, L S1, L putamen
2) R prefrontal ctx., R PMC
1) Full view of CB
not available
2) In single task
7/11 gained explicit
awareness
Doyon et al.,
1996
14 HC PET SRT task
using R-
hand
1) New Learning
2) Highly Learned
3) Explicit
Knowledge
1) R pre-striate ctx., R CB, L SMC, L
SMA, L PMC, L thalamus
2) R ventral striatum, R dentate, R
pre-striate ctx., R posterior parietal
ctx., Bilat. anterior cingulate ctx.
3) R CB, L mid ventrolateral frontal
area
CB activation in
explicit knowledge
condition likely seen
because sequence
is not yet fully
automated
t o
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Authors Subjects Scan
Type
Task Timing /
Conditions of
Scan
Results - Activated Brain Regions Comments
Grafton,
Hazeltine,
Ivry, 1995
12 HC PET SRT task
(R-hand)&
tone count
or SRT task
alone
1) Dual Task
2) Single Task
1) L M1, L SMA, L putamen, L rostral
prefrontal ctx.
2) R dorsolateral prefrontal ctx., R
PMC, R putamen
1)ln single task 7/12
gained explicit
awareness
Jenkins et al.,
1994
12 HC PET Key-press
sequence
learned by
trial and
error
1) Rest
2) Pre-learned
sequence
3) New sequence
1) Both sequences (new and old)
evoked SMC
2) New sequence activated prefrontal
cortex, PMC, parietal association
areas, putamen, CB. Hippocampus
decreased its action over practice.
CB involved in
automation of
sequence, putamen
important for both
learning and
retrieval.
Kawashima
et al., 1994
20 HC PET Reaching to
7
remembere
d spatial
locations
1) Early learning
2) late learning
1) During reaching and the delay
period PMC and M1 were active.
Within M1 fields of neurons active
during delay were adjacent to fields
active during reach
M 1 fields engaged
in efferent control of
action. PMC active
for sensory
guidance of action
Seitz et al.
1994
8 HC PET R-handed
trajectory
writing
1) Early Learning
2) Over Learning
1) R Dentate nucleus, L posterior CB,
LM1
2) L posterior cerebellum, R lateral
cerebellum, Bilat PMC, L M1, SMA, L
S1
New learning
involves the CB,
while overlearning
activates the PMC
Grafton et al.,
1992a
6 HC PET Pursuit
Rotor (R-
hand)
After practice L M1, L SMA, L pulvinar thalamus No view of CB
Grafton et al.,
1992b
18 HC PET Visuomotor
Tracking
(R-hand)
1) Increasing
temporal complexity
2) Increasingly
spatial complexity
1) Bilat. SMA
2) Bilat. dorsal parietal lobe, bilateral
precunate
CB activation not
altered by these
conditions, anterior
CB activation level
changed with motor
effector
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Authors Subjects Scan
Type
Task Timing /
Conditions of
Scan
Results - Activated Brain Regions Comments
Seitz &
Roland, 1992
9 HC PET Complex R-
hand finger
movement
sequence
1) Initial Practice
2) Advanced
Learning
3) After Learning
had occurred
1) L M1, L SMA, L SSA, L S1
2) Rant CB, L S1, L PMC, L M 1
3) L ant CB, L M1, L S1, L PMC, L
SSA
All data acquired in
one-session and
likely reflects
performance rather
than learning
Note: Studies are listed in chronological order with most recent first. (HC=healthy controls, M1 =primary motor area, S1 =primary sensory
area, SMC=sensorimotor cortex, SMA=supplementary motor area, PMC=premotor cortex, SSA=supplementary sensory area,
CB=cerebellum, DLPFC=dorsal lateral prefrontal cortex, SRT=serial reaction time, R=right, L=left)
N >
co
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 3. Summary of Relevant Human Patient Group Studies Using Implicit Motor Sequence Learning Tasks.
Authors Subjects Paradigm Conditions Results Significance
Gentilucci et 1 SMA CVA Reach- Reach-grasp trials Individual with SMA lesion only SMA assembles the
al., 2000 5 HC grasp,
reach-
grasp-
pace
tasks
compared with
trials where
placement of object
was added
impaired by the position of the reach-
grasp target.
sequence; the BG updates
and stores the motor plan
Helmuth et 24 PD; 24 SRT task Varied location of Individuals with PD can learn PD patients able to learn
al., 2000 HC (motor
and
perceptual
)
sequence: 1)
location and
sequence same, 2)
location same, &
3) sequence same
sequence of spatial locations but are
impaired at learning motor-response
sequences
SRT sequence perceptually
but are impaired in learning
motor response sequence
Vakil etal.,
2000
16 focal BG
CVA
16 matched
HC
1) SRT
task, R
hand only
2) SRT
task non
motor
version, R
hand
1) 10-element
ambiguous
sequence practiced
4 blocks (40 trials),
1 block random,
retention test
2) 10-element
ambiguous
sequence watched
occasional (30%)
motor response
1) Individuals with focal BG stroke
learn significantly less than HC group
for both motor and non-motor SRT
tasks
2) Individuals with BG stroke and HC
showed similar levels of EK of
practiced sequence
All subjects practiced SRT
task with R hand. However,
11 CVAs R-side and 5 L-
side.
Doyon et al.,
1998
1)11 PD; 11
CBD; 9
CVA; 9, 11,
14 HC
2)15 PD; 12
CB; 8 CVA;
15,12,11 HC
1) Dual
task SRT
2) SRT
task
1) 10-element
ambiguous
sequence practiced
8 blocks (240
times), bilateral
2) Re-tested 8-16
months (4 blocks)
1) All subjects show implicit learning,
however, they were slower for dual
than single task. Even after 240 reps
PD and CB patients still had not
automated sequence 2) PD and CB
patients show much less retention
than Frontal and HC subjects.
1) Striatum and CB
important for automation of
sequence learning
2) Striatum and CB sub
serve long term retention of
skill and are important for
automation
N >
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Authors Subjects Paradigm Conditions Results Significance
Gomez-
Beldarrain et
al., 1998
14 focal
CB CVA
10
matched
HC
SRT task 2 separate 10-
element ambiguous
sequence practiced
3 blocks (30 trials),
2 blocks random
1) Individuals with CB CVA do not
learn implicitly when using hand
ipsilateral to lesion but do learn
implicitly when using hand
contralateral to lesion
CB important for implicit
learning but its function
appears to be highly
lateralized to the ipsilateral
hemisphere
Doyon et al.,
1997
14 PD; 12
CB; 9
frontal
CVA; 15,
12, 14 HC
SRT task 10-element
ambiguous
sequence, 40
times/day, 6 days,
bilateral
PD (stage 2) and CB patients impaired
in implicit learning of SRT task
compared to HC. PD patients more
impaired for later learning. Frontal
lobe damage does not impair SRT
task learning.
Striatum and CB important
for implicit learn,
automating sequence (later
learning) appears to rely on
BG and CB
Dominey et
al., 1997
7 PD; 6 HC SRT task,
Tower of
Toronto
3 separate 24-
element-spatially
organized (8
locations)
sequences. Each
governed by same
rule. Explicit rule
information given to
participants
With explicit rule knowledge
individuals with PD able to
demonstrate implicit sequence
learning. Further, individuals with PD
were able to transfer this learning to
other sequences governed by same
rule.
Self directed learning is
impaired in individuals with
PD, while role based
learning is not.
Molinari et
al., 1997
8 focal
cerebellar
sroke
SRT task 8-element, 10-
element, and full
explicit knowledge;
subjected practiced
with both hands
separately
Severely impaired implicit SRT task
learning using both the hand contra-
and ipsilateral to brain damage. Full
explicit knowledge prior to practice
significantly reduced response times
and facilitated implicit learning.
Cerebellar lesions severely
impair implicit learning,
however, explicit knowledge
able to benefit SRT task
performance.
Ackermann
et al., 1996
1 individual
with focal L
SMA
damage
SRT task,
mirror
tracing
task
10-element
ambiguous
sequence (5-
choice), 7 blocks
(1=10 trials)
Impaired implicit learning of both
tasks.
SMA important for initiation
and learning of implicit
sequences (e.g. without
explicit knowledge)
ro
cn
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Authors Subjects Paradigm Conditions Results Significance
Knowlton &
Squire, 1996
12
amnesic,
20 PD, 15
HC
Probabilisti
c
classificati
on task
Task practice and
explicit knowledge
questionnaire
1) Amnesic patients learn probabilistic
task, have no explicit awareness
2) PD patients gain explicit knowledge
of task, do not demonstrate implicit
learning of probabilistic task
Neostriatum essential for
implicit learning
Jennings,
1995
9 PD, 8 HC Cued SRT
task
Cue either accurate
or no information; 3-
element sequence,
360 trials
PD patients could use predictive
information to change sequences but
not to prepare several responses in
advance.
PD impairs coordination
among motor program and
execution
Pascal-
Leone et al.,
1993
20 PD, 15
CB
degenerati
on, 30 HC
SRT task 8 to 10 elements, 7
blocks of 100 trials
in 1 day
1) PD patients learned SRT task,
needed more practice, could use
explicit knowledge to benefit
performance
2) CB patients did not show implicit
learning , could not use explicit
knowledge
BG may use explicit
information from prefrontal
cortex. CB critical for
implicit learning.
Scoville &
Milner, 1957
1 amnesic
(H.M.)
Star mirror
tracing
3 days practice Significant decrease in number of
errors without any awareness of task
Indicative of intact implicit
memory system despite
explicit deficits
Note: Studies are listed in chronological order with most recent first. (HC=healthy controls, PD=Pakinson’s Disease, S1=primary
sensory area, SMC=sensorimotor cortex, SMA=supplementary motor area, PMC=premotor cortex, SSA=supplementary sensory area,
CB=cerebellum, PD=Parkinson’s Disease, CBD=cerebellar degeneration, CVA=cerebrovascular accident, BG=basal ganglia,
SRT=serial reaction time, R=right, L=Left)
N 3
05
27
Neural Substrates Subserving Procedural Learning
Unlike the declarative learning and memory system no single or focal
region of the brain appears to subserve implicit motor-sequence learning.
Together evidence from human imaging studies, investigations of the abilities
of human subjects following neurologic damage, and animal lesion work,
reveals three major brain regions to be critical for implicit motor-sequence
learning: the cerebellum, basal ganglia, and sensorimotor cortical areas. Each
of these regions likely has a distinct role in implicit motor-sequence learning
(see Tables 1, 2, and 3 for an overview of studies using animal models,
imaging, and human patients).
Cerebellum and Implicit Motor-Sequence Learning
The cerebellum has long been postulated as a critical region for motor
learning. One function of the cerebellum during implicit motor-sequence
learning may be the chunking of individual pieces of information into larger
working units. The importance of the cerebellum for associating a stimulus
and response during conditional learning has been demonstrated by its role in
the eyeblink conditioned response (CR) in rabbits (McCormick, Steinmetz,
Lavond, Ivokovick, & Thompson, 1981; Clark, McCormick, Lavond, &
Thompson, 1984; Lavond, Steinmetz, Yokaitis, & Thompson, 1987).
Specifically, lesions to the ipsilateral interpositus nucleus completely abolished
the eyeblink CR without any effect on the unconditioned response (UR)
(Steinmets, 1992; Lavond 1984; Woodruff-Pak 1985). This effect persists
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
despite training following lesions (Lavond 1984) and is even seen in rabbits
following decerebration (Mauk & Thompson, 1987).
Association of individual pieces of information into larger units or
‘chunks’ facilitates faster processing of sequential responses (Hsiao & Reber,
1998; Perruchet & Amorim, 1992). This process allows advance preparation
of upcoming movements, and leads to faster responses. Chunking, therefore,
is critical for normal motor-sequence learning. If chunking is blocked, by either
a secondary task or long pauses between responses (analogous to trace
conditioning), implicit learning is prevented (Stadler, 1995). If the cerebellum
is functioning to chunk information during procedural implicit motor-sequence
learning damage to this structure should severely diminish implicit learning.
An alternative view of cerebellar function during motor learning posits
that it is involved in monitoring and updating movements on line using sensory
feedback (Gao, Parsons, Bower, Xiong, Li, & Fox, 1996; Bower, 1997;
Jueptner & Weiller, 1998). In this view the cerebellum is not necessary to
generate movement, however, it is critical for optimizing movement (during
learning) via its processing of sensory data during motor performances (Gao
et al., 1996). Evidence to support a sensory integration function for the
cerebellum stems from the finding that the lateral cerebellum is active (as
noted by PET scans) both during active and passive movements (Gao et al.,
1996; Jueptner & Weiller, 1998). Although this activation is differential (less
for sensory information i.e. passive movements), it does demonstrate a role for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
the cerebellum during sensory processing. Additionally, the cerebellum has
also been noted as critical for context dependent learning. During context
dependent learning, the appropriate motor program must be selected using
sensory information. Following focal lesions in the vermal and paravermal
cerebellar cortex the capability to select the correct motor plan based on
available sensory information is lost (Lewis & Tamargo, 2001). These findings
relate directly to implicit motor-sequence learning as demonstrated by Nixon
and Passingham (2001) who proposed that the function of the cerebellum may
be to prepare the next response in a series of “predictable” sensory events
during implicit motor-sequence learning. Nixon and Passingham (2001)
trained monkeys to make rapid movements to a target before selectively
lesioning the lateral nuclei of the cerebellum bilaterally. Following cerebellar
damage, the monkeys failed to demonstrate any reaction time savings when
the movement cue was predictable. Thus incorporation of sensory feedback,
and use of this information to predict and prepare the next response, may also
be a critical role of the cerebellum during implicit motor-sequence learning.
As might be expected, selective lesioning in a monkey model has
confirmed that damage to the cerebellum decreases implicit motor-sequence
learning. Using a monkey model (n=2) Lu, Hikosaka, and Miyachi (1998)
induced selective lesions by muscimol injections in different regions of the
cerebellar nuclei. Deficits in both learning of new, and memory for previously
learned sequences, was noted when the ipsilateral dorsal and central dentate
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
were inactivated. No effect on learning was apparent when the ventral or
lateral dentate, interpositus or fastigial nucleus were lesioned. These data
indicate that for this task (i.e. 2X5 sequential button press) ipsilateral dorsal
and central dentate function was critical for the learning of new sequences
(Table 1).
Neuro-imaging techniques (e.g. PET, fMRI, SPECT) have established
activity in the cerebellum corresponding to the early stage of implicit learning.
However, in examining this literature, some inconsistencies are apparent
across studies. These discrepancies are likely due to differences in
paradigms used, the stage of learning when brain scanning occurred, and the
field of view of the scanning device (Doyon, 1997a). For example, Grafton,
Mazziotta, and Presty (1992a) using PET examined the functional
neuroanatomy of visuomotor-skill learning during the pursuit rotor task.
Neurologically intact, control subjects were scanned while tracking a point on a
rotating disc with a stylus held in the right hand. In this study, no cerebellar
activation was recorded. However, only very early learning was examined and
more importantly a full view of the cerebellum was unavailable. Grafton,
Hazeltine, and Ivry (1995) repeated this work with the PET scanner set to
optimally view the cerebellum and recorded during early and later visuomotor
learning; subjects were scanned while learning the task and again the next
day following extensive practice. Learning-related increases in activation were
seen in the ipsilateral anterior cerebellum and parasagittal vermal areas. In
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
addition, left anterior cerebellar activity correlated with the speed of learning.
Following extended practice, regional cerebral blood flow (rCBF) was no
longer increased above baseline in the cerebellum (Table 2).
Seitz, et al. (1994) using neurologically intact healthy right-handed
subjects, has also noted activation of the cerebellum during early learning.
Subjects were asked to write ideograms, and were examined using PET
during early practice as well as after over-learning. In the early phase of
learning, the ipsilateral dentate nucleus significantly increased its activation.
This finding is supported by Jenkins, Brooks, Nixon, Frackowiak, and
Passingham (1994) who used PET techniques to examine the learning of a
sequence of key-presses by trial and error and eliminated the potentially
confounding effects of different speeds of performance during practice.
Compared to the performance of a pre-learned sequence, new sequence
learning elicited more extensive and intense cerebellar activity. Jenkins et al.
(1994) concluded that the cerebellum is particularly involved in the process “by
which learned tasks become automatic” (p. 3785). This conclusion gains
support from work by Nixon and Passingham (2000, experiment 3) who
demonstrated a failure to automate sequence learning following bilateral
excitotoxic lesions (in the cerebellar nuclei of 3 monkeys). Prior to lesioning
the cerebellum, these monkeys demonstrated sequence learning (4-element
SRT task); following bilateral cerebellar lesions they executed the same
sequence as well as they had prior to brain damage. However, new sequence
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
learning was disrupted following bilateral cerebellar lesioning. Although these
monkeys could learn a new sequence, examination of their behavioral data
indicated that their reaction times never reached asymptote; perhaps
demonstrating that they were not able to automate their learning over periods
of extended practice.
Changes in cerebellar activation patterns have been noted during early
and late practice of a maze task as well as with performance with both hands
(vanMier, Tempel, Permutter, Raichle, & Petersen, 1998). Ipsilateral anterior
cerebellar activity was associated with changing hand use (left anterior
cerebellar activity was noted with left-hand use, etc.); this pattern of activation
was related to movement execution and was seen early in practice. Further
support for the cerebellum’s role in early learning comes form the work of Toni,
et al. (1998) who noted that during trial and error learning of an 8-element
sequence (a unimanual task) that there was early bilateral activation of the
cerebellum. Subsequently over practice, contralateral activation decreased
and there was a small, but significant, increase in ipsilateral cerebellar activity.
It is difficult to assess the amount of isolated implicit motor-sequence
learning that occurred in each of the above studies as none of them controlled
the amount of explicit knowledge of their subjects. In an effort to address this
factor, Doyon, Owen, Petrides, Sziklas, and Evans (1996) scanned (PET)
healthy control subjects who were performing the SRT task when they were
unaware of the existence of a sequence, during random responses, and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
following the acquisition of explicit knowledge of the sequence. Subtraction of
activation during the random condition from that during the implicit learning
condition showed significant activation in the ipsilateral dentate nucleus.
However, after partial explicit knowledge of the sequence was gained dentate
activation returned to baseline levels.
The importance of cerebellar function for implicit motor-sequence
learning also has been shown in studies that demonstrate learning deficits in
individuals with cerebellar damage (Doyon 1997a; Doyon et al., 1997b;
Molinari, et al., 1997; Pascal-Leone et al., 1993; Table 3). Pascal-Leone et al.
(1993) examined the performance of individuals with bilateral cerebellar
degeneration as they practiced the SRT task. In the first condition, subjects
were unaware that they were practicing a 10-element sequence and no implicit
learning occurred. Further, explicitly teaching individuals with bilateral
cerebellar degeneration about the sequence prior to physical practice did not
facilitate implicit learning. Nearly all (9/10) cerebellar patients failed to
explicitly recognize the transition between the sequence and random
conditions during practice. Diminished implicit learning in individuals with
bilateral cerebellar dysfunction has been noted even with extended practice (6
training days / 40 trials of the sequence; Doyon et al., 1997b). Decreased
cognitive function, mood disturbances, or motor impairments could not explain
this effect. These data indicate that bilateral damage to the cerebellum
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
severely impairs implicit learning capability and that providing explicit
knowledge may not ameliorate this effect.
Despite numerous studies describing the activation of the cerebellum
during implicit learning as well as dysfunctional motor-sequence learning
following damage to this structure, the particular role of the cerebellum during
implicit motor-sequence learning remains unclear. Blood flow changes during
imaging studies indicate that the cerebellum is active early in learning - during
the encoding process (Doyon, 1997a). Doyon et al. (1998) concluded that the
cerebellum plays a role in the automatization process in later learning as well.
This was based on their findings that individuals with cerebellar damage did
not demonstrate implicit motor-sequence learning during dual task practice
conditions (SRT task and Brooks Matrices Test) initially or during re-test 10-18
months later. Data from Nixon and Passingham (2000) seem to strengthen
this conclusion. It appears, however, that once full automatization of the skill
has occurred, cerebellar activation is no longer critical (Flament, Ellerman,
Ugurbil, & Ebner, 1994; vanMier et al., 1994, 1995, 1998). Therefore, the
cerebellum is involved in motor-sequence learning but likely is not the storage
site for the memory trace of the skill. One putative role of the cerebellum
during implicit motor-sequence learning may be to form higher order
associations between responses and chunk items together (Curran, 1998). An
alternative (although not mutually exclusive) hypothesis postulates that the
cerebellum optimizes the motor plan by integrating sensory feedback and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
facilitating prediction of the next response during implicit motor-sequence
learning (Gao et al., 1996; Bower 1997; Nixon & Passingham, 2001).
Most investigations have used individuals with cerebellar degeneration
as a model for examining the role of this structure in learning. However,
several recent studies suggest lateralization of the function of the cerebellum
during motor-sequence learning. Toni et al., (1998) noted differential patterns
of activation on the contra- and ipsilateral sides of the cerebellum during trial
and error learning (see above for details). Gomez-Beldarrain, Garcia-Monco,
Rubio, and Pascal-Leone (1998) examined a group of individuals with
unilateral cerebellar stroke and demonstrated a dissociation during SRT task
learning using the hand contralateral versus ipsilateral to the lesion. Subjects
in this study did not demonstrate implicit learning when using the hand
ipsilateral to brain damage, however, when using the contralateral hand
normal learning was recorded. In direct contrast to Gomez-Beldarrain et al.
(1998), there has been one report of severe disruption of SRT task learning
following unilateral cerebellar damage using both the hand contralateral and
ipsilateral to brain damage (Molinari, et al., 1996). Also of interest in Molinari
et al.’s (1996) work, was the finding that explicit knowledge prior to SRT task
practice led to significant reductions in response time.
In aggregate, these data suggest several important distinctions
regarding the role of the cerebellum in implicit motor-sequence learning. First,
it appears that the cerebellum is critical for implicit motor-sequence learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
Second, the function of the cerebellum is likely to group individual elements of
the sequence into working units of movement and integrate sensory
information into the motor plan. These processes facilitate prediction and
faster processing of information. Third, cerebellar function appears to be
lateralized to the ipsilateral hemisphere during implicit learning of unimanual
movements. Last, it is unclear whether providing explicit information to
individuals with cerebellar damage during implicit motor-sequence learning will
be beneficial. These conclusions formed the bases for the following
hypotheses:
1) Following unilateral damage to the cerebellum implicit learning of motor-
sequences will be diminished relative to that for healthy controls. Normal
cerebellar function is essential for the learning of implicit motor-sequences
but this function is highly lateralized to the ipsilateral hemisphere during
unimanual tasks. Therefore, individuals with unilateral damage to the
cerebellum will be able to demonstrate some degree of implicit motor-
sequence learning using the arm contralateral to the lesion.
2) However, implicit motor-sequence learning following unilateral cerebellar
damage will not be affected by the nature of explicit knowledge.
Basal Ganglia and Implicit Motor-Sequence Learning
The basal ganglia likely function to coordinate the various elements of
movement sequences. It has been proposed that the role of the basal ganglia
in the production of movement sequences is to inhibit the elements of the
movements that are not being used while the current one is performed
(Bischoff, Arbib, & Winstein, 1997). Specifically, the putamen is involved in a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
cortical-subcortical motor loop that regulates voluntary movement (Alexandar
& Crutcher, 1990; Berridge & Whishaw, 1992). The importance of the striatal
projections within this loop has been demonstrated by animal lesion studies
that show grooming sequences in rats are completely abolished by striatal
aspiration. This occurs despite the animals continued ability to produce the
individual, isolated movement components (Berridge & Whishaw, 1992).
Further, different regions within the basal ganglia appear to support
various aspects of motor-sequence learning. Using two monkeys and focal
injections of muscimol Miyachi, Hikosaka, Miyashita, Karadi, and Rand (1997)
showed that chemical blockage of the anterior putamen and caudate disrupts
the learning of new sequences. Conversely, chemical blockage of the middle-
posterior putamen did not interfere with learning but prevented the execution
of previously learned sequences. These data are supported by other work in
which dopamine was depleted unilaterally in the monkey anterior striatum. In
this case the learning of new sequences with the arm contralateral to the
lesion was disrupted and an inability to shift strategies (to use the ipsilateral
arm) was noted (Matsumoto, Hanakawa, Maki, Graybiel, & Kimura, 1999).
Further, it has been demonstrated that neurons within the monkey globus
pallidus show ensemble activity in response to specific aspects of a sequence
(Mushiake & Strick, 1995). Globus pallidus activation has also been shown to
increase as sequences become more complex (Boecker et al., 1998). This
globus pallidus neuronal activity has been postulated as encoding detailed
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
information about the spatial-temporal characteristics of the sequence that is
being learning.
Similar to the cerebellum, the basal ganglia’s role in implicit motor-
sequence learning in humans has been demonstrated through its activation in
neuro-imaging studies (Doyon et al., 1996; Seitz & Roland, 1992; Jenkins et
al., 1994; Hazeltine, Grafton, & Ivry, 1997; Grafton et al., 1995, 1992; vanMier
et al., 1998; Toni et al. (1998) and by the impairments shown by individuals
with diseases or damage (Doyon et al., 1997b, 1998; Pascal-Leone et al.,
1993; Jennings, 1995). Using PET scans during the practice of different
sequencing tasks Jenkins et al. (1994) and Toni et al., (1998; both studying
key-press learning by trial and error) and Seitz and Roland (1992; studying
finger movements) have all demonstrated increased activation in the putamen
during both new learning and the retrieval of well-learned sequences. These
findings indicate that the basal ganglia are involved both in establishing and
regulating the final movement program (Table 2).
In contrast, Grafton et al. (1992) did not note any striatal activation
during the early learning stages of the pursuit rotor task. These conflicting
findings may be resolved by those of Doyon et al. (1996) who performed PET
scans during practice of the SRT task in both the early and late stages of
learning (subjects were scanned while performing a highly learned sequence
and while learning a new one). In this experiment, the ventral striatum was
distinctly active (as compared to a random condition) during performance of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
the pursuit rotor task under highly practiced conditions. If the role of the basal
ganglia is to coordinate the sequence via suppression of the functional
elements not presently in use, then it is not surprising that increased levels of
activation are noted in this region during later learning, a time when the
functional elements have already been established.
Specific regions of neuronal activation during implicit motor-sequence
learning of the SRT task have also been identified. Rauch, Whalen, Savage et
al. (1997) used fMRI in conjunction with behavioral data to demonstrate
significant right sided caudate and putamen activation (above baseline levels)
as healthy controls (n=10) practiced the SRT task bilaterally (first two fingers
of each hand). Further, the magnitude of the intensity of the signal from the
putamen correlated with the amount of RT decrease. Two intriguing
conclusions may be derived from this work. First, the right caudate and
putamen must be critical during the early stages (at least) of implicit learning of
the SRT task and second, the putamen is particularly important for decreasing
reaction times that are recorded during implicit SRT motor-sequence learning.
Once explicit knowledge is gained, the function of the basal ganglia is
presently unclear. Doyon et al. (1996) reported that once explicit knowledge
was gained, no striatal activity was noted during sequence practice (PET
scans). However, a separate study reported increased activation in the
putamen even after explicit knowledge had been gained by 7/12 subjects
(Grafton et al., 1995). This discrepancy may stem from the fact that only a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
little more than half of Grafton et al.'s (1995) subjects actually gained explicit
knowledge. Perhaps more robust levels of explicit knowledge are necessary
for striatal activity to decrease during sequence performance. It appears that
increased activity in the cerebellum and basal ganglia are not necessary after
explicit knowledge is acquired. One explanation for this is that the declarative
memory system is able to assume the function of these neural regions once
explicit knowledge is robust. The largest functional difference, therefore,
between the cerebellum and basal ganglia is their different roles in early
versus later learning. A prerequisite for the inhibiting and switching function of
the basal ganglia would be the early grouping of information into units by the
cerebellum.
Numerous other studies have demonstrated decreased implicit learning
in individuals with diseases affecting the basal ganglia such as Parkinson’s
(PD) and Huntington’s Disease (Pascal-Leone et al., 1993; Doyon et al.,
1997b, 1998; Jennings, 1995; Knopman & Nissan, 1991; Jackson etal.,
1995). Doyon et al. (1997) noted that individuals with PD failed to
demonstrate implicit learning in the SRT task under single (1997b) or dual
(1998) task conditions. Doyon et al. (1998) re-tested the same individuals with
PD 10-18 months later and noted that for those whose disease status had
progressed (from Hohn & Yahr Stage I to II), impairment of long-term retention
for the previously learned sequence was found. Thus, the striatum must play
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
a role in both the automatization, and long-term storage of, some elements of
the motor plan for implicit motor-sequence learning (Table 3).
The use of predictive information during performance of motor-
sequences allows rapid selection of and switching between motor responses.
Evidence from Jennings (1995) suggests that the basal ganglia are important
for preparing the next movement in a learned sequence. In this study,
Jennings used a cueing response paradigm to show that healthy control
subjects used advance information to prepare their first response as well as
subsequent movements. Individuals with PD, however, used advance
information only for their first response - no advance preparation of the
following movements was noted. Thus, basal ganglia dysfunction resulting
from PD interfered with the ability to move among elements in the motor
program - i.e. predicatively suppressing the last response and switching to the
next one. This finding is consistent with single cell neural activity records from
the monkey that demonstrated that the caudate delivers instructions for the
next movement in a learned sequence of responses (Kermadi, Jurquet, Arxi, &
Joseph, 1993). These data may be extended to propose that during sequence
practice and learning, the basal ganglia function to coordinate chunks of
information regarding upcoming movements by suppressing those not
currently in use (Bischoff, 1998).
To date only one study has examined the impact of unilateral basal
ganglia damage on implicit motor-sequence learning. Vakil, Kahan,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
Huberman, and Osimani (2000) studied SRT learning in 16 individuals with
unilateral focal stroke in the basal ganglia. These individuals practiced a 10-
element sequence over one day (40 trials) and returned the following day for a
retention test. Compared to age-matched healthy controls, SRT learning was
impaired in those with basal ganglia damage. Further, the authors
demonstrated decreased non-motor-sequence learning (subjects pressed the
spacebar on a keyboard in response to the third position stimulus being
illuminated) in the same individuals with basal ganglia stroke. One confound
in this study, however, was the use of the right hand for all SRT task practice.
Of the 16 individuals studied, 11 had a right-sided stroke and 5 a left-sided
stroke (all were right-handed). Thus, 5 of the 16 were asked to perform the
SRT task with their involved arm. As response time was the actual dependent
measure (not reaction time as stated in the article) the use by some subjects
of their motor-involved arm likely biased median group response times and
may have inflated the differences between the stroke and control groups.
Regardless of this problem in experimental design, Vakil et al.’s finding of
diminished SRT learning in this group suggests that bilateral basal ganglia
activation is necessary for normal implicit motor-sequence learning, and that
focal unilateral damage is sufficient to disrupt this process.
Unlike those with cerebellar damage there is some evidence that
individuals with PD may use explicit knowledge to compensate for diminished
function of the basal ganglia. Pascal-Leone et al. (1993) demonstrated
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
impaired implicit learning using the SRT task in those with PD who were
unaware of the sequence. These same individuals then participated in a
declarative instructional session in which they had to verbally practice the
sequence until they could reproduce it without errors. Individuals with PD
were eventually able to use the explicit knowledge they had gained to
decrease their RT, however, this process required significantly more practice
compared to that for control subjects. Further individuals with PD failed to
decrease their RTs to the same degree as control subjects.
Recent work, however, suggests that the ability to use explicit
information (in this case knowledge of results and varied practice in a rapid
goal directed arm movement) to optimize learning may be impaired in
individuals with PD (Onla-or, 2001). Individuals with PD practiced a unimanual
upper extremity arm movement over two days under either constant or
variable practice conditions. Typically individuals without any neurologic
damage have been shown to benefit from conditions of variable practice (for
examples see Shea & Morgan, 1979; Shea, Kohl, & Indermill, 1990).
However, Onla-or (2001) demonstrated that individuals with PD were unable
to benefit from conditions of varied practice, and that they learned only in the
context in which they had practiced. This study led to an important distinction
in the consideration of the role of the basal ganglia during motor learning. It
appeared that the motor learning of these individuals with PD was negatively
affected by a “set switching deficit” in which they were unable change motor
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
strategies (based in this case on varied practice conditions). Thus the motor
learning ability of individuals with PD was strongly context-dependent. These
data pose an important question relevant to this study: will individuals with
unilateral stroke in the basal ganglia be able to take advantage of explicit
knowledge to improve motor-sequence learning? If the function of the basal
ganglia is to shift between motor strategies incorporating explicit knowledge
into the motor plan, then explicit knowledge may not prove to be beneficial.
Indeed, the implicit motor-sequence learning of individuals with basal ganglia
stroke may be deleteriously affected by attempts to shift motor strategies
based on explicit knowledge. One goal of this dissertation is to determine if
the presence of explicit knowledge is beneficial, neutral, or detrimental to
implicit motor-sequence learning following focal unilateral stroke in the basal
ganglia.
In sum, previous findings point to an integral role of the basal ganglia
for coordinating a series of responses during implicit motor-sequence learning.
It appear that dysfunction resulting from damage in the basal ganglia,
however, may be lessened by explicit knowledge (at least for the SRT task) as
the declarative memory system functions to coordinate and anticipate the next
response in a sequence. It is unclear, however, how much explicit information
is necessary for the declarative memory system to assume this role and
compensate for basal ganglia damage. This study will also determine if
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
explicit knowledge prior to practice can facilitate the implicit learning of a
relatively complex motor-sequence-tracking task.
We propose that following damage to the basal ganglia, switching
among elements of the sequence will be disrupted, limiting implicit motor-
sequence learning. Presently, it is unclear if the provision of full explicit
knowledge regarding the sequence being practiced will reduce this dysfunction
following unilateral focal basal ganglia stroke. Therefore we hypothesize that:
3) Following unilateral damage to the basal ganglia implicit learning of motor-
sequences will be diminished relative to that for healthy controls.
4) During implicit motor-sequence learning, the basal ganglia function to
coordinate serial responses. Therefore, unilateral damage to the basal
ganglia will result in absent implicit learning when subjects are without and
when they have explicit knowledge of the sequence.
Sensorimotor Cortical Areas and Implicit Motor-Sequence Learning
The sensorimotor cortical areas that are most important for implicit
learning include the primary motor cortex, supplementary motor area, and
premotor area. It is likely that these regions are not the only cortical areas
important for implicit learning. For example, recent evidence suggests that the
pre-supplementary motor area contains neurons that are specific to certain
sequences of actions (Nakamura, Sakai, & Hikosaka, 1998). Within a
behavioral context, however, we are limited to studying the roles of the neural
regions that most directly contribute to implicit motor-sequence learning and
thus will focus on these three critical areas.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
Primary Motor Area (Ml)
Putative roles for the primary motor area (Ml) during implicit motor-
sequence learning include movement initiation and fine motor coordination
(Berridge & Whishaw, 1991; Barone & Joseph, 1989; Table 1). During
unimanual movements, contralateral Ml has been shown to be active in
numerous imaging studies (Honda et al., 1998; Hazeltine et al., 1997; Grafton
etal., 1992; Grafton et al., 1995; vanMieretal., 1998; Seitz & Roland, 1992;
Kawashima, Roland, & O’Sullivan, 1994). Contralateral Ml activation has
been noted as being more robust for practiced sequences (even after
relatively small amounts of practice in one session) than for random ones
(Kawashima et al., 1994; Shibasaki, et al., 1993). Changes in Ml over
relatively short amounts of practice are likely those mediated by “fast” learning,
in which early within session improvements are noted with motor-sequence
practice (Kami, et al., 1998). These changes likely are reflective of the
generation of task-specific motor processing within Ml such as the direction of
the next movement (Kakei, Hoffman, & Strick, 1999). Following this early
improvement, ongoing evolution of skill or “slow” learning takes place
consisting of delayed gains (consolidation) in performance that follow practice.
Ml may also be involved in mediating the specific changes in force for
the groups of muscles involved in a new task. One example of this possible
role for Ml comes from Muellbacher, Ziemann, Boroojerdi, Cohen, and Hallet
(2001) who studied the changes in Ml excitability (using rTMS) while healthy
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
individuals learned a pinch task. Brief practice led to both behavioral
improvement as well as an increase in TMS-stimulated motor evoked
potentials (MEP) in the muscles directly used for the pinch task; resting motor
threshold was unchanged. This increase in MEP was seen only for direct
stimulation of Ml by TMS and not after stimulation to the brainstem or cervical
spinal segments. These data may be interpreted as reflecting the early
phases of Ml reorganization during motor learning.
Ipsilateral Ml also appears to have a critical role in the learning of
unimanual motor-sequences (Chen, Cohen, & Hallett, 1997a; Chen, Hallett, &
Cohen, 1997b; Grafton et al., 1992; Tinazzi & Zanette, 1998; Shibasaki et al.,
1993; Winstein, Merians, & Sullivan, 1999). Using PET imaging during the
execution of complex finger movements Shibasaki et al., (1993) noted
increases in ipsilateral Ml (in addition to activation in contralateral Ml). This
activation of ipsilateral Ml was not observed during the execution of simple
repetitive finger movements. Additionally, Chen et al. (1997a) demonstrated
the importance of ipsilateral Ml during the learning of hand sequences by
stimulating this region during sequence practice using repetitive transcranial
magnetic stimulation (rTMS). The authors found that stimulation ipsilateral to
the hand being moved resulted in timing errors during the execution of both
simple and more complicated sequences. Further, different roles for left
versus right Ml were noted. Stimulation of left Ml was more disruptive to
complex sequences than it was to simple ones. In addition, left-sided
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
stimulation resulted in disruptions that continued into the post-stimulation
period while right-sided rTMS impaired sequence learning only during
stimulation (Table 2). In sum, these data indicate that the roles of ipsilateral
and contralateral Ml differ by laterality (right versus left) and task complexity.
Ipsilateral Ml (in combination with the already active contralateral Ml area)
appears to be vital during the execution of more complex movement patterns.
Premotor Cortex
Premotor cortex (PMC) is likely important for transitioning between
movements and spatial working memory (Mushiake, Inase, & Tanji, 1990,
1991). Regardless of the hand being used, right PMC has been noted to be
active during motor-sequence learning using the SRT task paradigm
(Hazeltine et al., 1997; Grafton et al., 1995; Honda et al., 1998), the pursuit
rotor task (Grafton et al., 1992), and during maze tracing (vanMier et al.,
1998). Grafton et al. (1995) postulated that the PMC was active during
visuomotor-sequence learning as a result of the spatial working memory
demands of the SRT task. This is consistent with the finding of increased PMC
activation early in practice as stimulus-response maps were being formed
(Grafton et al., 1995; Jenkins et al., 1994; Kawashima et al., 1994). However,
when awareness of the sequence was prevented, using a dual task version of
the SRT task, no activation in PMC was noted (Hazeltine et al., 1997; Grafton
et al., 1995; Honda et al., 1998; Table 2). PMC therefore, makes a larger
contribution to sequence learning when movements are being directed by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
external cues (e.g. conscious hypothesis testing as in trial and error sequence
learning; Curran, 1995).
Further evidence for PMC’s regulation of movements when declarative
knowledge is present, comes from the findings of Seitz et al. (1994). In this
work 8 healthy individuals were scanned using PET as they learned a
unimanual task. In a period of overlearning of the movement (when subjects
had gained explicit awareness of the task), it was noted that activity had
significantly increased in bilateral PMC; however, PMC had not been active in
the early learning process. These data suggest that the amount of explicit
awareness possessed by subjects’ plays a role in the recruitment of specific
neural systems. A selective role for PMC during sequence learning, when
declarative information is available, is further supported by its neuronal
connections to the prefrontal cortical regions (e.g. DLPFC). Prefrontal areas
(DLPFC) have also been shown, as being active when explicit knowledge is
present. The interactions between these regions (both functional and
neuroanatomical) will be discussed in detail in a later section of this chapter.
Supplementary Motor Area
Contrary to the finding of diminished PMC activity under conditions of
implicit motor-sequence learning (e.g. subjects have not gained explicit
awareness of the sequence or task), supplementary motor area (SMA)
appears to increase its activity when explicit knowledge is unavailable. Single
cell recordings from monkey SMA have demonstrated sequence specific
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
neurons and support SMA’s putative role in regulating internalized sequences.
These neurons were active only during internally guided (i.e. implicit)
sequences and sub-sets were active only for particular sequences (Mushiake
et al., 1990, 1991; Table 1). Grafton et al. (1995) and Hazeltine et al. (1992)
have demonstrated that SMA increased its activation when SRT task learning
was implicit under dual task conditions (in the dual task condition, subjects are
cued externally for each response in the primary sequencing task, however, a
secondary explicit task engages the declarative system and prevents
formation of awareness of the repetitive aspect of the practiced sequence;
Table 2). Further, it has been reported that SRT task learning is severely
impaired following unilateral left-sided focal stroke in the SMA (Ackermann,
Daum, Schugens, & Grodd, 1996; Table 3).
There has been recent evidence that the SMA may be functionally
segregated into rostral and caudal components. Boecker, et al. (1998) used
PET scanning to investigate the performance of previously learned unimanual
sequences in 7 healthy individuals. Increasing sequence complexity was
correlated with contralateral rostral SMA action. Caudal SMA was noted to be
more active during sequence execution (compared to rest), however, it did not
selectively increase its action in response to a more complex motor task. It
may be that the rostral SMA plays a role in modulation and control of more
complex motor-sequences, while caudal SMA is active for motor execution.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
SMA appears to be important for the selection of responses in a
sequence when choices are based on internal or pre-learned information.
PMC and SMA are differentially important for response selection based on
available information. Recent work by Toni et al. (1998) examined sequence
learning by trial and error (a task in which explicit knowledge of the task is
provided at the outset) and demonstrated the functional interplay between
PMC and SMA. In this task PMC was highly active early, when explicit
strategies were being used to learn the sequence. Then, as the sequence
was discovered and became more automatic, PMC decreased and SMA
increased its activity. Therefore, PMC and SMA’s unique functional
contributions to response selection may depend on the specific information
available - PMC uses external information and SMA coordinates internally
guided responses. From these data we propose that:
5) Following unilateral damage to the sensorimotor cortical areas implicit
learning of motor-sequences will be diminished relative to that for healthy
controls.
6) Unilateral damage to the sensorimotor cortical areas will cause a decrease
in implicit motor-sequence learning when no explicit knowledge is provided.
Implicit motor-sequence learning ability will increase as greater amounts of
explicit knowledge are provided.
Summary
A thorough understanding of the interactions among neural substrates
during implicit motor-sequence learning has not yet been gained, however,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
some preliminary relationships may be proposed based on what is presently
understood about neural connections and the work discussed above (Figure
2).
SMA
Pre-SMA (anterior)
Posterior SMA
Anterior = Ant.
Putamen & Caudat
Posterior = Middle
Post. Putamen
Globus Pallidus
internal
Cerebellum
Dentate Nucleus
Primary Motor
Thalamus
Basal Ganglia
Motor
Command ^
Figure 2. Schematic representation of the neural network underpinning
implicit learning.
The cerebellum functions to chunk individual pieces of information into larger
functional units, and integrate sensory information into the developing motor
plan, during early practice. Information from the cerebellum is relayed to
either PMC (when sequence performance is guided by explicit or external
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
declarative knowledge) or SMA (when sequence performance is guided by
implicit or internal knowledge) via the thalamus. SMA contains information on
the overall sequence and what movement is currently being performed.
This information is sent to the anterior basal ganglia (anterior putamen and
caudate), which function to coordinate the ongoing movement with upcoming
ones. SMA also sends information about the sequence to Ml which controls
the execution (and aids in the control of the timing) of the movements. If any
of these processes or connections are disrupted, implicit motor-sequence
learning will be diminished.
When explicit knowledge is available to the learner the prefrontal
association areas, in particular DLPFC, are invoked into the network
supporting implicit motor-sequence learning. DLPFC has direct connections to
PMC and basal ganglia as well as an indirect connection with cerebellum via
thalamus (Middleton & Strick, 1994; Selemon & Goldman-Rakic, 1985;
Alexander, Delong, & Strick, 1986). These connections (via thalamus) likely
support the integration of explicit information into the implicit motor plan (their
putative roles will be discussed in detail in the next section). An overview of
the predicted impact of explicit knowledge and lesion location on implicit
motor-sequence learning has been presented in table 4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
Table 4. Predicted impact of explicit knowledge condition and lesion location on implicit
motor-sequence learning.
Group
________________ HC_______ CB BG SMC
EK +
No-EK + +
Note: EK = explicit knowledge; No-EK = no-explicit knowledge; T+ = demonstrate implicit
learning, the size of this character is designed to reflect the relative magnitude of change in
performance associated with learning; + = Increased implicit learning beyond that see
without EK; - = No implicit learning.
Other neural substrates likely are also important in implicit motor-
sequence learning (e.g. sensory cortex for afferent feedback, frontal lobes for
attention; thalamus for integration of information; specific modality processing
regions for visual versus auditory responses, anterior cingulate gyrus for
visuomotor attention, etc). The regions proposed for study, however, appear
to contribute distinct and unique functions to implicit motor-sequence learning.
Therefore, from the data presented it appears very likely that if the cerebellum,
basal ganglia, or sensorimotor cortical areas are damaged or their function is
disrupted, implicit learning will be compromised and not occur normally.
t
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Interactions between Implicit and Explicit Learning and
55
Memory Systems
Functional Interactions
The interaction between explicit and implicit memory systems during
motor-sequence learning has yet to be well understood. Recently it was
demonstrated that extended, focused implicit practice could be used to
promote the explicit memory function in an individual with profound declarative
memory loss (Glisky & Schacter, 1987, 1988, 1989). In this work a severely
amnesic patient was taught the knowledge and skills necessary for success at
a complex computer data-entry job. Training was accomplished using the
“method of vanishing cues” (Glisky & Schacter, 1987, p. 894) which used
priming (a function of the preserved implicit memory system) to complete
fragments of words. With this method, individuals are initially given as many
letters as they need in order to identify a target response. Subsequently the
length of the word fragment is reduced across trials until the appropriate
response can be made without any cueing. This method of training is also
structured around “domain specific knowledge” (Glisky & Schacter, 1987,
p.894) in which only very specific skills are taught. Using these strategies the
patient H.D. was able to learn 250 discrete pieces of information regarding
data entry rules. However, H.D. was unable to generalize her new learning to
other tasks; large decrements in performance were observed if any small
changes occurred in her routine (Glisky & Schacter, 1989). It should also be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
noted that the learning demonstrated by H.D. required extended amounts of
practice over long periods of time (1-2 hours / day, 2-3 times / week, over
several months). Glisky and Schacter (1993) confirmed these findings,
showing that six individuals with amnesia were able to learn new semantic
information although not at a normal rate. Clearly, this work demonstrates that
declarative learning may take place with extended practice using the
procedural learning system. This is likely not a direct or robust effect as
indicated by the extensive practice required for explicit knowledge of the task
to be gained.
Less studied is the impact of prior explicit or declarative information on
motor-sequence learning in cases of an impaired procedural memory system.
Thus far, it appears that the effect of explicit knowledge on implicit motor-
sequence performance is dependent on the type, timing, and meaningfulness
of the information that is provided. For example, Reber (1976) showed that
explicitly directing individuals without neurologic damage to search for the
rules governing an artificial grammar task degraded their performance as
compared to giving neutral instructions prior to practice. Individuals who were
instructed to search for grammar structure, learned less and tended to invent
rules. This suggests that for this experiment, the information gained during
implicit learning had an advantage over that provided by explicit knowledge.
Similarly, in an implicit motor-sequence task, subjects who were explicitly
given probability instructions performed poorer than uninstructed subjects in a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
complex, fine-motor catching task (Green & Flowers, 1991). In this task,
explicit knowledge was provided regarding the probable trajectory of the target
(75% of trials). Instructed subjects attempted to apply an overt strategy based
on their explicit knowledge that resulted in the production of a less efficient
movement pattern. Thus, in Green and Flowers’ (1991) study, task
instructions appeared to interfere with the development of a correct and
accurate strategy for success. These data and others (Wulf & Weigelt, 1997)
suggest that the cognitive demand of instructions may disrupt the formation of
the implicit motor plan.
In each of these studies it appears that explicit knowledge had a
particular interference effect on performance (Bliss, 1892; Boder, 1935). It
may be that for each of these studies (Reber, 1976; Green & Flowers; 1991;
Wulf & Weigelt, 1997), the rules necessary for successful task completion
were not ones that could be expressed explicitly. Subjects were being told to
search for and use rules that they were not likely to find (Reber, 1976) and not
likely to find useful (Green & Flowers, 1991). Specifically, each of these tasks
had a relatively complex motor solution. It may be that for more complicated
motor tasks; explicit information is less helpful in the development of the motor
plan than is discovering a motor solution using the implicit system alone.
Further evidence for an interaction between implicit motor-sequence
learning and explicit knowledge stems from the reported benefit of explicit
knowledge for the SRT task for both individuals without neurologic damage
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
(Curran & Keele, 1993) and individuals who have had a stroke (Boyd &
Winstein, 2001; Molinari et al., 1997). For each of these studies, access to
explicit knowledge of the sequence prior to practice benefited performance
(demonstrated by larger decreases in reaction time over practice). In
particular, Boyd and Winstein (2001) found that following stroke (located in
the sensorimotor cortical areas and their output tracts) even partial explicit
knowledge of the sequence significantly reduced reaction time on the SRT
task as compared to individuals post-stroke without explicit knowledge. In the
Boyd and Winstein (2001) study, acquisition (and amount) of explicit
knowledge that subjects gained prior to physical task practice was evaluated
via a pre-test. In a separate study Reber and Squire (1998) did not find that
explicit knowledge prior to practice of the SRT task aided performance in
neurologically healthy individuals. One explanation for this discrepancy in
findings is Boyd and Winstein’s (2001) use of individuals who were post-stoke
versus Reber and Squire’s (1998) study of individuals without any neurologic
damage. Another explanation for the differences between these two studies
may stem from the failure of Reber and Squire to ensure that explicit
knowledge had occurred prior to sequence practice (i.e. no pre-test).
Additionally, Reber and Squire provided explicit knowledge by having subjects
watch the sequence 10 times. It is likely that this form of explicit knowledge
lacked salience and interest for subjects, preventing them from making use of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
it. Thus, the manner in which explicit knowledge is provided may be an
important factor for its use.
Several authors have postulated that under normal circumstances
implicit and explicit learning occur in parallel, but that explicit awareness of the
correct strategy for success does not develop until a certain degree of motor
success has been achieved (Brooks, Hilparth, Brooks, Ross, & Freund, 1995;
Gentile, 1998; Willingham & Goedert-Eschmann, 1999). This is supported by
the finding that explicit knowledge of the existence of the sequence in the SRT
task often does not develop until after implicit learning (as indicated by
significant reductions in RT) has occurred (Nissan & Bullemer, 1987; Grafton
et al., 1995). Further, in other more complex implicit learning paradigms
explicit awareness is commonly not gained even after extended amounts of
practice (Pew, 1974; Magill 1998; Masters, 1992; Wulf & Schmidt, 1997;
Lewicki, 1987, 1998). For example, in Wulf and Schmidt’s (1997) study,
subjects practiced the same pattern 60 times a day for 4 days. When asked if
they thought some part of the pattern had been repeated, only 4 of 42
answered yes indicating that they had some explicit awareness of the
existence of a pattern in their responses. Further, when told that 1/3 of the
pattern had repeated itself and asked to guess which part, only 14 of 42 were
able to correctly do so. These data indicate that the processing involved in
implicit learning likely needs to be advanced to a fairly automatic level before
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
sufficient cognitive resources can be freed to permit explicit awareness of the
implicit task.
It appears that at least three prerequisites are necessary for explicit
knowledge to promote implicit learning: 1) the information must be salient, 2)
subjects must demonstrate explicit knowledge of the information, mere
exposure is insufficient, and 3) the motor task must be simple enough for the
learner to be able to incorporate explicit knowledge into performance. Little
research presently exists to determine what impact prior explicit knowledge
has on the implicit motor-sequence learning of more complex motor responses
and to date no investigations have systematically examined the impact of
explicit knowledge on implicit learning following damage in the cerebellum or
basal ganglia.
Neuroanatomic Interactions
Evaluation of the anatomic organization of the neural regions that
support the declarative and procedural memory systems during implicit motor-
sequence learning indicates that there may be several avenues for their
interaction. Specifically, there are two prominent pathways that likely are
involved in the mediation of information between the declarative and
procedural systems - the cortico-basal ganglia-thalamo-cortical loop (Parent &
Hazrati, 1995) and the cerebello-thalamo-cortical pathway (Middleton & Strick,
1994). Further, other data suggest that there may be some degree of
interaction between these two pathways.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
61
The basal ganglia project to four regions within the frontal lobe
functionally creating five basal ganglia-thalamocortical circuits. The prefrontal
(DLPFC and lateral orbitofrontal cortex (LOFC)), motor (precentral motor
fields), occulomotor (frontal and supplementary eye fields), and limbic (anterior
cingulate and medial orbitofrontal cortex) areas of the cortex are all targets of
the basal ganglia (Alexander & Crutcher, 1990; Parent & Hazrati, 1995). Of
these general circuits, there are two prefrontal projections from basal ganglia,
one to DLPFC and one to LOFC. For this discussion the direct connection
between DLPFC and basal ganglia is the most intriguing. The projection from
DLPFC terminates within the caudate (Selemon & Goldman-Rakic, 1985).
Interestingly, parallel projections arise in the posterior parietal lobe and PMC,
and also terminate in the caudate. Despite these parallel projections their
terminal arborizations do not overlap but sit adjacent to one another within the
basal ganglia (Selemon & Goldman-Rakic, 1985). Reciprocal connections
back to the DLPFC partially close this loop. These connections pass through
the thalamus (ventrolateral and ventromedial anterior thalamic nucleus) en
route back to the prefrontal areas. Miyachi et al. (1997) illustrated the
functional relationship between these neural regions by blocking the anterior
basal ganglia and abolishing (in monkeys) the ability to learn sequences by
trial and error (a task that requires explicit information processing and
therefore must invoke the pre-frontal cortex; Figure 3).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
Medial Temporal Lobe
Hippocampus
Peri-hippocampal
structures i
Motor Cortex
. (Ml)
PMC
DLPFC
Basal Ganglia
Anterior = Ant. <
Putamen & Caudate
Thalamus
VLo
VLc
J W MD Posterior = Middle
Post. Putamen
Pons
Globus Pallidus
internal
Cerebellum
Dentate Nucleus
Motor
Command
Figure 3. Putative interrelationships between implicit and explicit neural
networks.
Middleton and Strick (1994) have illustrated another avenue for interaction
between the procedural and declarative neural networks. Retrograde
transneuronal transport (herpes simplex virus) was used to identify the
neurons that project to thalamus and subsequently to prefrontal cortical areas
(i.e. DLPFC). This labeling technique identified neurons from two distinct
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
regions that projected via thalamus to DLPFC - dentate nucleus of cerebellum
and internal portion of the globus pallidus from the basal ganglia (Figure 4).
Basal ganglia and cerebellum both project to ventrolateral nucleus in
thalamus, however, they have separate target zones within these projections
(Alexander & Crutcher, 1990; Houk, 1997; Middleton & Strick, 2001). None of
these projections overlapped with those to the motor areas of the cortex, thus
they constitute a separate, parallel pathway for information transfer and
processing between the procedural and declarative memory systems. Support
for this argument comes from recent work demonstrating that cerebellar output
from the dentate projects to the prefrontal cortex from the thalamus. Further
within the dentate nucleus, separate output channels exist that have the
potential to influence motor (to Ml) and cognitive (to prefrontal cortex)
separately (Middleton & Strick, 2001). In addition, the pathways from the
cerebellum (dentate) and basal ganglia (globus pallidus internal) to Ml
constitute a very small percentage of total output (15 and 30%, respectively).
Thus, the majority of output from these two regions is directed elsewhere in
the cortex (Hoover & Strick, 1999). These data suggest that the output of the
cerebellum and basal ganglia are connected with the regions associated with
declarative knowledge.
Based on what is currently understood about the function of the DLPFC
and basal ganglia it is tempting to conclude that the prefrontal cortico-basal
ganglia-thalamo-cortical circuit represents the primary neuroanatomic pathway
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
for information transfer between the declarative and procedural memory
systems during motor-sequence learning. The cerebello-thalamic-cortical
pathway may be able to influence and share information with cortico-basal
ganglia connections as they both project to DLPFC; a site where information
could be entered into the basal ganglia loop from the cerebellar loop for
integration into the motor plan (Figure 4).
Prefrontal
Cortical
Column
Basal Ganglia
Thalamus
BG
Striatum
G lobus
P allidus
Prefrontal
Cortical
Column
Cerebellum
Thalamus
CB
Pons
D entate
Adapted from Houk, 1997
Figure 4. The interrelationships between the basal ganglia and cerebellum
during implicit motor-skill learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
It appears then that the basal ganglia are critical for the integration of
declarative information (e.g. explicit knowledge about the task, instructions)
into the implicit motor plan. Therefore, damage to the basal ganglia should
profoundly disrupt implicit motor-sequence learning as well as alter the
integration of explicit knowledge into the motor plan.
Summary
From this review it is apparent that there is both a functional and
neuroanatomic dissociation between the declarative and procedural memory
systems. Discrete regions in the medial temporal lobe mediate declarative
memory and learning while procedural memories are subserved by a
distributed neural network. The focus of this dissertation will be on the three
most prominent members of the network for procedural learning, the
cerebellum, basal ganglia, and sensorimotor cortical areas. A second focus of
this work is the interaction between these two memory systems. The work of
Glisky and Schacter (1987, 1988, 1989,1993) provides behavioral evidence
for information transfer from the implicit to the explicit system. Neural
connectivity seems to support the potential for transfer of knowledge from the
explicit to the implicit memory systems; however, data regarding this
interaction are less clear and may be somewhat task dependent. A goal of
this dissertation is to provide evidence that will elucidate the potential transfer
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
of information from the declarative to the procedural memory systems as well
as the distinct roles that the basal ganglia, cerebellum and sensorimotor
cortical areas play in this process.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
CHAPTER 3
Neuropsychology of Implicit Motor-Sequence Learning
Following Stroke
Introduction
This chapter will discuss motor learning following stroke with respect to
the neuropsychology of procedural learning. Particular emphasis will be
placed on previous work of motor learning following stroke and the capability
of individuals with post-stroke related brain damage to demonstrate implicit
motor-sequence learning. In addition, the principles of motor learning relevant
to this dissertation (i.e. retention and transfer test designs) will be reviewed.
At the conclusion of this chapter the hypotheses that were derived from this
review are presented.
Motor Learning after Stroke-Related Brain Damage
The ability to learn motor skills following stroke-related brain damage is
critical for individuals to return to independent function. Further, most
individuals with stroke-related brain damage are older and the normal effects
of aging may impact their learning capability. It is well documented that with
age voluntary movement slows (for examples please see Pratt, Chasteen, &
Abrams, 1994; Pohl & Winstein, 1998). In the past, this slowing was attributed
largely to peripheral changes, but more recently it has been demonstrated that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
there are also alterations in the manner in which movements are planned and
implemented (Haaland, Harrington, & Grice, 1993).
Even with the challenges associated with aging that were briefly
presented above, following stroke individuals do retain some ability to learn
new motor skills (Winstein Merians, & Sullivan, 1999; Hanlon, 1996; Platz,
Denzler, Kaden, & Mauritz, 1994). For example, Winstein et al. (1999)
demonstrated that although performance was not as accurate, similar
improvements were seen over practice and retention of a discrete visuomotor
movement task for individuals with and without stroke-related brain damage.
Similarly, Platz, Denzler, Kaden, and Mauritz (1994) studied 20 individuals
who had near complete functional recovery following stroke (minimal residual
affected arm paresis as indexed by strength grades). These individuals
practiced a three dimensional motor learning task and their ability to learn was
evaluated by a retention test on the day following the practice session.
Following stroke, it was noted that the ability to solve spatial motor problems
and improve performance with practice was comparable to that of healthy
control subjects. Despite these improvements, several deficits in the
movements made by the individuals with stroke were noted. First, their
movements were less consistent. In addition, following stroke it was noted
that movements were slower and more corrections were necessary. The
ability of these individuals with stroke to improve, however, demonstrates an
ability to learn new motor skills with practice.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
69
Following unilateral stroke individuals can learn following the same
learning schedule as do those without neurologic damage. Hanlon (1996)
investigated the contextual interference effect in 24 individuals with unilateral
stroke. In this work, half of those with stroke related brain damage practiced a
arm movement sequence using the paretic limb under random practice
conditions and half under blocked practice conditions. Similar to previous
work with non-neurologically affected individuals, Hanlon’s (1996) findings
demonstrate that following unilateral stroke, random or varied practice was
more effective for learning (as evidenced by retention test performance).
Some caution must be used in interpreting these data; from this study it is not
clear what region of the brain was affected by stroke. It has recently been
demonstrated that random practice does not benefit individuals with PD during
learning a rapid arm movement (Onla-or, 2001). It is tempting to combine the
results from these two studies and infer that individuals in Hanlon’s (1996)
work did not have lesions that extended into the basal ganglia. Ultimately
however, Hanlon’s data do add to a growing body of literature that suggest
that following stroke related brain damage, individuals can demonstrate motor
skill learning (see Winstein et al, 1999; Sullivan, 1998).
In most motor learning investigations of individuals with stroke, subjects
had conscious, explicit awareness of, and feedback (knowledge of results) for,
the task goals and requirements during practice. Thus, they were able to use
their declarative memory system to augment motor performance. In contrast,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
when explicit knowledge of the task constraints is not provided, we have
shown that implicit motor-sequence learning is diminished (Boyd & Winstein,
1998a, 1998b, 2000, 2001; Boyd, Baker & Winstein, 1999).
In a series of experiments we (Boyd & Winstein, 2001) assessed
implicit motor-sequence learning (using the SRT task) in individuals with
unilateral brain damage following stroke under three separate practice
conditions: when subjects were 1) unaware of the sequence and practice for
one day, 2) unaware of the sequence but allowed extended practice (three
days), and 3) provided with explicit knowledge of the sequence’s existence
and composition prior to implicit practice. Three hypotheses were tested: 1)
individuals with unilateral stroke-related sensori-motor area brain damage
using the arm ipsilateral to stroke would demonstrate implicit motor-sequence
learning (as demonstrated by decreases in reaction time (RT)), 2) extended
practice would facilitate the magnitude of implicit learning, and 3) explicit
knowledge provided prior to physical sequence practice would potentiate the
degree of implicit motor-sequence learning. Twelve subjects participated; and
all had CT or MRI verified unilateral sensori-motor area brain damage
following stroke (except one who had a unilateral brain stem lesion).
For each session the same practice order was followed. To reduce the
impact of task familiarity (non-specific learning), subjects initially practiced a 4-
choice random sequence. After one block of practice in the random condition
(6 trials of a 9-element non-repeating sequence) a repeating 9-element
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
sequence was practiced 18 times (3 blocks). Next, a second block of 4-choice
random responses was practiced; followed by one block of the repeating
sequence. Each day, the repeating sequence was practiced a total of 24
trials; transitions between random and repeating sequence practice were
transparent. All subjects used the arm ipsilateral to the side of focal brain
damage.
After practice a questionnaire was used to assess subjective explicit
awareness by asking subjects if they had ever noticed a pattern. Recognition
was tested by showing each subject schematic representations of the
repeating sequence and two foil sequences. For each presentation, subjects
were asked to judge if the schematic was a sequence that they had seen
before. Recall was determined by asking subjects to complete a 4-element
fragment of the sequence by filling in one blank color with one of four-color
choices. If they were unsure subjects were encouraged to make a best guess
(forced choice).
Median RT was calculated for each block of practice. For each of the
four blocks of repeating sequence practice median RT was subtracted from
the median RT of the second 4-choice random block to calculate a change
score. A paired T-test was used to determine if the difference between
median RT for the last random and sequence blocks was reliable (significance
p<0.05).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
To assess the ability of participants who are post-stroke to learn under
SRT conditions, four individuals with a unilateral stroke practiced four blocks of
the repeating sequence in one day a sequence without awareness of
sequence existence (ST-Unaware). Immediately following the last practice
block explicit awareness was tested.
i—
0 1
c
< L >
g -10
c a
6 -20
.1 -30
1 -40
« -50
0
^ -60
-70
3 4 1 2
Blocks
Figure 1. RT change across sequence block practice for ST-Unaware group.
At the conclusion of practice there was no difference between RTs for the
random and sequence practice conditions. Error bars are the standard error of
the mean (SEM). The zero line is the median RT for the second random
sequence block.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
RT did not change across practice for the ST-Unaware group (p=0.96;
mean median RT change 5.4 msec; Figure 11 ). One ST-Unaware subject
gained subjective explicit knowledge of the sequence following practice, and
none of the ST-Unaware group demonstrated recognition or recall. From
these data we concluded that contrary to our hypothesis, following unilateral
stroke, when subjects are unaware of the specific task being practiced, implicit
motor-sequence learning is impaired.
We hypothesized that with more practice, those with stroke would
demonstrate implicit motor-sequence learning. Another hypothesis was that
the provision of explicit knowledge prior to sequence practice would promote
the rate of SRT task learning in individuals with unilateral stroke. These
hypotheses formed the basis for our next two experiments.
Four different individuals with unilateral stroke (ST-Extended Practice)
practiced the SRT task over three consecutive days. The procedure was
identical to that described above with a few exceptions. The same 9-element
sequence was practiced 24 times a day over three consecutive days (72-
practice trials, 12 blocks total). The ST-Extended Practice group was not
explicitly informed of the existence of the sequence. As in experiment 1, at the
conclusion of practice on day three, explicit knowledge of the practiced
sequence was tested.
1 Healthy controls tested under the same conditions show a 29.8 ms decrease in RT (Boyd & Winstein,
1998b)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
By the end of practice on day three there was no difference between
median RT from the last block of repeating sequence and the last block of
random sequence for any of the ST-Extended Practice subjects (p=0.237;
mean median RT change = 3.4 msec; Figure 2). None of those in the ST-
Extended Practice group gained explicit awareness of the existence or content
of the practiced sequence.
30 i
| 20
h 10 H
tx
C 0
ra-10
c
C O
£-20
| -30
"S -40
C -50
c a
I -60
-70
-A
1
2
Days
Figure 2. RT change across 3 practice days for ST-Extended Practice group.
Each point represents median RT change score for that day’s practice. The
zero line is the median RT for the second random condition from the last day
of practice. Despite extended practice mean median RT did not decrease for
the sequence relative to the random condition. Error bars are the standard
error of the mean (SEM). The zero line is the median RT for the second
random sequence block.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
The third experiment determined the impact of explicit knowledge prior
to sequence practice. The task and sequence were explicitly explained to four
different individuals with unilateral stroke (ST-Explicit Knowledge). Before
practice, subjects were given schematic representations to study that
contained both spatial and color information of the repeating sequence. ST-
Explicit Knowledge subjects were not allowed to physically practice the
sequence and were instructed to indicate to the tester when they felt they
knew it well enough to begin practice; each spent approximately 15 minutes
studying the sequence. To ensure that the sequence was explicitly learned
prior to SRT task practice a pre-test was administered. This test comprised
the recognition and recall elements of the standard explicit testing procedure
used in experiments 1 and 2. Following the pre-test the 9-element sequence
was practiced following the same procedure described in experiment 1.
Immediately following practice, recall and recognition were re-tested.
Over practice, subjects in the explicit knowledge group decreased
median RT on average by 46.7 msec. RT at the end of practice was
significantly less for the repeating sequence compared to the random
sequence block (p=0.016; Figure 3). Analysis of the pre-test data
demonstrated that those in the ST-Explicit Knowledge group gained only
partial explicit knowledge; three of four subjects demonstrated recognition of
the correct sequence and recall was at a chance level. Following SRT task
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
practice, all four ST-Explicit Knowledge subjects demonstrated sequence
recognition, yet recall only improved for one subject.
£
o n
c
< D
O )
c
c d
•O
( 1 )
c
to
< 1 >
30
20
10
0
-10
-20
-30
-40
-50
-60
-70
2 4 1 3
Blocks
Figure 3. RT change across practice for ST-Explicit Knowledge group. A
significant decrease in RT for the last block of sequence practice as compared
to the random condition was noted. Error bars are the standard error of the
mean (SEM). The zero line is the median RT for the second random
sequence block.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
< D
D )
C
0
I —
0 1
c
ffl
T J
< D
C
ca
( D
80
70
60
50
40
30
20
10
0
ST-Extended ST-Explicit ST-Unaware
Practice Knowledge
Figure 4. Mean change in RT for ST-Unaware, ST-Extended Practice, and
ST- Explicit Knowledge groups. Positive numbers demonstrate a decrease in
sequence RT relative to the random condition and reflect implicit learning.
The ST-Explicit Knowledge group demonstrated a significantly larger decrease
in median RT when compared to both the ST-Unaware and ST-Extended
Practice groups.
To compare the results across experiments, a one-way ANOVA was
performed using Group as the between-subject factor and Median RT change
score for the last block of practice as the dependent measure. A significant
group main effect was found (p=0.001; Figure 4). Post-hoc testing (Scheffe
test) revealed significant differences between the ST-Explicit Knowledge and
ST-Unaware groups (p=0.003), as well as between the ST-Explicit Knowledge
and ST-Extended Practice groups (p=0.002).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
The inability of individuals post unilateral stroke to demonstrate implicit
motor-sequence learning when they are unaware of the sequence (both under
conditions of brief and extended practice), and using the limb ipsilateral to the
lesion demonstrates an important deficit in SRT motor learning for this
population that has not been previously described. The fact that these deficits
are attenuated by the provision of explicit knowledge prior to physical practice
complements previous work in motor learning post-stoke in which subjects
received feedback (knowledge of results) and detailed instructions regarding
strategies for successful task completion. In this study, subjects in the
extended practice group did not gain explicit awareness, (despite three times
as much practice as those in the explicit knowledge group) and were unable to
demonstrate implicit motor-sequence learning; lack of awareness of the
sequence disrupted implicit motor-sequence learning even when practice was
increased.
As has been previously discussed (Chapter 2), the functional neuronal
substrates for implicit learning likely include the motor, premotor, and
prefrontal regions as well as areas of the basal ganglia and cerebellum. The
results from Boyd and Winstein (2001) suggest that at least deficits in SRT
implicit motor-sequence learning associated with damage in the sensorimotor
cortical areas may be attenuated with complementary use of the explicit
memory system (i.e. medial temporal diencephalic network) at least for the
SRT task. Additionally, the deficits noted in motor-sequence learning for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
unaware subjects, using the limb ipsilateral to brain damage, provides further
evidence for the role of the ipsilateral hemisphere in higher order planning and
organization of learned motor actions.
Our results, together with previous motor learning findings in stroke,
suggest that explicit knowledge prior to physical practice may benefit implicit
learning in individuals with stroke-related brain damage while extended
practice alone does not. Despite these preliminary findings, the exact
relationship between explicit knowledge and implicit learning during motor-
sequence acquisition remains unclear. More importantly, as this work
examined individuals whose stroke was located in the sensorimotor cortex (or
the output from the sensorimotor cortex), it is not known if and how explicit
knowledge can be used to lessen implicit learning deficits following damage to
the cerebellum or basal ganglia. Finally, it should be noted that the small
number of subjects in this study limits the generalizability of the findings. The
data collected for this dissertation were a direct outgrowth from the preliminary
work discussed above. It is a goal of this current work to address the impact
of lesion location on SRT learning, with and without, explicit knowledge in a
larger group of individuals with stroke-related brain damage.
Task Type and Implicit Motor-Sequence Learning
The ability of individuals with stroke-related brain damage to
demonstrate implicit motor-sequence learning, and the benefit of prior explicit
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
knowledge may depend in part on the type (e.g. level of cognitive and motor
complexity, unique motor demands, information processing load) of the task
being learned as well as the content of the instructions provided. Information
from studies of implicit motor-sequence learning demonstrate that there is a
relatively inverse relationship between task demands (defined in terms of
sequence length, composition, and attentional requirements) and the rate of
implicit learning; if the amount of practice is equated across tasks, as
complexity (indexed by longer sequences, dual tasks, and low probability
between responses) increases implicit motor-sequence learning decreases
(Stadler, 1992; Stadler, 1995; Wulf & Shea, 1998).
For the purpose of this discussion task type will be considered in two
ways, first from a psychological construct that reflects the “cognitive” demand,
and second, from a motor skill demand (or information processing load) point
of view. In an implicit sequencing task, cognitive demand may be manipulated
in three ways - sequence length, composition, and attentional demand. As
sequences become longer it has been shown that they in turn become more
difficult (i.e. require more practice) to learn (Cleeremans & Jimenez, 1998).
Stadler (1992) demonstrated that for the SRT task as the statistical
structure of a sequence (e.g. probability or rules that govern composition)
increased so did the magnitude of reduction in reaction time. An example of
this type of structural predictability would be a sequence in which the color red
was always followed by blue then green. In this instance, subjects would only
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
have to learn the rule red, blue, green and not a more complex or ambiguous
relationship. When sequences are governed by some type of rule, such as the
example above, they are considered to be “probabilistic”; conversely, when
sequences consist of randomly presented stimuli that repeat they are
“ambiguous” (Cleeremans & Jimenez, 1998).
An additional factor that increases the difficulty of implicit sequence
learning is the inter-trial interval. Stadler (1995) showed that when the time
interval between the individual elements of the sequence are long (in this
experiment 2000 msec) an interference effect occurred and healthy control
subjects failed to demonstrate implicit motor-sequence learning of the SRT
task. It may be that over the course of practice, when long pauses are
inserted between individual responses the formation of associations within
each sequence unit is disrupted. This in turn, negatively impacts the ability to
predict the next response based on prior experience with the current one, and
implicit motor-sequence learning is diminished.
It has also been shown that as attentional resources are devoted to
other (typically explicit-type) tasks during implicit motor-sequence learning, the
magnitude and/or speed of implicit motor-sequence learning decreases
(Frensch et al., 1994, Cleeremans & Jimenez, 1998). In the dual task design,
subjects are asked to simultaneously practice an implicit motor-sequencing
task and perform a second “distracter” task that evokes the declarative
system. These secondary tasks often involve counting auditory stimuli. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
true effect of imposing a second task on implicit motor-sequence learning has
been hotly debated. It is presently unclear whether the interference effect of
the secondary distracter task is on attention (Frensch et al., 1994), the
development of explicit knowledge for the task (Jimenez et al., 1996), or the
development of implicit knowledge (Stadler, 1995). A broad discussion of
these topics is beyond the scope of this work, however, it is undisputed that a
secondary task that places demands on attention during implicit motor-
sequence learning poses a significantly more complex motor problem.
Motor demand is the second level of task type important for implicit
sequence learning. Simpler tasks might be defined as those that necessitate
less amounts of information processing. Past work has defined simpler tasks
as those invoke fewer degrees of freedom of motion (Bernstein, 1967), or take
less practice to reach asymptotic performance (Wulf, 2001). Conversely, more
difficult or complex tasks would promote higher information processing
demand and require more practice to achieve stable performance levels (Wulf
2001).
Until very recently very little attention has been paid to the impact of
motor demand using implicit sequence learning paradigms (Pew, 1974; Green
& Flowers, 1991; Wulf & Schmidt, 1997), however, this topic has been
considered in the motor learning literature (for examples see Wulf & Shea,
2001; Wulf, Shea, & Matschiner, 1998; Wulf, Hob, & Prinz, 1998). Very
recently Shea, Wulf, Whitacre, and Park (2001, in press) showed that younger
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
(University students) individuals without any neurologic damage were able to
show implicit learning of a complex stabilometer task. Subjects’ errors or
deviations from the target were assessed while “surfing” a waveform on a
stabilometer; unknown to these individuals the middle third of the target
waveform was identical and repeated (experiment 1). By the retention test,
the errors made while surfing the repeated middle third were significantly less
than those made on the random first and last thirds. To our knowledge, this
study is the first (and only) to assess whole body (many degrees of freedom)
implicit motor-sequence learning.
In their exhaustive review on the subject of task complexity in motor
skill learning, Wulf and Shea (2001) came to several conclusions. In this work,
complex motor skills were considered to be those that required the control of
several degrees of freedom, that were novel to the learner, and required large
amounts of practice before performance reached asymptote. Several factors
were noted as having the opposite effect on complex and simple skills
including, contextual interference, attentional focus, feedback frequency, and
physical guidance. Intuitively, learning more complex skills should require
more practice. In addition, Wulf and Shea (2001) questioned the validity of
generalizing findings from studies using simple motor skills to more complex or
real life tasks / situations. Although this work did not distinguish between
motor skill and implicit motor-sequence learning, it appears likely that the
same conclusions would result from a similar consideration in implicit motor-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
sequence learning—as motor task demands increase more practice will be
necessary for successful implicit learning.
Task complexity therefore, may be manipulated by altering length,
attentional load, sequence composition, or information processing demands.
To date most research focused on implicit motor-sequence learning has
investigated the effect of changing cognitive demand (e.g. sequence length,
composition, and attentional load). This work seeks to extend this work and
examine implicit motor-sequence learning of two different tasks; one discrete
motor task requiring finger movement (SRT task) with a speed goal (move as
quickly as possible), and a continuous upper extremity pursuit tracking task
with an accuracy goal (track as accurately as possible). To date no work has
systematically examined the performance changes of the same group of
individuals on two different task types, and it is unclear whether the same
pattern of change will be evident across tasks. From the data reviewed we
formed the following hypothesis:
7) Implicit sequence learning (as indexed by change in reaction time or
tracking error) will be less for a continuous visuomotor-sequence tracking
task as compared to a discrete stimulus-response sequence task. Further,
following unilateral damage to the sensorimotor cortex, basal ganglia, or
cerebellum deficits in implicit motor-sequence learning will be more
pronounced for a visuomotor-sequence tracking task compared with that
for a stimulus-response sequence task.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
Retention and Transfer Test Designs and Implicit Motor-Sequence
Learning
Retention Test Design
Classically defined, motor learning “is a set of processes associated
with practice or experience leading to relatively permanent changes in the
capability for movement” (Schmidt & Lee, 1999, p. 264). The issue of which
aspects of task practice induce a lasting or “permanent change” in motor
abilities is important in assessing the amount or degree of motor learning that
has occurred. When studying skill learning the distinction between the
temporary effects of practice and the more permanent effects of learning are
critical. To this end, Salmoni et al. (1984) argued that the use of a retention
test design is one method for separating temporary effects of practice on
performance from the lasting changes in skill associated with learning.
Several examples of the importance of this distinction were illustrated by
Salmoni et al. (1984) who thoroughly reviewed the literature and demonstrated
that constant feedback during motor skill learning appears to be beneficial
during practice but actually degrades learning (as shown via a retention test).
In the motor learning literature the distinction between performance and
learning is commonly considered, and is a typical feature of study design.
Unfortunately very few studies that have focused on the implicit learning
system have employed such experimental controls (for exceptions see Wulf &
Schmidt, 1997; Frensch, et al., 1998; Frensch, et al., 1999; Vakil et al., 2000;
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
Boyd & Winstein, 2001; Shea et al., 2001). It is a goal of this dissertation to
separate the immediate performance effects of practice from the long-term
changes associated with implicit motor-sequence learning via the use of
retention tests.
Transfer Test Design
Transfer of implicit learning to related tasks would indicate whether the
knowledge acquired during implicit learning is stored as an abstract motor
program or as a specific movement plan. Glisky and Schacter (1987, 1988,
1989) found that the knowledge acquired by a severely amnesic patient was
relatively inflexible - she demonstrated difficulty transferring learning to novel
situations. However, in a follow-up study Butters, Glisky and Schacter (1993)
studied 6 individuals with amnesia and showed not only that they were able to
transfer newly learned semantic information but that this effect was improved
by increased practice. Shimamura and Squire (1988) also demonstrated
transfer in a sentence completion task that was similar to that seen in normal
control subjects. However, they used familiar words and subjects may have
been able to use past knowledge that would support improved transfer.
Specific to implicit motor-sequence learning, several authors (Pew,
1974; Wulf & Schmidt, 1997) have demonstrated that implicit learning of a
continuous tracking task was stored as an abstract representation and could
be transferred. Wulf and Schmidt (1997) had young healthy control subjects
practice a continuous 30 second tracking task unaware that the middle third of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
the curve was a repeating pattern. Subjects were not only able to significantly
decrease the number of errors over the repeated segment with practice but to
transfer this learning to novel speeds (temporal scaling) of the pattern.
Transfer of learning is increased when the original sequence is governed by a
particular set of rules or probabilities (Reed & Johnson, 1994; Stadler, 1992;
Berry & Broadbent, 1988; Bapi, Doya, & Hamer, 2000).
Transfer learning has also been demonstrated when the spatial
dynamics of the task are altered (Bapi, Doya, & Harner, 2000). Following trial
and error sequence learning, subjects were able to show transfer effects when
the keypad they used to respond to the repeated sequence was rotated in
space. This demonstrates one form of effector independence in the learning
of the sequence task, as the rotated condition required different finger
movements. In Bapi et al.’s (2000) study trial and error sequence learning
was used and thus, subjects had explicit knowledge of the sequence. There is
one report (Lee & Vakoch, 1996) that attempted to separate explicit and
implicit learning and assess transfer effects. Subjects completed a simple
rule-number (explicit) task, in which they were aware of the rules for success
and a complex rule-line (implicit) task where they were unaware of the
regularities in their responses. When asked to transfer their learning to new,
yet similar, simple and complex tasks, subjects demonstrated positive transfer
of their explicit learning but not of their implicit learning. Lee and Vakoch
(1996) concluded that explicit knowledge of the simple task allowed individuals
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
to take advantage of what had been learned previously and incorporate this
into the transfer task. Implicit knowledge may have caused a type of
interference effect, where subjects were unable to beneficially adapt the
general rules that had already been learned to the new task.
Some authors have speculated that one function of explicit knowledge
is to promote the transfer and generalizability of the implicit motor plan (Segar,
1994; Curran, 1989). This conceptualization is in direct contrast to that
presented by Lee and Vakoch (1996). Therefore the manner for and
mechanisms by which transfer is affected by explicit knowledge remain
unclear. Thus, we propose to test this hypothesis as a part of our research
objectives.
8) The information acquired during implicit motor-sequence learning is stored
as a fundamental, abstract dynamic motor plan and not as a static
representation of the practiced movements. Therefore, if implicit learning
occurs, the fundamental nature of the learned pattern (i.e. motor program)
will facilitate performance during novel scaling of similar sequences (i.e.
transfer).
Summary
In sum the work discussed in this chapter leads to several interesting
conclusions and tempting hypotheses. First, following unilateral stroke-related
brain damage individuals are able to demonstrate motor learning if they have
conscious, explicit awareness of, or feedback for, the task goals and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
requirements. Presently, it is unclear if these findings will generalize to implicit
motor-sequence learning and indeed it appears from our previous work that it
may not; individuals with brain damage resulting from stroke appear to be
impaired in implicit motor-sequence learning, particularly when they do not
have access to explicit knowledge. Confounding this conclusion, are the
effects of task complexity (particularly motor demand) on implicit motor-
sequence learning which have not been well investigated or described.
Further, the relatively permanent effects of implicit motor-sequence learning
have not been documented as very little of this work has incorporated
retention or transfer test designs. These data form the basis for the last two
hypotheses for this dissertation.
In the next chapter the methods used for this dissertation will be
presented. Then the following three chapters will discuss the data that was
collected to address the hypotheses that have been presented. In addition
interpretation of the results of this study in light of the extant literature will also
be discussed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
CHAPTER 4
Research Design and Methods
Overview
Chapter 4 will detail the methods used in this dissertation, including
descriptions of the research design, subject groups, testing procedures, and
outcome measures (dependent variables). The following three chapters (5, 6,
and 7) will present results. Specific statistical methodologies will be described
in the last three chapters in conjunction with the hypotheses they were
designed to test.
To determine the relationship between behavioral deficits in implicit
motor-sequence learning and focal brain damage, three groups of individuals
with unilateral stroke were studied along with an age-matched control group
(Hypotheses #1, 3, and 5). To address the influence of explicit knowledge
(EK) on implicit motor-sequence learning (Hypotheses #2, 4, and 6)
participants were further divided into two groups; half of all subjects were
randomly assigned to the explicit knowledge (EK) group and were
incrementally provided with explicit knowledge over the three practice days.
Subjects (healthy control and focal stroke) in the no-explicit knowledge groups
(No-EK) returned each day for implicit motor-sequence practice but were not
provided with any explicit knowledge or instructions for the tasks. Two
different motor tasks (SRT task and the continuous tracking (CT) task) were
used to evaluate the impact of motor task complexity on implicit sequence
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
learning (Hypothesis #7). Finally, to probe whether the learned skill was
stored as an abstract motor plan, transfer capability to novel sequences was
also tested (Hypothesis #8).
Research Design
Four groups of subjects were tested, three with focal brain damage
localized in either the cerebellum, basal ganglia, or involving the sensorimotor
cortical areas, and healthy controls. Both tasks used in this dissertation were
unimanual and to minimize the impact of disrupted effector control, individuals
with focal brain damage always used the “less” involved extremity (see details
below). Individuals in the healthy control group were matched for arm use. By
random designation, half of those in each group were provided with explicit
knowledge prior to practice and half were kept unaware of the sequences
being practiced. Every subject practiced two different motor-sequence tasks
(the SRT and CT tasks) over three days. On day 4, retention and transfer
were tested. The amount of explicit knowledge gained with practice by
subjects in the No-EK group was assessed via explicit knowledge tests
following transfer testing at the end of day 4. Explicit knowledge was tested at
the end of each day of practice for those in the EK group. In addition, on day
3 (when subjects in the EK group were provided with full explicit knowledge
prior to practice) it was tested both before and after practice (Table 1).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
Table 1. Experimental Design and Overview.
Task SRTI CT
(Practice order determined randomly)
Group
Cerebellar
Sub-
Group
Day 1
Day 2
Day 3
Day 4
EK
EKO
Practice
EK test
EK1
Practice
EK test
EK2
Practice
EK test
Retention
Transfer
No-EK
I
EKO
Practice
EKO
Practice
EKO
Practice
Retention
Transfer
EKT
Basal
Ganglia
Sensori
motor
Healthy
Controls
A
EK No-EK
1 1
/
EK
1
\
No-EK
1
EKO EKO
▼
EKO
▼
EKO
P
EKT
P P
EKT
P
EK1 EKO EK1 EKO
P
EKT
P P
EKT
P
EK2 EKO EK2 EKO
P
EKT
P P
EKT
P
R R R R
T T
EKT
T T
EKT
7 \
EK No-EK
I I
EKO
P
EKT
EK1
P
EKT
EKO
P
EKO
P
EK2 EKO
P P
EKT
R
T
R
T
EKT
Note: All subjects participated each day in random order in both the simple and complex motor
tasks. Each group was divided evenly into those who received explicit knowledge prior to
practice on days 2 and 3 and those who did not. (EK=explicit knowledge, EK0=no prior
explicit knowledge, EK1=prior awareness of sequence existence, EK2=prior full explicit
knowledge, P=practice, R=retention, T=transfer, EKT=explicit knowledge test)
Subjects
Thirty-seven individuals with stroke were recruited and divided into
three groups based on the site of neurologic damage: cerebellum (CB; n=7),
basal ganglia (BG; n=10), or sensorimotor cortical areas (SMC; n=10).
Inclusion criteria for individuals with stroke were: 1) focal, unilateral damage in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
the cerebellum, basal ganglia, or sensorimotor cortical areas (confirmed by
MRI or CT scan); and 2) at least 6 months post-stroke. To ensure
homogeneity between groups (focal brain damage and healthy controls) all
subjects were also right hand dominant (determined by participant self-report)
and did not present with any evidence of dementia (score of at least 27 on the
Mini-Mental State exam). Exclusion criteria for all subjects included the
following conditions (and were determined by subject interviews and / or
medical records review): 1) acute medical problems; 2) uncorrected vision
loss; 3) previous history of psychiatric admission; and 4) history of multiple
strokes, transient ischemic attacks (TIA), or extensive cortical white matter
disease. Individuals with stroke were recruited from the outpatient clinical
services at the University of Southern California Healthcare Consultation
Center, Rancho Los Amigos National Rehabilitation Center, and the South
Bay Stroke support group. Individuals in the healthy control group were
recruited from the local community. All participants signed an approved
institutional informed consent form as well as a medical records release form
prior to testing.
Lesion Location
Prior to inclusion in this study lesion location was determined for all
subjects in the stroke groups. Review of an existing MRI or CT scan
confirmed lesion location. Medical records were obtained with written consent
of each participant. Digital photographic images were made of each brain
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
scan. Representative images demonstrating the extent of each individual’s
brain lesion are presented by group in Appendix 1-3 (Appendix 1 Cerebellar
Stroke Group; Appendix 2 Basal Ganglia Stroke Group; Appendix 3
Sensorimotor Cortical Area Involvement Stroke Group). Specific descriptions
of the individuals that comprised the cerebellar (CB), basal ganglia (BG),
sensorimotor cortical involved areas (SMC), and healthy control (HC) groups
are detailed below.
Cerebellar Stroke Group
Seven individuals with unilateral focal stroke located in their cerebellum
participated in this study (Appendix 1). Four were randomly assigned to the
EK group (CB EK) and three to the No-EK group (CB No-EK). Of those in the
CB EK group, three had right-sided brain damage and one had left sided
damage. There were two males and two females in this group; average age
was 57.7 years while time post stroke averaged 31.5 months. In the CB No-
EK group there were three males, two with right sided and one with left-sided
brain damage. Average age was somewhat younger than the CB EK group
(41.3 years), however, time post stroke was longer (62.3 months). For a
summary of subject information including stroke impairment (upper extremity
Fugl-Meyer motor score) and cognitive status (mini-mental status exam) for all
groups see table 2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
Basal Ganglia Stroke Group
Ten individuals with unilateral focal stroke affecting their basal ganglia
were also included in this study (Appendix 2). Five were randomly assigned to
the BG EK group (BG EK; 4 right, 1 left-sided damage) and five to the No-EK
group (BG No-EK; 4 right, 1 left). Of those in the BG EK group, four were
male and one was female. Average age was 51.0 years and time post-stroke
was 27.8 months. In the BG No-EK group, there were also four males and
one female (mean age 58.2 years). These individuals were on average 10.4
months post-stoke.
Sensorimotor Cortical Area Stroke Group
Initially, it was intended that the SMC group would consist of individuals
with focal stroke that only affected cortical regions associated with motor and
sensory functions. However, after intensive subject recruitment only two
individuals were identified that met this criteria (SMC 2 and 7). Therefore, the
inclusion criteria for this group was expanded to permit individuals with
damage that included, as well as extended beyond, the sensorimotor cortical
areas (Appendix 3). Ten individuals met these criteria and were included in
this study. Of these 10, five were randomized into the EK group (SMC EK; 3
left and 2 right-sided brain damage). In the SMC EK group there were three
females and one male. Average age was 59.0 years and time since stroke
onset was 33.4 months. The No-EK group (SMC No-EK) was composed of
three right sided and two left sided lesions. These individuals were similar in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
age to the SMC EK group (58.6 years) and were on average 48.0 months post
stroke.
Table 2. Subject Characteristics.
Lesion
Side
Sex Age (sd)
Post-
Stroke
Duration
(sd)
MMSE
(sd)
Fugl-
Meyer
Motor*
(sd)
Cerebellum
CB EK 3 Right, 2 Male, 57.75 31.5 27.5 49.7
1 Left 2 Female (15.0) (29.0) (1.4) (23.2)
CB No-EK 2 Right, 3 Male 41.33 62.3 27.0 60
1 Left (15.6) (54.9) (1.6) (0.0)
Basal Ganglia
BG EK 4 Right, 4 Male, 51.0 27.8 28.0 47.8
1 Left 1 Female (9.8) (28.2) (1.4) (19.5)
BG No-EK 4 Right, 3 Male, 58.2 10.4 (5.6) 28.4 44.4
1 Left 2 Female (14.6)
(1.1)
(16.0)
Sensorimotor
SMC EK 3 Right, 2 Male, 59.0 33.4 29.0 30.2
2 Left 3 Female (10.5) (18.9) (1.2) (21.2)
SMC No-EK 3 Right, 4 Male, 58.6 48.0 27.8 26.8
2 Left 1 Female (19.2) (30.1) (9.8) (18.7)
Healthy Hand
Control Used
HC EK 3 Right, 1 Male, 55.4 29.8
_
2 Left 4 Female (11.0) (0.4)
HC No-EK 3 Right, 2 Male, 57.4
_
29.6
. .
2 Left 3 Female (16.1) (0.5)
Note: Post-stroke duration is in months. MMSE = Mini-Mental Status Exam. Sd = standard
deviation. *=Upper Extremity Fugl-Meyer Motor Score, 66=maximum.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
Healthy Control Group
Individuals in the healthy control group were approximately matched for
age and then randomly designated into either the EK (HC EK; n=5) or No-EK
(HC No-EK; n=5) groups. All were right handed, however, for testing
purposes they were matched for arm use. Thus, during practice six used their
right upper extremity and four their left. The designation of which subjects
would use their non-dominant hand was made randomly. The HC EK group
was composed of four females and one male. These individuals were on
average 55.4 years of age. In the HC No-EK group there were three females
and two males. On average this group was 57.4 years of age.
Instrumentation and task, procedures, and outcome measures for both
the SRT and CT tasks are described separately below.
Procedure: Serial Reaction Time Task
Instrumentation and Task
The purpose of this experiment was to characterize implicit sequence
learning of a relatively simple motor-sequence. Four different colored circles
(yellow, red, blue, and green) could be displayed on the computer screen (17
inch, color) placed directly in front of the subject. When a colored circle was
not displayed, a large black asterisk (20-point font) appeared in the center of
the screen as a fixation device. A standard keyboard was placed on the table
directly in front of the computer screen. Centrally located on the keyboard, the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
letters “v, b, n, and m” had been capped with the colors yellow, red, blue, and
green.
Displaying one of the four colored circles on the screen generated
stimuli for movement. Only one colored circle appeared at a time; the other
circles were transparent at this time, however, each colored circle maintained
its relative location on the screen always appearing in the same position
throughout testing. The inter-stimulus interval varied from 500 to 800 msec.
Responses were made by pressing one of the four keys corresponding (in
color and location) to the appropriately colored light.
iO • # •
Figure 1: Set-up for the serial reaction time task. Subjects were
seated in front of a computer screen with their hand resting on the
table top and four fingers lightly touching the “v”, “b”, “n”, and “m" keys.
The keys were colored yellow, red, blue, and green.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
A computer software program (L. Boyd, 2000, E-Prime software
platform, Psychological Software Tools, version Beta 5.0) controlled the
appearance of the colored circles and recorded subjects’ responses.
Following every response, time data (response time) was stored for future
analysis as a text file on a desktop computer (Dell P90; Figure 1).
Procedure
Sequence Practice
Subjects were seated facing the computer screen with one hand
(ipsilateral to brain damage for BG and SMC groups, contralateral to brain
damage for CB group, and matched hand for HC groups) resting on the
keyboard. Four fingers of one hand were used to respond (all except thumb).
Following the cue to respond subjects pushed the key corresponding to the
colored circle. The correct key corresponded to the stimulus in both its color
and relative location on the keyboard. Subjects were instructed to respond as
quickly as possible by pushing the appropriate colored key as soon as the
stimulus for movement appeared. Subjects were not required to press the
correct key; as soon as a response was made the next trial began. Response
errors were stored for later analysis. After a key was pressed, the colored
circle disappeared, and a brief fixation asterisk appeared before the next
stimulus (inter-stimulus interval varied from 500-1000 ms).
The implicit sequence practice procedure was identical for all groups.
All subjects practiced the same 10-element ambiguous sequence (B-Y-R-B-G-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
R-B-R-G-Y). The beginning and end of each sequence was not marked in any
way. Each block of responses was composed of 10 repetitions of the
sequence (100 responses). In each session, subjects practiced the sequence
a total of 50 times (5 blocks of practice). A short break (1-2 minutes) was
provided at the end of each block of responses.
To control for the effects of non-specific learning, an initial block of 4-
choice random sequence responses were practiced (100 responses). Next, 4
blocks of sequence practice and a second 4-choice random sequence practice
block was performed. Finally, subjects practiced one last block of the
repeating sequence. In sum, subjects practiced the repeating sequence for 5
blocks (50 times through the sequence; 500 responses) and the random
sequence for 2 blocks (200 responses) each day.
On day four, retention and transfer were tested. Retention indexed
learning of the SRT task and was assessed by a performance of one block of
the sequence. SRT task transfer learning was tested by re-ordering the
practiced sequence. The transfer sequence was composed of three sub-units
(chunks) of the practiced sequence placed in a new order (i.e. B-R-G-Y—B-Y-
R—B-G-R).
Explicit Testing
Explicit knowledge was tested at different time points for the two
groups. For the EK group explicit testing occurred at the end of practice on
days 1 and 2, and at both the beginning and end of practice on day 3. Explicit
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
101
testing was performed only on the last day (following retention and transfer
tests) for those in the No-EK group (day 4). The first explicit test consisted of
the following: A short questionnaire designed to assess subjective awareness
of the sequence’s existence and composition was administered first (adapted
from Willingham et al., 1989; Table 3).
Table 7. Subjective Questionnaire for Explicit Tests.
SRT task
1)
Did you ever notice anything about the task?
2)
Did you ever notice a pattern?
3)
If yes, what was it?
CT task
1)
Did you ever notice anything about the task?
•
If noticed a sequence
Which portion of the screen was it located in?
•
If did not notice a sequence
There was a sequence. Which 1/3 of the screen do you think it was located in?
The beginning, middle, or end?
As those in the EK group were provided with explicit knowledge of the
existence of a sequence at the beginning of day 2, subjective awareness was
assessed only in the first test. Playing three different 10-element sequences
tested recognition memory. Subjects watched, without making responses, and
were asked to judge if they recognized the sequence as one they had seen
before. The true sequence that had been practiced and two foils were shown,
and for each the subject was asked, “Is this a sequence that you recognize?”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
102
Subjects also were required to rate their confidence in their recognition
decision on a scale of 0-100 (100 being totally confident they recognized the
sequence and 0 being not sure at all). Last, sequence recall was tested.
Three separate sub-units of the sequence were presented. In each, one
response was blank and the subject asked to fill in the appropriate color based
on what they recalled from their practice of the repeating sequence. Subjects
who claimed to have no awareness of the correct color were asked to make a
best guess (forced choice).
Outcome Measures
Response time (RT; reaction plus movement time) was the time
between stimulus onset to key-press. RT was measured and stored for each
trial. Median RT was calculated for each block of practice. To allow
comparison across subjects, and to eliminate the effect of grossly different
RTs, a change score was calculated for each block of practice (change RT =
median RT from the second block of random sequence practice on day 1
(SRT Rand2) minus median RT from repeating sequence blocks 1 to 15). We
chose to use the median response time from the second random block
(Rand2) in all our change score calculations. The Rand2 block was selected
for the change score calculation for two reasons. First, at this time the amount
of practice and explicit knowledge was equivalent between the EK and No-EK
groups. Second, enough practice had occurred by the Rand2 block to greatly
reduce (if not virtually eliminate) non-specific learning effects (non-specific
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
103
learning effects may have been manifested in the SRT task as reductions in
RT even across random sequences due to becoming familiar with the task).
Learning was assessed by calculating the difference between the median RT
from SRT Rand2 and the median RT from the retention test. Transfer learning
was evaluated in the same way (median RT from SRT Rand2 minus median
RT from transfer block).
Explicit testing was evaluated in two ways. First, a percent correct was
calculated for the subjective responses (e.g. 80% or 4 of 5 subjects noticed
the repeating sequence). Next, a recognition and recall percent correct were
calculated. There were three possible responses for each of these tests
(subjects could correctly identify the repeated sequence and correctly state
that the two foils had not been previously seen). Thus, by group the total
number of correct scores was divided by total responses, and a percent
correct was calculated. Finally, confidence scores were averaged across
groups.
Procedure: Continuous Tracking Task
Instrumentation and Task
A lightweight lever was attached to a frictionless vertical axle and
secured to a table, positioned parallel to the floor. To accommodate different
arm lengths, the handle at the end was adjusted for each subject. A linear
potentiometer was attached to the transducer at the base of the vertical axle
and the analog signal from this transducer was converted to digital by a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
National Instruments A/D board (shielded multifunction I/O board, #PCI-
6024E). This digital signal was then sampled at 200 Hz.
A target cursor (white X) moved from left to right across the screen (1.1
cm/sec) and was visible on a black background (LabView software, National
Instruments, Inc.). The task was to track the vertical path of the target with
movements of the lever. Subjects’ lever movements appeared as a green
open square. Subjects were unable to see the entire pattern (no visual
feedback); only the relationship between the target cross and subjects’
movements could be seen as both moved across the screen.
Unknown to the subjects the middle third of each tracking pattern was
repeated and identical across practice, retention, and transfer. This pattern
was constructed using the polynomial equation as described by Wulf and
Schmidt (1997) with the following general form:
/(x) = b0 + ai sin (x) + b -i cos (x) + a2 sin (2x) + b2 cos (2x) +...+ a6 sin (6x) +b6
cos (6x)
The middle (repeated) segment was constructed by using the same
coefficients for every trial (bo=2.0, a-i=-4.0, b-i=3.0, a2=-4.9, b2 =-3.6, a3 =3.9,
b3 =4.5, a4=0.0, b4=1.0, a5 =-3.8, b5 =-0.5, a6 =1.0, and b6 =2.5; Wulf & Schmidt,
1997). The first and third segments of the tracking pattern were generated
randomly using coefficients ranging from 5.0 to -5.0 and were different for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
every trial. In each third of the tracking pattern there were 10 separate
direction reversals. To ensure that smooth transitions existed between the
three segments of each tracking pattern, the end of segment one (first third)
and the beginning of segment three (last third) were adjusted vertically by the
software so they connected the ends of the repeated segment. Every
waveform was then screened visually prior to testing to ensure smooth
transitions across the three elements (see Figure 2 for sample waveforms).
Random 1 Repeating Sequence Random 2
140
120
100
6000 0 1000 2000 3000 4000 5000
Samples
Figure 2. Sample waveforms from the continuous tracking task with the
repeating third highlighted. The middle third of every trial was identical and
repeated. The first and last thirds were generated randomly and differed for
every trial.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
106
Procedure
Sequence Practice
Prior to each testing session the lever was calibrated. Practice
conditions were identical on days 1, 2, and 3. Subjects sat in front of the
monitor with their arm resting on the lever. Subjects in the CB group used the
arm contralateral to their brain lesion, those in the BG and SMC groups used
the ipsilateral arm, and the HC group was matched for arm use. Individuals
made arm motions from neutral to approximately 90 degrees of internal
rotation with the start position at 45 degrees (Figure 3).
Monitor
Figure 3. Set-up for the continuous tracking task. Subjects were seated
facing a dark computer screen with one arm resting on a frictionless lever.
Subjects moved the lever in an arc from 0 to 90° of internal rotation as they
tracked the target across the screen. Internal rotation moved the cursor down
while external rotation moved it up on the screen.
Trial duration was 30 seconds; a short break was taken between trials.
Subjects were instructed to track the target with movements of the lever as
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107
accurately as possible. Each day all groups tracked 50 trials (5 blocks). A
brief rest break (1-2 minutes) was provided approximately halfway through
each practice session following the third block.
As in the SRT task, on day four retention and transfer were assessed.
Retention testing was composed of one block of tracking practice. CT task
transfer testing consisted of two separate one-block sessions - to assess
spatial and temporal transfer. Re-scaling the repeated waveform (and the two
random segments) by 130% tested spatial transfer. Shortening the duration of
the entire trial to 24 seconds tested temporal transfer; the repeating waveform
was compressed to 80% of its original length; none of the features of this
sequence were altered. Each third of the target waveform still consisted of 10
reversals, however, the target now moved at 1.375 cm/sec.
Explicit Testing
For the EK group explicit knowledge of the repeated segment was
tested at the conclusion of day 1 and 2, and via a pre-test prior to practice on
day 3. Explicit knowledge was also tested for this group at the end of day 3.
For the No-EK group explicit knowledge was only assessed after the transfer
test on day 4. Explicit knowledge was tested first by subjectively asking
subjects if they had noticed anything about the task. If a repeating segment
was noticed, subjects were asked to locate it in the overall pattern. When
subjects failed to notice a repeating pattern in the sequence they were
informed that one had existed and asked to guess if it was in the beginning,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
middle, or end (adapted from Wulf & Schmidt, 1997; Table 3). Recognition
memory was tested by playing one true and two false cursor movement
patterns on the computer screen and asking subjects whether they believed
they had seen them before. As with the SRT testing, subjects were asked to
rate their confidence in their recognition decisions on a scale of 0 to 100 (100
being completely confident they had seen the pattern before).
Outcome Measures
Motor performance was evaluated across practice, retention, and
transfer in two ways for the CT task. First root mean squared error (RMSE)
was calculated for every trial. RMSE reflects the accuracy of the kinematic
pattern produced and is the average difference between the target pattern and
subjects’ actual movement trace.2 This score was calculated separately for
the two random and the repeating segment components of the pattern. As the
first third of every trial was random and unpredictable, subjects never knew
where the target would initially appear. Thus, it was noted that the first few
seconds of each trial were often spent catching-up to the target, and
subsequently, significantly higher RMSE scores were noted for this portion of
the waveform. Therefore, RMSE from the last random third of the trial was
used to index non-specific learning. To minimize the effects of non-specific
learning and the provision of explicit knowledge on random tracking
performance, RMSE from the last third of the waveform for the last block (10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
trials) of practice on day 1 were averaged and used as an index of error during
tracking performance (RMSE-Rand2). A change score was then calculated for
each block of practice (every 10 trials) by averaging RMSE from the repeated
middle third and subtracting RMSE-Rand2. A learning score was calculated in
the same fashion (average RMSE from the retention test minus RMSE-
Rand2). Change scores were also calculated for the transfer tests.
To determine the accuracy of subjects’ tracking across practice as well
as the magnitude of the time difference (lag time) between subjects’
movements and the target cursor, a time series analysis (TSA) was also
performed. In the TSA, the tracking pattern from the repeated middle third of
the trial was serially correlated with the target pattern. Both the tracked and
target waveforms are composed of an array of 2000 data points. In essence,
the subjects sequence tracking data point array was slid along the target data
point array until a maximal correlation coefficient was achieved. This
correlation coefficient was converted to a R2 value, which reflects the overall
accuracy of the subject’s tracking performance. The magnitude of the amount
that the tracking data were moved (or slid) along the target data was
converted to msec and represents the time lag of the subject’s tracking
performance. Therefore, this analysis derived average time lag and accuracy
of tracking performance for the repeated middle third of the pattern (Figure 4).
2RMSE = (I(x ,-T )2/n ) 1 /2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
Subject’s response “slid” on target
until max correlation found
Subject A
140
Target
120
100
r = 0.94, time
lag =
155 msec (3 1
intervals)
60
0 500 1000 1500 2000
Samples
B Example of maximally correlated subject track. The distance the waveform was
moved was converted to time and expressed as msec “lag”.
140
120
100
2000 500 1000 1500 0
Samples
Figure 4. Sample target and subject waveforms from the second day of practice. (A)
Example of the repeated middle third of the waveform along with a representative
subject response. The correlation coefficient (r) and the time lag of the subject’s
response are also presented. (B) For the time series analysis the subject’s response
was “slid” along the target, with correlation coefficients calculated serially -- for every
interval the subject response was moved. When the correlation coefficient reached a
maximum, the two waveforms were considered a best fit. The magnitude of the
distance the subject’s waveform was moved was converted to time (msec) and
expressed as tracking lag.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
111
For comparison, the same TSA was calculated for the Rand2 segment
(the last 10 trials of the second random segment of the waveform at the end of
day 1 of practice). The Rand2 TSA calculation was performed using 1900
data points (rather than the 2000 used for the repeated middle third) due to
end effects.
Explicit testing was summarized using the same method described in
the SRT task outcome measures with two exceptions. First, the percent of
subjects who correctly identified the middle third as the repeating segment
was calculated (e.g. 1 of 5 guessed middle third or 20% correct). Second,
there was not a recall component to the CT task explicit testing procedure.
Implicit and Explicit Task Order
Practice order of the two tasks was randomly designated on day 1 and
counterbalanced across the next three days for each subject. Once task order
was determined, each day of practice followed the same protocol. For the EK
groups, day 1 consisted of sequence practice only. At the conclusion of
practice explicit knowledge was tested. On day 2, EK subjects were informed
that there was a repeating element to each task and following practice,
recognition and recall memory for the sequences were tested. On day 3, EK
subjects were explicitly instructed regarding the existence, composition, and
location of the sequence. EK subjects were also provided with a study period
and a schematic drawing of the sequence for both tasks. They were allowed
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
112
only to look at, but not physically practice the sequence during this study
period. A pre-test of explicit knowledge was then administered prior to
practice to test whether the sequence had been learned explicitly. This pre
test consisted of the recognition and recall elements of the explicit testing
protocol; explicit testing of recognition and recall was also performed after
practice. For the No-EK groups, days 1,2, and 3 consisted of practice only;
no instructions or explicit knowledge regarding the sequences were provided.
Explicit knowledge of the No-EK group was assessed at the end of day 4.
On day 4, both groups were given retention and transfer tests.
Retention was tested via one block of sequence practice for both tasks.
Transfer was assessed by one block of practice of a sequence that was
statistically similar to the one previously practiced.
Statistical Analyses
The specific statistical methods utilized are presented in Chapters 5, 6,
and 7 in conjunction with the hypotheses they were designed to test. For all
statistical calculations in this dissertation SPSS statistical software (9.0.1) was
used. Statistical significance was set at p<0.05 and when appropriate post-
hoc testing was performed using a Scheffe procedure.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
113
Summary
In order to answer the hypotheses presented in this dissertation four
groups of subjects were studied - three with focal stroke, and a healthy control
group. These individuals were also divided into groups that either received
explicit knowledge for the sequences being practiced or did not. Finally, to
better understand the impact of task type on the interaction between explicit
knowledge and implicit motor-sequence learning all participants practiced both
the SRT and CT tasks. Outcome measures indexed the amount of change in
performance on the repeating of the two tasks with respect to random
sequences.
The next chapter (5) will address the first group of hypotheses, detailing
the ability of individuals with focal stroke in the CB, BG, or SMC to
demonstrate implicit motor-sequence learning as compared to a group of age
matched healthy controls. Data will be presented from both tasks, with the
SRT task considered first, followed by the results from the CT task.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
114
CHAPTER 5
The Impact of Focal Brain Damage
on Implicit Motor-Sequence Learning
Overview
The purpose of this chapter is to address the first specific aim of this
dissertation; to determine the impact of focal brain damage on implicit motor-
sequence learning. In this analysis and discussion only data from the No-EK
groups (n=18) will be considered. The impact of explicit knowledge on implicit
motor-sequence learning will be discussed in Chapter 6. The data and
analyses presented here are specific to the three hypotheses that concern
implicit motor-sequence learning following focal brain damage. These
hypotheses proposed that
Following unilateral damage to the cerebellum implicit learning of motor-
sequences will be diminished relative to that for healthy controls. Normal
cerebellar function is essential for the learning of implicit motor-sequences
but this function is highly lateralized to the ipsilateral hemisphere during
unimanual tasks. Therefore, individuals with unilateral damage to the
cerebellum will be able to demonstrate some degree of implicit motor-
sequence learning using the arm contralateral to the lesion. (Hypothesis
# 1 )
Following unilateral damage to the basal ganglia, implicit learning of motor-
sequences will be diminished relative to that seen for healthy controls.
(Hypothesis #3)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
115
Following unilateral damage to sensorimotor cortical areas, implicit motor-
sequence learning will be diminished relative to that demonstrated by
healthy controls. (Hypothesis #5)
In this chapter, first the methods, including statistical analyses that were
used to address these hypotheses will be described. These will be followed
first by the results and lastly, by a discussion relative to previous work.
Methods
All of the individuals in the No-EK group practiced both the SRT and CT
tasks, and data from both will be presented in this chapter. A brief description
of the subjects is presented first, followed by data analysis methods used for
the SRT and CT task data respectively.
Subjects
In order to assess the ability of individuals with focal brain damage to
demonstrate implicit motor-sequence learning, data from the No-EK groups
(HC=5; CB=3; BG=5; and SMC=5) were analyzed. For a complete description
of subject characteristics see Chapter 4, table 2. All of the No-EK subjects
followed the practice and retention test schedule outlined in Chapter 4.
SRT Task Dependent Measures and Data Analysis
Response time (RT) was the primary dependent measure for the SRT
task. Practice data were grouped into 15 blocks (1 block =10 times through
the sequence); there was also a one block retention test. First, data were
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
116
screened for wrong responses. Across groups incorrect key presses were
uncommon and ranged from 2% for the HC group to 4% for the SMC group
(both the BG and CB groups made 3% incorrect key presses). Incorrect
responses were discarded from the data set and not included in any portion of
the data analyses. Response time is commonly viewed as highly variable and
susceptible to distraction. Therefore, in accordance with the bulk of other work
using the SRT task (see Nissan & Bullemer, 1987; Curran & Keele, 1993) we
calculated median RT by block. Thus, a median RT for each of the 10
repetitions by subject of the repeating (150) and random (60) sequences was
generated. Finally, mean median RT across subjects and by block for the
repeated sequence and selected random sequences (see below) were
calculated for each of the groups. This standard procedure was followed for
all of the RT data presented in this dissertation.
For the SRT task, implicit learning is indexed by comparing RT for the
repeated sequence with response time for random sequences. We choose to
index random sequence performance with the last random block of practice on
the end of day one (SRT Rand2). This point in practice was selected for two
reasons. First, it is late enough in practice (subjects have had 5 blocks of
exposure to the SRT task) that there should no longer be any significant non
specific learning effects. Second, the SRT Rand2 block of practice is the
latest point in practice where the exposure to explicit knowledge is equivalent
between the EK and No-EK groups. This is not relevant to this chapter as only
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
117
data from the No-EK groups is considered, however, for consistency all
repeating sequence SRT task performance and learning was referenced to
this point in practice.
To evaluate acquisition of the SRT task, a change score was calculated
for each individual in the No-EK groups by block of practice. The change
score was comprised of median SRT repeating sequence RT minus median
SRT Rand2 RT. This score reflects the amount of change or decrease in RT
that occurred across practice days (1 to 3) relative to performance of a random
response series.
To assess learning of the SRT task, the retention test data were used.
These were evaluated in two ways. First, absolute RT from the repeating and
SRT Rand2 sequences were compared by focal stroke group to that of the HC
group. As it has been previously reported that following stroke even
movements of the less involved arm are slowed (Pohl & Winstein, 1999), this
allowed consideration of the differences in speed of responses between the
groups. Next, a learning score was calculated by subtracting median
repeating sequence RT from SRT Rand2 for each individual. The learning
score reflects the magnitude of change seen at the retention test by
determining the difference between SRT Rand2 and the repeating sequence.
Therefore, a positive learning score indicates that implicit learning of the SRT
task has occurred.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118
To test our hypotheses we examined the SRT data in three ways. First,
to determine the reliability of change in the repeated sequence relative to SRT
Rand2 across acquisition, a Group (HC, CB) by Sequence (repeated, SRT
Rand2) by Block (day 1, day 2, day 3) ANOVA was performed with a repeated
measures correction for block and RT as the dependent measure. This same
analysis was repeated for the BG (Group (HC, BG) by Sequence (repeated,
SRT Rand2) by Block (day 1, day 2, day 3) ANOVA) and CB group (Group
(HC, CB) by Sequence (repeated, SRT Rand2) by Block (day 1, day 2, day 3)
ANOVA). Next we examined retention test data in two ways. First, to
determine whether the RTs of the groups were significantly different at
retention, a Group (HC, CB, BG, SMC) by Sequence (repeating, SRT Rand2)
ANOVA was conducted with retention test RT as the dependent measure.
Next, to test the reliability of differences in learning scores a one-way ANOVA
was performed using group as the factor (HC, CB, BG, SMC) and the learning
score as the dependent measure. When appropriate a Scheffe test was used
post-hoc to determine the locus of between group differences (p<0.05).
CT Task Dependent Measures and Data Analysis
Two different analyses yielded three separate dependent measures for
the CT task - an error score, tracking time lag, and tracking accuracy. First,
RMSE (a general measure of tracking error) was calculated. In the same
fashion as with the SRT task data, RMSE scores were calculated separately
for the repeated and random sequences. RMSE scores were averaged by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
119
block (every 10 trials) separately for random and repeated sequence
components. The CT task had two random components (the first and last
third), however, to index repeating segment performance only scores from the
last third (CT Rand2) were considered. Similar to the SRT task, RMSE from
the last one-third of each trial of the last block of practice at the end of day one
was used to index random segment performance. Given that the target for the
first third of every tracking trial initially begins on the left side of the screen,
and that this portion of the sequence is entirely random in its start position,
subjects had no way of knowing where it would appear. Thus, the first few
moments of sequence practice were often spent “catching up” to the target; a
situation which can artificially inflate the RMSE score for the first third of the
tracking trial. Therefore, RMSE from the first third of the CT task were not
considered in data analyses.
A change score was calculated using RMSE data and was equal to
mean RMSE CT Rand2 minus mean RMSE repeating segment by block. This
score reflects the amount of decrease in tracking error that occurred by block
across the three days of practice. Change score data from the last block of
practice for each day were used in statistical analyses. To index learning of
the CT task via changes in tracking error the repeated versus CT Rand2
sequences were compared by group for the retention test. Additionally, a
learning score (CT Rand2 RMSE minus retention test repeating sequence
mean RMSE) was calculated.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
As the CT task involved a more complex movement solution compared
to the SRT task, we were able to consider these data in temporal and spatial
dimensions. Therefore, a time series analysis was also performed on the CT
task data where the amount of time lag of the subject’s tracking movements
could be evaluated in conjunction with the accuracy of the response. The time
series analysis yielded two measures of tracking performance; 1) time lag
(msec) or the average time distance that the subject is behind the target as
they track it across the screen, and 2) a best fit correlation coefficient (r) which
was calculated by comparing the target and subject’s trajectory once a
correction for time lag was implemented. The correlation coefficient indexed
target tracking accuracy; (the procedure used to generate the time lag and
tracking accuracy data are described in Chapter 4 and summarized in Figure
4). For data presentations (figures) r was converted to a R2 value. Thus, as
subjects learn the repeated waveform, time lag should decrease as tracking
accuracy increases.
To test our hypotheses of altered implicit learning following focal stroke
compared to that for healthy controls we considered both acquisition and
retention data. To analyze tracking error across acquisition we used the same
analyses as previously described for the SRT task. For the CB group, a group
(HC, CB) by Sequence (repeated, CT Rand2) by Block (day 1, day 2, day 3)
ANOVA with a repeated measures correction for block and tracking error
(RMSE) as the dependent measure was completed. This procedure was
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
121
repeated for the BG and SMC groups. To evaluate learning, the retention test
data were considered. First, the absolute magnitude of tracking error for the
repeating sequence at retention and CT Rand2 were compared via a Group
(HC, CB, BG, SMC) by Sequence (repeating, CT Rand2) ANOVA with tracking
error (RMSE) as the dependent measure. Last, a one-way ANOVA
determined the reliability of between group differences with Group as the
factor (HC, CB, BG, SMC) and learning score as the dependent measure.
Post-hoc comparisons were done using a Scheffe test (p<0.05).
The time lag and tracking accuracy data were evaluated similarly. For
acquisition, time lag data from each focal stroke group was compared to the
HC group via a Group (e.g. HC, CB) by Sequence (repeated Rand2) by Block
(day 1, day 2, day 3) ANOVA with time lag as the dependent variable and a
repeated measures correction for block. Time lag at the retention test
essentially indexes subjects’ ability to predict the path of the target. Retention
test between group differences were tested via a Group (HC, CB, BG, SMC)
by Sequence (repeated, Rand2) ANOVA with time lag as the dependent
measure. All correlation coefficients were converted to Fisher Z scores prior
to statistical analysis. Then using the Z scores, accuracy during acquisition
was assessed for each focal stroke group compared to the HC group. Thus,
another series of Group (e.g. HC, CB) by Sequence (repeated, Rand2) by
Block (day 1, day 2, day 3) ANOVAs with repeated measures corrections for
block and Z scores as the dependent measure were conducted. To determine
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
122
between group differences at the retention test, a Group (HC, CB, BG, SMC)
by Sequence (repeated, Rand2) ANOVA was performed with Z scores as the
dependent measure. For all of the between group comparisons a Scheffe test
was used to identify the locus of between group differences when appropriate
(p<0.05).
Explicit Knowledge
Tests for explicit knowledge of the existence and composition of the
practiced sequence were administered at the end of day 4. Subjective
responses were tallied, while recognition memory was expressed as a percent
correct by group. Confidence ratings for recognition memory were also
calculated and averaged by group.
Results
SRT task
Acquisition Performance
Across the three days of acquisition, all groups showed progressive increases
in change scores indicating faster RTs with practice (Figure 1).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
123
H C
C B
B G
S M C
\-
< D
£ -50
o
Q .
( /)
< D
o c -150
c
< D
“ -250
c
(0
s z
O
-350
Day 1 Day 2 Day 3
Days of Practice
Figure 1. Change in response time (RT) for the repeated sequence across
acquisition on the SRT task for the No-EK groups. Error bars are the SEM.
The zero line represents random sequence RT. Data below this line illustrate
decreased RT and reflect implicit learning. All groups (healthy control (HC),
BG, CB, and SMC) decreased RT for the repeated sequence relative to
random with practice.
The SMC group showed the largest decrease in RT, as evidenced by their 257
msec change score at the end of day three. The next largest amount of
change in RT was made by the HC group, who decreased their RT by 103
msec at the end of day three. The BG and CB groups showed less change in
RT with practice (89 and 79 msec, respectively) at the end of day three (Figure
1).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
124
HC versus CB Group
Our comparison of HC and CB groups across SRT task acquisition
demonstrated that both groups improved with practice (Main Effect of Block
F(2,24) = 7.04, p=0.004; Figure 1). Further, across acquisition both groups
also decreased their repeating sequence RT relative to SRT Rand2 (Block By
Sequence Interaction F(1,12) = 7.75, p=0.016). The HC and CB groups
demonstrated similar ability to decrease their repeating sequence RT as
demonstrate by the absence of a Group by Sequence by Block interaction
(p=0.559).
HC versus BG Group
Again, both HC and BG groups improved with practice (Main Effect of
Block F(2,32) = 11.61, p=0.000; Figure 14). A Sequence by Block interaction
demonstrated that both groups were able to decrease repeating sequence RT
relative to SRT Rand2 (F(2,32) = 11.61, p=0.002). As demonstrated with the
CB group, the absence of a Group by Sequence by Block interaction
(p=0.529) showed that both groups similarly changed across practice.
HC versus SMC Group
Individuals in the SMC and HC groups, also showed significant change
across acquisition (Main Effect of Block F(2,32)=5.88, p=0.007) and a
Sequence by Block Interaction (F(2,32) = 5.88, p=0.007), indicating that both
HC and SMC were able to show significantly faster RTs for the repeating
sequence as compared to their SRT Rand2 score. There was not a significant
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
125
difference in the way that the HC and SMC groups changed across practice
(Group by Sequence by Block interaction p=0.799).
Retention Test
At the retention test the SMC group demonstrated the slowest repeating
sequence RTs (659 msec), followed by the BG group (550 msec), CB group
(454 msec) and HC group (343 msec). A Main Effect of Group identified a
significant between group difference in the speed of responding (F(3,28) =
4.89, p=0.007). Post-hoc testing demonstrated that it was the significantly
slower responses of the BG group (p=0.050) and SMC group (p=0.001) that
were the locus of this difference (Figure 2). In addition Main Effect of
Sequence was noted (F(1, 28) = 5.48, p=0.026) demonstrating that RT for the
repeated sequences were faster than that for the random ones and reflecting
implicit motor-sequence learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126
R a n d I I R e p
HC CB BG SMC
Groups
Figure 2. RT for repeated versus random sequence at the retention test on
the SRT task. Error bars are SEM. A Main Effect of Sequence showed that for
all groups RT was faster for the repeated as compared to that for random
sequences. Statistically reliable differences demonstrated that the repeated
sequence RTs for the BG and SMC groups were larger than that for the HC
group(Main Effect of Group).
Evaluation of the learning scores demonstrated no Main Effect of Group
(p=0.139). As a trend for between group differences in the learning score was
shown we calculated an effect size for the difference between the HC and
SMC groups. The analysis demonstrated that there was a large meaningful
difference between the magnitude of the learning score for the HC and SMC
groups (ES=0.82). These results show that all groups were able to reliably
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
127
decrease their repeating sequence RTs by retention. In addition, that the
magnitude of this change was either fairly similar to that seen for the HC group
(CB and BG groups) or exceeded it (SMC group; Figure 3).
500
400
o
< L >
( /)
E ,
< D
E
i-
8 1 300
c
o
Q .
V)
0
a:
200
< L >
O l
« 100
O
HC CB BG
Groups
SMC
Figure 3. Retention test learning scores for the SRT task. Error bars are SEM.
Positive numbers show a decrease in repeating sequence RT relative to the
random condition and reflect implicit learning. A trend for between group
differences was found. This was due to the larger magnitude of change in
repeating sequence RT shown by the SMC group compared to that for the HC
group. An effect size calculation confirmed this difference (ES=0.82).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
128
Explicit Knowledge
Analysis of the explicit knowledge test data identified that 80% (4/5) of
those in the HC group subjectively expressed having noticed the sequence,
however, they were only 66% correct for recognition and 40% correct for recall
memory (Table 1). Of the focal stroke groups the CB subjects acquired the
next most amount of explicit awareness by the end of practice for the SRT
task (66% noticed the sequence, 44% recognition and 11 % recall). The BG
group appeared to acquire the least explicit awareness of the sequence (0%
noticed, 46% recognition and 6% recall), while the SMC group was slightly
better (20% noticed, 53% recognition and 33% recall).
CT Task
Acquisition Performance-Tracking Error
For the CT task, all groups were able to decrease their tracking error
across the acquisition days (Figure 4). Again, the largest amount of decrease
was demonstrated by the SMC group, 8.2 degrees. Following the same
pattern as the SRT task, change scores were next largest for the BG group
(5.9), followed by the CB group (4.1), and then the HC group (3.8) by the end
of day three.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
129
— ▼—
H C
C B
■
B G
- -♦
S M C
Day 1
Day 2
Days of Practice
Day 3
Figure 4. Change in tracking error across acquisition for CT task No-EK
groups. Error bars are the SEM. The zero line represents random sequence
tracking error. Data below this line show decreased tracking error and reflect
implicit learning. All groups significantly decreased repeating segment
tracking error relative to that seen for the random segments across acquisition.
HC versus CB Group
Both the HC and CB groups decreased their tracking error across
acquisition (Main Effect of Block F(2,24)=10.23, p=0.001; Figure 17).
Additionally, these decreases in tracking error for the repeated segment made
it reliably different than the CT Rand2 for this group by the end of acquisition
(Block by Sequence interaction F(2,24) = 10.23, p=0.001). The two groups
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
130
changed similarly across practice (no Group by Sequence by Block interaction
p=0.809).
HC versus BG Group
Tracking error also decreased reliably across acquisition for the BG and
HC groups (Main Effect of Group F(2,32) = 20.01, p=0.000). A significant
Block by Sequence interaction confirmed that both groups improved their
repeating segment tracking error relative to CT Rand2 (F(1,12) = 4.745,
p=0.045. Similar, to the CB group, the two groups made similar changes
across acquisition (Group by Sequence by Block interaction p=0.191).
HC versus SMC Group
The SMC and HC group significantly reduced their tracking errors
across acquisition (Main Effect of Block F(2,32) = 6.59, p=0.004). This
improvement in repeated segment tracking performance was confirmed by a
Block by Sequence interaction (F(2,32) = 6.59, p=0.017) indicating that the
errors made for the repeated segment were less than those made on the CT
Rand2. The SMC group made larger tracking errors than the HC group on
both the repeated and Rand2 segments (e.g. SMC repeated 20.9, HC 13.0 on
day three of practice) and this contributed to a significant Group by Block
(F(2,32)=3.30, p=0.050) and Group by Sequence by Block interactions
(F(2,32) = 3.30, p=0.050).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
131
Retention Test - Tracking Error
The magnitude of repeating sequence tracking errors at the retention
test were much larger for the SMC group (20.9) than any of the other groups
(BG = 13.0, CB = 12.1, and HC = 11.2; Figure 5).
HC CB BG SMC
Groups
Figure 5. Tracking error for repeated and random sequences at the retention
test, CT task No-EK groups. Error bars are SEM. A Main Effect of Sequence
showed that for all groups less tracking errors were made for the repeated as
compared to that for random segments. The SMC groups’ repeated segment
tracking error was reliably larger than that for the HC, CB, or BG groups (Main
Effect of Group).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
132
Evaluation of the magnitude of tracking error made at the retention test
demonstrated a Main Effect of Group (F(7,28) = 8.11, p=0.008). A post-hoc
test revealed that this was due to the significantly larger magnitude of tracking
errors made by the SMC group as compared to the HC (p=0.001), CB
(p=0.018), and BG (p=0.008) groups (Figure 5). A Main Effect of Sequence
showed that tracking errors for the random segment were reliably larger than
the repeated segment (F(1,28) = 8.11, p=0.008).
15 r
HC CB BG SMC
Groups
Figure 6. Retention test learning score for tracking error for the No-EK groups.
Positive numbers show a decrease in repeating sequence RT relative to the
random condition and reflect implicit learning. No between group differences
were identified for the learning score; all of the groups demonstrated implicit
motor-sequence learning of the tracking task.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
133
Despite these large errors and contrary to both hypotheses of diminished
implicit learning for the BG and SMC groups, there were no between group
differences for the learning score (p=0.313; Figure 6). In sum, these findings
indicate that all the groups learned implicitly, and that the focal stroke groups
showed similar amounts of change to that demonstrated by the HC group.
Acquisition Performance - Time Series Analysis
HC versus CB Group - Time Lag
The time series analysis revealed several very interesting and distinct
findings. First, for the CB and HC groups there was not a main effect of Block
(p=0.152).
450
HC
380 CB
BG
SMC
310
240
170
1 _L
3
X
100
Days
Figure 7. Time lag of repeated segment tracking across acquisition. Error
bars are the SEM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
134
This was the result of the CB group’s inability to decrease their time lag across
acquisition; the HC group showed improvements in tracking time lag (CB time
lag day 1 206 msec, day 2 205 msec, day 3 196 msec, HC time lag 205 msec,
183 msec, 176 msec; Figure 7). Further no Sequence by Block interaction
was found demonstrating that there was little difference in the time lag of
tracking for the repeated as compared to the random segments (p=0.172).
This was likely due to the fact that the CB groups’ random segment tracking
time lag (223 msec) was not very different from their repeated sequence time
lags.
HC versus BG Group - Time Lag
In contrast, both the BG and HC groups were able to decrease their
time lag with practice (Main Effect of Block F(2,32) = 9.94, p=0.002; Figure
20). In addition, by the end of acquisition, time lag for the repeated segment
was reliably shorter than for the CT Rand2 segment for both groups as shown
by a Sequence by Block interaction (F(2.32) = 9.94, p=0.001. These two
groups changed similarly across practice as demonstrated by the lack of a
Group by Sequence by Block interaction (p=0.303).
HC versus SMC Group - Time Lag
Across acquisition the SMC group had the largest time lags (Figure 20).
They showed a similar pattern of reducing tracking time lag across practice to
that demonstrated by the HC group (Main Effect of Block F(2,32) = 5.93,
p=0.006). There was no Sequence by Block interaction demonstrating that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
135
the differences between the time lag of tracking for the repeated and Rand2
segments was not reliably different (p=0.191).
HC versus CB Group - Tracking Accuracy
Tracking accuracy improved across practice for all groups (Figure 8).
0.50
H C
0.60
C B
B G
0.70
S M C
3
O
o
<
0.80 O )
c
o
C O
1 —
0.90
1.00
1 2 3
Days
Figure 8. Tracking accuracy across practice. Error bars are SEM. All groups
significantly improved their tracking accuracy relative to random.
For the CB and HC groups this was evident by a Main Effect of Block (F(2,24)
= 5.25, p=0.013). For these two groups these improvements were larger for
the repeated as compared to the Rand2 segments (Sequence by Block
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
136
interaction F(2,24) = 5.25, p=0.035). The two groups’ tracking accuracy
improved similarly across practice (Group by Sequence by Block interaction
p=0.798).
HC versus BG Group - Tracking Accuracy
The BG and HC groups’ tracking accuracy also was aided by practice
(Main Effect of Block F(2,32) = 3.818, p=0.033), and more accurate
performance was demonstrated for the repeating segment as compared to the
Rand2 (Sequence by Block interaction F(2,32) = 5.111, p=0.038). There was
not a between group difference across practice (Group by Sequence by Block
interaction p=0.955).
HC versus SMC Group - Tracking Accuracy
The same pattern was true for the SMC group. This group
demonstrated the least accurate tracking performance (Figure 8), however,
they were helped by practice, and a Main Effect of Block was found (F(2,32) =
4.14, p=0.020). However, no significant Sequence by Block interaction was
found (p=0.188). This was largely due to the less accurate tracking
performance of the SMC group compared to the HC group (day three
repeating segment SMC 0.71; HC 0.92).
Retention Test - Time Series Analysis
As time lag was accounted for before tracking accuracy was calculated
(see methods for this procedure) a cross plot demonstrating time lag by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
137
accuracy (R2) was generated for the retention test data to illustrate this
interrelationship (Figures 9 and 10). Figure 9 displays the time lag versus
accuracy of tracking by group for CT Rand2, while figure 10 shows the time
lag versus tracking accuracy by group at retention test.
300
o
< D
i/i
£
245
< D
£
i- 190
D )
(0
_ J
c
C O
(1 )
S
135
0.40 0.52 0.64 0.76 0.88 1.00
Mean R2
Figure 9: Random sequence tracking accuracy and time lag. Error bars are
the SEM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
138
o
d >
( / >
£
E
i-
O )
c a
_i
c l
a
< u
300
245
190
135
80
■
- • - i ------------- 1
▼ HC
• CB
■ BG
♦ SMC
0.40 0.52 0.64 0.76 0.88 1.00
Mean R2
Figure 10. Time lag and tracking accuracy for the repeated segment at
retention test. Error bars are SEM.
All three focal stroke groups had larger time lags than the HC group, yet
only the SMC group appeared to be less accurate. These observations were
confirmed by statistical tests. Comparison of time lag showed a reliable
between group difference (Main Effect of Group F(7,28) = 6.72, p=0.001).
Post-hoc tests revealed that this finding was due to significantly slower
tracking time lags for the three focal stroke groups compared to the HC group
(CB p=0.025, BG p=0.005, SMC p=0.001; Figure 23). Significant between
group differences for tracking accuracy (Main Effect of Group F(7,28) = 3.29,
p=0.035) were identified by post-hoc tests to be the result of poorer tracking
accuracy for the SMC group compared to the HC group (p=0.005; Figure 10).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1. Explicit Knowledge of the No-EK Groups.
SRT Task CT Task
N
Subject
% Noticed
Recognition
% Correct % Sure
Recall
% Correct
Subject
% Noticed
Seq.
Repeat
% Correct
Recognition
% Correct % Sure
HC 5 80
(4/5)
66.0 66.0 40.0 0
(0/5)
20.0 66.0 71.0
CB 3 66
(2/3)
44.0 57.0 11.0 0
(0/3)
0.0 44.0 70.0
BG 5 0
(0/5)
46.0 64.0 6.0 0
(0/5)
0.0 46.0 58.0
SMC 5 20
(1/5)
53.0 80.0 33.0 0
___ IQ/5) .....
40.0 46.0 70.0
Note: All explicit testing for the No-EK groups was performed at the end of day 4. Subjective percent noticed denotes the percentage of
subjects in each group that expressed having noted a sequence or some form of repetition in their responses. Recognition and recall
percent correct were calculated in the following manner. For each memory type (recognition, recall) three sequences were shown to the
subject, two foils and one true. Therefore for each test, a total score of three was possible. Total number of correct responses was then
divided by the number of total trials which yielded a percent correct score. Segment repeated provides percentage of subjects in each
group who after being informed that some element of the tracking pattern repeated, were able to correctly identify (guess) the correct
third. The confidence score on average indicates how sure (scale of 0 to 100) subjects were of their recognition responses. For details
of the subjective, recognition, and recall tests see Chapter 4.
co
C D
140
Explicit Knowledge
Explicit testing found that not one of the 18 subjects in the No-EK
groups subjectively noticed the CT repeated sequence (Table 1). When
informed that there had been a repeating sequence and forced to choose if it
was located in the beginning, middle, or end of the tracking pattern no subjects
for either of the four groups were able to do so at higher than chance levels; in
two groups (CB and BG) no one correctly identified the middle as the repeated
third. In the SMC group 40% (2/5) guessed the middle third, and in the HC
group only 20% (1/5) guessed correctly. Recognition memory was the best for
the HC group (66% correct), while the BG and SMC groups both
demonstrated recognition memory of 46% correct. The CB group showed
nearly the same recognition memory pattern, explicitly identifying 44% of the
tested sequences correctly as either having or not having been seen
previously.
Discussion
Implicit learning of regularities within sequences has been
demonstrated across practice for healthy non-neurologically impaired
individuals (for examples see Nissan & Bullemer, 1987; Wulf & Schmidt, 1997;
Reber & Squire, 1998) and individuals with neurologic damage in the medial
temporal lobe that is associated with explicit memory (for examples see
Nissan & Bullemer, 1987; Glisky & Schacter, 1987; Reber & Squire, 1998).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
141
However, very little work has focused on implicit learning capability following
unilateral focal damage to the neural regions supporting the implicit memory
system (for exceptions see Gomez-Beldarrain et al., 1998; Vakil et al, 2000;
Boyd & Winstein, 2001). It is important to begin by noting that in this
dissertation, preserved implicit motor-sequence learning was demonstrated, in
each of the three unilateral focal stroke groups. Additionally, similar degrees
of implicit learning were noted for both the SRT and CT tasks. These findings
will be discussed with respect to the three initial hypotheses that formed the
basis of this study.
The first hypothesis proposed that following unilateral cerebellar stroke
implicit learning would be preserved as subjects practiced using their less
involved upper extremity and (by extrapolation) the less involved cerebellar
hemisphere. Additionally, we hypothesized that despite this finding of implicit
learning capability following cerebellar stroke, the magnitude of these
individuals’ implicit motor-sequence learning would be less than that seen for
healthy controls. Our data only partially support this hypothesis; following
unilateral cerebellar stroke, implicit motor-sequence learning was noted and
did not differ from that seen for healthy control subjects. Individuals in the CB
No-EK group did demonstrate the same degree of implicit learning of both
tasks as evidenced by decreased RT and tracking error for the embedded
sequence. This result is not surprising considering the previous work of
Gomez-Beldarrain et al. (1998) who showed that individuals with unilateral CB
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
142
stroke implicitly learned (using the SRT task paradigm) when using the less
involved cerebellar hemisphere (arm contralateral to side of brain damage).
Implicit learning, however, was not demonstrated when individuals used the
arm ipsilateral to brain damage (and by implication the damaged cerebellar
hemisphere). Gomez-Beldarrain et al. (1998) suggest that the cerebellum is
important for implicit motor-sequence learning, however, it appears that
cerebellar function is highly lateralized to the ipsilateral side. Our findings
would appear to confirm those of Gomez-Beldarrain et al. (1998), as we found
that when the less involved arm was used following cerebellar stroke implicit
motor-sequence learning was demonstrated for both tasks.
The second portion of our hypothesis stated that following unilateral
cerebellar stroke implicit motor-sequence learning would be diminished
relative to age matched healthy controls. This hypothesis was not supported
by our first two dependent measures - RT and tracking error. The CB group
did not differ from the HC group for either RT or RMSE learning scores.
However, in our time series analysis we identified tracking performance lag
time across acquisition and at the retention test, and identified significant
deficit in the CB groups’ ability to use experience to reduce their lag time.
Across practice time lag of tracking for the repeating segment did not change
and was not different from that of the random segment condition (Figures 7
and 9). At retention Time lag of tracking CB group was significantly longer
than that seen for the HC group (Figure 10); interestingly, this occurred
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
143
despite the fact that tracking accuracy was essentially the same for the CB
and HC groups. Despite the CB group’s ability to track accurately, they were
unable to track the target closely. In fact, evaluation of their time lag values
(range 206 to 196 msec across acquisition and retention) suggests that they
may have been relying on a visually guided tracking mode. It is tempting to
conclude from this finding that the cerebellum is primarily involved in timing of
responses while tracking accuracy in controlled elsewhere. This is not a new
conceptualization of cerebellar function for motor control (for examples see
Jueptner, Rijntjes, Weiller, Faiss, Timmann, Mueller, & Diener, 1995; Ivry,
Keele, & Diener, 1988; Ivry & Keele, 1989), however, it is a novel concept of
cerebellar function during implicit motor-sequence learning.
Thus, the time series analysis of the CT task performance revealed an
anticipatory timing deficit for the CB group that has not been previously
described for implicit motor-sequence learning. Indeed, from this finding it
appears that contrary to Gomez-Beldarrain et al.’s (1998) conclusions, the
implicit motor-sequence learning system is compromised even when invoking
the less involved cerebellar hemisphere. Another possibility is that this deficit
in implicit learning was only revealed with the more CT task and that for the
SRT task, ipsilateral cerebellar function is sufficient for implicit motor-
sequence learning. This interaction will be explored in detail in Chapter 7
when the performance data from the two tasks are directly compared.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
144
The next hypothesis proposed that following unilateral basal ganglia
damage implicit motor-sequence learning would be diminished relative to that
seen for healthy controls. The BG group did demonstrate implicit motor-
sequence learning for both tasks, however, they were significantly slower in
responding at the retention test (absolute RTs for the SRT task, Figure 2)
compared to the HC group. This finding, however, did not extend to the CT
task - there was not a reliable difference in tracking error between the BG and
HC groups. Learning scores for both the SRT and CT tasks indicated that the
BG group was, by the retention test, able to decrease both RT and tracking
error. From these data it seems that contrary to our hypothesis individuals
with unilateral basal ganglia damage are able to demonstrate implicit motor-
sequence learning and that their capability to change performance with
practice (and maintain changes at retention) was similar to that seen for the
HC group.
These findings of preserved capability for implicit motor-sequence
learning following unilateral basal ganglia stroke are contrary to those of Vakil
et al. (2000) who used the SRT task in a similar investigation. One
explanation for the difference in our findings and Vakil et al.’s (2000) was their
mixing the use of impaired and intact hemispheres (all subjects used their right
hand for responses, yet not all had right sided brain damage). A second
possible reason for the differences between our and Vakil et al.’s findings is
the amount of practice that was provided during acquisition. In our study, all
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
145
individuals practiced the repeated sequence 150 times (15 blocks), however,
only 50 trials of the repeated sequence were practiced in Vakil et al.’s (2000)
study. Unfortunately, Vakil et al. do not provide mean data and no direct
comparison of change in RT can be made with our data at the conclusion of
50 trials. However, it is a distinct possibility that with more practice, Vakil et al.
(2000) might have demonstrated implicit motor-sequence learning similar to
that seen in this dissertation.
It is likely that the basal ganglia function to coordinate the individual
elements of the sequence into a working group, allowing faster and more
accurate responses with practice (Berridge & Whishaw, 1992; Matsumoto et
al., 1999). This supposition is supported by findings from our time series
analysis. Individuals with basal ganglia stroke were found to have significantly
larger lag times at the retention test than those in the HC group -
demonstrating a deficiency in the capability to predict the sequence based on
prior experience. Once adjusted for the time lag, however, tracking accuracy
did not differ from that of the HC group. Prior work by Jennings (1995)
showed that individuals with Parkinson’s Disease were unable to use
information to predict and prepare responses in advance. It may be that
incorporation of information (predictive or other) into a motor plan requires
shifting and alteration in upcoming responses. Individuals with Parkinson’s
Disease have been shown to have great difficulty in altering their motor plan,
demonstrating a set-shifting deficit, and fail to benefit from conditions of varied
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
146
practice (Onla-Or, 2001). It may be that the sensitivity of the time series
analysis allowed us to demonstrate a similar deficit, although on a smaller
scale, in individuals with unilateral focal basal ganglia stroke.
An alternate interpretation of the finding of larger time lag following
basal ganglia stroke might be that an effector deficit exists, leading to reduced
capability to produce tracking movements. However, this interpretation is
weakened by the fact that subjects used the arm that was ipsilateral to basal
ganglia brain damage. The fact that implicit motor-sequence learning was
decreased as compared to the HC group and that greater lag times were
recorded for the BG group indicates that bilateral basal ganglia function must
be critical for implicit motor-sequence learning. Imaging data confirm that the
basal ganglia are active bilaterally (putamen) and relatively early (caudate)
during unimanual implicit motor-sequence learning (Toni et al., 1998). From
these data and the extant literature it appears that there is support for the
notion that the basal ganglia function during implicit motor-sequence learning
to coordinate the individual elements of an implicit motor-sequence. When
damaged, deficits in using prior experience to predict and prepare responses
in advance, thereby prolonged lag times for tracking a repeated pattern, are
apparent.
The last hypothesis addressed in this chapter proposed that following
unilateral stroke involving the sensorimotor cortical areas, implicit motor-
sequence learning would be decreased relative to individuals without
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
147
neurologic damage. For both tasks the SMC group demonstrated implicit
motor-sequence learning. However, at retention the SMC group demonstrated
significantly slower RTs and larger tracking errors compared to the HC group.
However, the SMC group also showed the greatest improvement, evident in
their large learning scores for both tasks at retention. It has been
demonstrated that Ml is highly active, reorganizing rapidly during the early
phases of motor learning (Kami et al., 1989; Muellbacher et al., 2001).
Disruption of this process likely lead to very poor initial responses, however,
with practice significant improvements were made. Likely, this pronounced
practice effect related to the use of the arm ipsilateral to brain damage;
function of one sensorimotor cortical hemisphere was preserved and implicit
motor-sequence learning occurred.
Time series analysis showed that despite the fact that the SMC group
decreased its time lag with practice, they had a significantly longer lag time
compared to that for the HC group at retention. In fact, the SMC group had
the longest lag times throughout acquisition and at retention (Figure 7 and 10).
The SMC group also was significantly less accurate compared to the HC
group at retention (R2 0.77 for SMC, 0.94 for HC). It has been previously
demonstrated that ipsilateral Ml is highly involved in the regulation of timing
during the production of simple sequences (Chen et al., 1997a). In fact, Chen
et al., (1997a) also noted that if Ml’s function is disrupted (via rTMS in this
case) that even more timing errors are made, particularly for complex
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
148
sequences (compared to simple ones). These results are consistent with and
extend those of Chen et al. (1997a), as large timing errors were evident in the
slowness of the RT for the SRT task and the pronounced time lag for the CT
task in the SMC group. The timing errors made by the SMC group may be
considered distinct from those made by the BG and CB groups in one
important way. They were accompanied by significantly less accurate tracking
movements. It may be that the ability of individuals in the SMC group to use
experience to predict the path of the target and reduce their time lag was
hampered by an impaired ability to produce accurate movements. In this
scenario, subjects are producing inaccurate or incorrect responses and
therefore, are not forming either a spatial or temporal representation for the CT
task. This combined deficit in spatial accuracy for, and temporal proximity to,
the target suggests that both of these functions are regulated by the
sensorimotor cortical areas. Therefore, it appears very likely that during
implicit motor-sequence learning the sensorimotor cortical areas function
bilaterally to coordinate the early encoding of the motor plan as well as
ipsilaterally to regulate the timing of sequence production.
Finally, the degree of explicit knowledge gained during sequence
practice should be considered. Observation of the explicit testing data (Table
8) showed that for both tasks the HC group demonstrated more explicit
knowledge compared to the focal stroke groups. For the simple SRT task,
only the HC and CB groups demonstrated subjective awareness of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
149
sequence, and a very low percentage of those in the focal stroke groups
demonstrated recognition; even fewer demonstrated any recall. In the
complex CT task, none of the subjects in either of the four groups subjectively
noticed the sequence, and only the HC group demonstrated recognition
memory that was only slightly above a chance level.
It is tempting to conclude that individuals with focal stroke did not gain
much explicit knowledge of the task because they had not yet learned the
sequence as well as the HC group. Evaluation of the data reveal that despite
demonstrating implicit motor-sequence learning, none of the focal stroke
groups was able to reduce either their repeating sequence RTs or repeating
segment tracking error to absolute values that were comparable to those of
the HC group (see Figures 2 and 5). It has been proposed that explicit
knowledge and implicit learning develop in parallel (Willingham & Gooedart-
Eschmann, 1999); explicit knowledge only becomes accessible as learning
becomes robust (Nissan & Bullemer, 1987; Brooks et al., 1995; Gentile, 1998)
and cognitive resources are freed (Doyon, 1997a). It is not surprising that less
explicit knowledge was gained for the more complex CT task than for the
simpler SRT task. Previous work has shown that for complex tracking tasks
(Pew, 1974; Wulf & Schmidt; 1997) explicit knowledge is rarely acquired even
for the non-neurologically impaired learner after extensive practice. Therefore,
it may be that our finding of low explicit knowledge for the CT task at the
conclusion of day four relates to the fact that implicit motor-sequence learning
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
150
was not yet robust enough to permit the awareness and development of
explicit knowledge. A second factor also must be considered. The continuous
nature of the CT task likely made it difficult for subjects to express at a verbal
level the regularities of the practiced sequence. It is tempting to consider
whether it is even possible for explicit knowledge to be formed and / or
expressed for tasks that do not easily lend themselves to declarative (fact
based, verbalizable) descriptions. A thorough discussion of the interactions
between implicit motor-sequence learning and explicit knowledge will be
considered in the next chapter.
Summary
The results of this chapter permit us to draw several conclusions. First,
all of the groups demonstrated the capability for implicit motor-sequence
learning. For the BG and SMC groups this finding is novel and has not been
previously described. Interestingly, for the CB group a timing deficit was
detected even after implicit motor-sequence learning had occurred; this may
reflect a fundamental role for the cerebellum during the production of implicit
sequences and leads to the supposition that the cerebellum is critically
involved in timing regulation during motor-sequence learning. The BG group
demonstrated poor predictive capability - a finding that is consistent with
previous work and leads to the supposition that the basal ganglia may be
important for updating and augmenting the motor plan based on prior
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
151
experience. The SMC group showed the largest amount of error and yet the
greatest degree of change across practice. It appears that the ipsilateral
sensorimotor cortical areas are initially impaired, yet, remain very capable of
benefiting from practice. The combination of large time lag and poor tracking
accuracy shown by the SMC group, however, illustrate that bilateral
sensorimotor cortex function is necessary for the production of unimanual
sequential movements.
As none of the groups showed large degrees of explicit knowledge at
the end of day four, the declarative memory system likely did not aid in the
formation of implicit motor-sequence learning. From these data two important
questions arise. First, how might the provision of explicit knowledge during
practice affect the implicit motor-sequence learning capability of these groups?
And second, as we have already seen differential results for the SRT and CT
tasks, will the impact of explicit knowledge be similar for the two tasks? These
questions form the basis for the next two chapters. Chapter 6 will address the
impact of explicit knowledge on implicit motor-sequence learning following
focal stroke and Chapter 7 will consider the interactions between stroke
location, explicit knowledge, and task.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
152
CHAPTER 6
The Effect of Prior Explicit Knowledge on
Implicit Motor-Sequence Learning following Focal Brain Damage
Overview
The manner and mechanism for the interaction between explicit
knowledge and implicit motor-sequence learning has received some attention
in the scientific literature (for examples see Reber, 1976; Green & Flowers,
1991; Curran & Keele, 1993; Reber & Squire, 1998; Boyd & Winstein, 2001;
Shea, et al., 2001). To date, there is little consensus regarding the impact of
explicit knowledge on implicit motor-sequence learning. Some have reported
a benefit of explicit knowledge (Curran & Keele, 1993; Boyd & Winstein,
2001), while others have demonstrated an interference effect (Reber, 1976;
Green & Flowers, 1991; Shea et al., 2001). These disparate findings likely
result from the combined divergent influences of task (i.e. motor demand),
type and timing of explicit knowledge, and the salience of the information
provided to the learner. Further, very little of this work has been extended to
include analyses of particular neural regions and their contributions to the
interaction between explicit knowledge and implicit learning. To date very few
(if any) studies have sought to relate damage in specific neural regions to an
altered (beneficial or detrimental) interaction between explicit knowledge and
implicit motor-sequence learning. The second specific aim of this dissertation
sought to assess the effect of prior explicit knowledge on implicit motor-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
153
sequence learning following focal brain damage. This goal formed the basis
for Chapter 6. The data and analyses presented in this chapter were focused
on addressing three hypotheses regarding the interaction between explicit
knowledge and implicit motor-sequence learning. These were:
Implicit motor-sequence learning following unilateral cerebellar damage will
not be affected by the nature of explicit knowledge. (Hypothesis #2)
During implicit motor-sequence learning, the basal ganglia function to
coordinate serial responses. Therefore, unilateral damage to the basal
ganglia will result in absent implicit learning when subjects are without and
when they have explicit knowledge of the sequence. (Hypothesis #4)
Unilateral damage to the sensorimotor cortical areas will cause a decrease in
implicit motor-sequence learning when no explicit knowledge is provided.
Implicit motor-sequence learning ability will increase as greater amounts of
explicit knowledge are provided. (Hypothesis #6)
Also addressed in this chapter was the capacity for subjects to
generalize implicit learning via transfer tests where the practiced sequences
were re-ordered (SRT) and re-scaled (CT; separately in the spatial and
temporal dimension). Transfer of implicit learning would indicate that the
knowledge acquired with practice was stored as an abstract motor plan, which
is flexible and capable of being re-scaled to new, yet similar, sequences.
Presently, it is unclear how transfer function is impacted by explicit knowledge.
As transfer is one measure of the robustness of motor learning, it is tempting
to conclude that if explicit knowledge aids implicit motor-sequence learning,
then it will also benefit the transfer of that learning. Accordingly, it has been
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
154
proposed that one function of explicit knowedlge is to promote the
generalizability of the implicit motor plan (Segar 1994; Curran, 1989). We
addressed this possibility by testing the hypothesis that:
The information acquired during implicit motor-sequence learning is stored as
a fundamental, abstract dynamic motor plan and not as a static
representation of the practiced movements. Therefore, if implicit learning
occurs, the fundamental nature of the learned pattern (i.e. motor program)
will facilitate performance during novel scaling of similar sequences (i.e.
transfer). (Hypothesis #8)
First in this chapter an overview of the methods including subject
characteristics and the specific statistical analyses used will be presented.
Next results will be presented. As the focus of this chapter is on the impact of
explicit knowledge within each group (HC, CB, BG, SMC), the results section
was organized by subject group. Finally, a discussion of the findings will be
integrated with our current understanding of the interactions between implicit
and explicit learning systems with respect to the specific neural contributions
of the cerebellum, basal ganglia, and sensorimotor cortical areas.
Methods
Two groups of subjects (No-EK n=18 and EK n=19; 37 total) were
studied for this chapter. Data from both the SRT and CT tasks were included
and discussed by subject group. Findings from these two tasks were
integrated in the discussion section of this chapter, however, the impact of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
155
task was not directly compared as this level of analysis was undertaken in
Chapter 7.
Subjects
To evaluate the relationship between implicit motor-sequence learning
and explicit knowledge, each group of subjects (HC, CB, BG, and SMC) were
evenly divided such that one-half of all participants was not provided with any
explicit knowledge (No-EK), and therefore predominantly learned the tasks
implicitly. The second half were provided with incrementally larger amounts of
explicit knowledge over the course of task practice (EK; details are provided
below). Some data from the No-EK group were already considered in Chapter
5, where it was demonstrated that all No-EK subjects, to varying degrees were
capable of implicit motor-sequence learning in both the SRT and CT task
paradigms.
Thus, in this chapter data from an additional 19 subjects (EK group
HC=5; CB=4; BG=5; SMC=5) will be presented and integrated with the data
from the 18 No-EK subjects previously described in Chapter 5 (HC=5; CB=3;
BG=5; SMC=5). A thorough description of the subjects was presented in
Chapter 4 and will not be repeated here.
Experimental Manipulation of Explicit Knowledge
Subjects in the EK group were progressively given explicit knowledge
for each task over the three days of practice. On day one, no information was
provided, subjects were instructed to “Respond as quickly as possible” to the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
156
SRT task and to "Track as accurately as possible” for the CT task. Explicit
knowledge was tested at the conclusion of day one of practice for all EK
subjects. At the beginning of day two, subjects were instructed that there was
some sequential and repeating aspect to both tasks. They were reminded of a
repetition in some responses / trials prior to beginning each task. Again, on
day two explicit knowledge was tested at the end of practice. At the beginning
of day three, subjects were told that they had been practicing a sequence;
before practicing either task, the location and composition of the sequence
was explained verbally. Then for each task a representation of the sequence
was provided (see Chapter 4, Figure 2), and subjects were asked to study,
without physically practicing, the information. A pre-test of explicit knowledge
was administered when subjects indicated that they were ready (typically
subjects took 5-10 minutes to study the information). Following practice on
day three, explicit knowledge was re-assessed.
In contrast, subjects in the No-EK groups were not given any indication
of the presence of repetition in their responses, and if they verbalized noticing
or asked about a sequence the investigator remained neutral - no direct
answer or response was provided. Explicit knowledge was tested only at the
conclusion of day four, after the retention and transfer tests were completed
for both tasks.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
157
Dependent Measures
SRT Task Dependent Measures
As previously described in Chapter 4, the dependent measure for the
SRT task was RT. Median RT was calculated by block for practice, retention,
and transfer. To control for large differences in absolute response time across
groups, RT data were converted to a practice change score (median practice
RT by block minus Rand2 RT). The SRT Rand2 value represented median
RT for a block of random sequence practice at the end of day one, a time
when the effects of non-specific learning had been minimized. A learning
score (median retention test RT minus Rand2 RT), and a transfer score
(median transfer RT minus Rand2 RT) were also calculated.
CT Task Dependent Measures
Three outcome measures were used to index performance on the CT
task, tracking error (RMSE), time lag of tracking (msec), and tracking accuracy
(r). The primary outcome variable for the CT task was RMSE, which indexed
tracking error. Identical to the SRT task, mean RMSE was calculated by block
for practice, retention, and transfer. These data were then converted to a
change score by taking the difference between mean RMSE for the repeated
third of the sequence and subtracting Rand2. Therefore, we calculated
change scores by block (practice), and a learning (retention), and transfer
(transfer) score.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
158
Time lag gave a measure of how close in time subjects’ responses
were to the target cursor. This measure may be used to evaluate how well
subjects’ were able to predict the path of the target cursor. Theoretically,
subjects’ might know the sequence well enough to anticipate the target’s path
and be ahead of it in time (yielding a negative time lag), however, this never
occurred. Mean time lag was calculated for the repeating third of the CT task
for the last block of each day of practice and the retention test. It was also
determined for Rand2.
Tracking accuracy represented the relationship between the target
waveform and subject’s responses after time lag was corrected. It was
calculated as the correlation coefficient (r) between the target and subjects’
waveforms, and due to our time factor correction was unbiased by the time lag
of subjects’ responses. A mean correlation coefficient was calculated for the
last block of practice on each day, and for retention and transfer tests. These
values were then converted to Fisher Z-scores for all statistical analyses. To
provide a basis for comparison, this procedure was repeated for Rand2.
Statistical Methods
Each of the four outcome measures (median RT, mean RMSE, mean
time lag, and mean Fisher Z-scores) were similarly evaluated for reliable within
and between group differences. Data from practice and retention were
assessed separately for all four of the dependent measures, and transfer data
was examined for median RT and mean RMSE.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
159
Acquisition Performance
As each of our hypotheses concerned changes within each group (HC,
CB, BG, and SMC) with and without explicit knowledge we determined the
impact of this factor within each group across acquisition. Each group’s
performance was examined separately for SRT task acquisition performance
by one factor Knowledge (EK, No-EK) ANOVA with a repeated measures
correction for block (day 1, day 2, day 3) with the RT change score as the
dependent measure. This same procedure was repeated three times, with
tracking error, time lag, and tracking accuracy (Fisher Z-score) change scores
separately used as dependent measures.
Retention and Transfer Tests
Performance on the retention test was used to index implicit motor-
sequence learning. For the SRT task, each group was evaluated using a one
factor Knowledge (EK, No-EK) ANOVA to determine the reliability of
differences using the RT learning score from the retention test as the
dependent measure. This was repeated for the CT task separately using the
tracking error learning score, time lag learning score, and tracking accuracy (z-
score) learning score as the dependent measures.
Last, data from the transfer tests were evaluated. For the SRT task, a
Group (HC, CB, BG, SMC) by Knowledge (EK, No-EK) by Sequence
(repeated, Rand2) ANOVA determined the generalizability of implicit motor-
sequence learning with RT as the dependent measure. As there were two
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
transfer tests for the CT task, separate Group (HC, CB, BG, SMC) by
Knowledge (EK, No-EK) by Sequence (repeated, random) ANOVAs were
performed. The first CT task transfer ANOVA evaluated amplitude transfer and
the second temporal transfer. When between group statistical differences
were detected, a post-hoc Scheffe test was used to determine their locus. In
instances where there were trends for statistical significance, effect sizes were
calculated to assess the meaningfulness of any between group differences
(Thomas et al., 1991).
Results
Healthy Control Group - Effect of Explicit Knowledge
SRT Task
Average RTs for practice, retention, transfer for the HC groups are
displayed in Figure 1. Across the three days of practice both EK and No-EK
groups showed deceased RTs. Analysis of change scores across acquisition
demonstrated that this was a significant improvement, both EK and No-EK
were faster in responding to the repeated sequence at the end of practice
(Main Effect of Block F(2,16)=15.16, p=0.005). However, there was also a
Main Effect of Knowledge (F(1,8)=8.34, p=0.020) due to the larger changes in
RT for the HC EK compared to the HC No-EK group. Across practice,
individuals in the HC EK group were able to significantly decrease their RT
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
161
when compared to the HC No-EK (Knowledge by Block interaction
F(1,8)=5.44, p=0.048).
-50
I s
i - -100
•--150
0 )
c n
S-200
-250
-300
-350
Day 3 Ret Transfer Day 1 Day 2
Blocks
Figure 1. RT change for the HC EK and No-EK groups. Both significantly
decreased RT with practice (Main Effect of Block). However, larger decreases
in RT were found for the EK group (Knowledge by Block interaction). There
were no between group differences for retention or transfer tests. Error bars
are SEM.
Despite differences for acquisition, by the retention test on day four,
there was not a significant difference between the two HC knowledge groups
(No Main Effect of Knowledge p=0.440), yet both showed significant levels of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
162
implicit learning relative to SRT Rand2 (reflected as the zero line in Figure 1).
Effect size calculations, however, demonstrated a large meaningful difference
(ES=1.1) between the two knowledge groups. Evaluation of the transfer
scores for the HC groups, showed that despite demonstrating faster RTs for
the transfer sequence (as compared to SRT Rand2) this difference was not
statistically reliable (Sequence Effect p=0.124). In addition, the two HC
knowledge groups did not differ from one another (Knowledge Effect p=0.746).
The amount of explicit knowledge possessed by individuals in the No-
EK groups (HC, CB, BG, and SMC) was presented and discussed in Chapter
5, Table 1. At the conclusion of day one, 80% (4/5) of individuals in the HC
EK group subjectively stated having noticed some degree of repetition in their
responses for the SRT task (Table 1). Despite this, recognition and recall
were below chance (40% each). By the end of the second day of practice,
recognition had improved to 73%, however, recall remained below chance
(40%). Healthy control EK subjects demonstrated that they had a high degree
of explicit knowledge of the SRT task prior to (86% recognition, 73% recall)
and at the end of practice on day three (recognition 100%, 86% recall).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
163
Table 1. Explicit Knowledge of the EK groups.
D ay 1 SRT Task
SRT Task
N
Subjective
% Noticed
Recognition
% Correct Confidence
Recall
% Correct
HC
5 80 (4/5) 40.0 58.7 40.0
CB 4 50 (2/4) 58.3 85.4 33.3
BG
5 20 (1/5) 53.3 60.9 40.0
SMC
5 20 (1/5) 46.6 60.9 26.6
D ay 1 CT Task
CT Task
N
Subjective
% Noticed
Sequence
Repeat
% Correct
Recognition
% Correct Confidence
HC
5 0 (0/5) 20.0 33.3 65.9
CB 4 0 (0/5) 0.0 33.3 80.8
BG
5 0 (0/5) 20.0 40.0 55.5
SMC
5 0 (0/5) 20.0 53.3 64.9
Pay 2
SRT Task CT Task
Recognition
% Correct Confidence
Recall
% Correct
Recognition
% Correct Confidence
HC
73.0 74.8 40.0 46.6 65.9
CB 50.0 74.6 33.3 33.3 70.4
BG
53.3 83.3 33.3 26.6 70.7
SMC
53.3 55.9 46.6 53.3 83.3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
164
Table 1 (continued)
Pre-Test Day 3
SRT Task CT Task
Recognition
% Correct Confidence
Recall
% Correct
Recognition
% Correct Confidence
HC
86.6 88.6 73.0 60.0 74.9
CB 66.6 80.8 33.3 41.6 80.4
BG
86.6 90.9 46.6 46.6 70.3
SMC
80.0 98.9 46.6 73.0 70.3
Post-Test Day 3
SRT Task CT Task
Recognition
% Correct Confidence
Recall
% Correct
Recognition
% Correct Confidence
HC
100.0 100.0 86.0 73.3 66.3
CB 83.3 81.6 33.3 50.0 75.8
BG
86.6 79.9 46.6 40.0 71.6
SMC
73.3 79.9 53.3 66.6 71.6
Note: Explicit testing of the EK groups was performed at the end of each day of practice. In
addition, a pre-test of explicit knowledge was administered at the beginning of day 3.
Subjective awareness of the sequence was only assessed at the end of day 1. For details see
Chapter 4.
CT Task
For the CT task, both HC knowledge groups again improved their
performance with practice as shown by decreased RMSE (Main Effect of block
F(2,16)=6.55, p=0.008; Figure 2). However, there was not an effect of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
165
knowledge (Knowledge Effect p=0.206), suggesting little effect of the presence
or absence of explicit knowledge during the practice of the CT task.
5
1
3
C D
■ 7
11
-15
Ret TxA TxT Day 1 Day 2 Day 3
Blocks
Figure 2. Change in tracking error for HC groups. A significant Main Effect of
Block was found showing that both groups (EK and No-EK) benefited from
practice regardless of the presence of explicit knowledge. At retention and
transfer tests (TxA, TxT) there was no effect of explicit knowledge. Error bars
are SEM.
There was a trend for a difference between the EK and No-EK groups
at the retention test (Main Effect of Knowledge F(1,8)=3.58, p=0.090) and a
large effect size confirmed this trend (ES=1.4). On closer examination, this
trend was primarily due to an increase in the tracking error for the HC No-EK
group compared to that for the end of day three. In contrast, tracking error
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
166
slightly decreased for the HC EK group at retention compared to that for the
end of day three. Neither HC knowledge group showed a reliable change in
transfer score for amplitude (Sequence Effect p=0.080) or timing (Sequence
Effect p=0.106). In addition, explicit knowledge did not have an effect on
transfer, as the transfer scores from the two EK groups did not differ from one
another (Knowledge Effect amplitude p=0.258; timing p=0.785).
Across acquisition, both HC knowledge groups were able to decrease
their time lag (Main Effect of Block F(2,16)=3.61, p=0.050; Figure 3).
However, this reduction in tracking delay was larger for the HC EK group than
the HC No-EK group as evidenced by the Main Effect of Knowledge
(F(1,8)=13.74, p=0.006; Figure 24C). At retention there was a strong trend for
a significant difference between the groups in time lag (Knowledge Effect
p=0.082). Observation of the data however, demonstrated that the HC EK
group had shorter time lag of tracking at retention (EK 109 msec; No-EK 160
msec; Figure 3) and a large effect size (ES=0.73 ) confirmed that this was a
meaningful difference (Thomas et al., 1991). Lastly, tracking accuracy for the
two HC knowledge groups was compared. No effect of block (p=0.192) or
knowledge (p=0.954) was found for practice (Figure 3). Similarly there was
not a difference between these two knowledge groups’ tracking accuracy at
retention (Knowledge Effect p=0.893).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
167
2 2 5
E K d1
200 d2
4 g 150
Q
I 175
o
125 □ d3
0 re t
Q d2
N o -E K d1
re t
d3
100
0.65 0.72 0.79 0.86 0.93 1.00
R2
Figure 3. Time series analysis demonstrated that both HC groups decreased
their time lag (Main Effect of Block) across acquisition. The EK group was
able to decreased time lag of tracking more than the No-EK group. There was
a trend for differences in time lag for the two HC groups at the retention test.
A large effect size confirmed this effect of explicit knowledge for tracking time
lag (ES=0.73). Tracking accuracy did not differ between the groups across
acquisition or at retention. Error bars are SEM.
Explicit knowledge data from all of the No-EK groups for the CT task
was presented in Chapter 5 and summarized in table 8. It will not be repeated
here. Explicit testing for the CT task on day one, revealed that none of the HC
EK subjects subjectively noticed the existence of the practiced sequence.
Further, only 20% correctly guessed that the middle third was the repeated
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
168
element and recognition was only 33%. At the conclusion of day two,
recognition was at chance (46%) and confidence was high (66%). On the pre
test on day three, recognition was 60% and this improved over practice to 73%
(Table 1).
Cerebellar Group
SRT Task
The pattern of change for the cerebellar groups across SRT task
practice was similar to that seen for the HC groups (Figure 4). Both CB EK
and CB No-EK benefited from practice as shown by a Main Effect of Block
(F(2,10)=5.67, p=0.023). However, the CB EK group performed significantly
better over acquisition, changing RT more than the No-EK group (Main Effect
of Knowledge F(1,5)=28.23, p=0.007). By retention (and similar to that for the
HC groups), this advantage had disappeared, and there was not a between
knowledge group difference (Knowledge Effect p=0.370). Evaluation of the
transfer scores, showed that despite demonstrating faster RTs for the transfer
sequence (as compared to SRT Rand2) this difference was not statistically
reliable (Sequence Effect p=0.124). The groups also did not differ from one
another (Knowledge Effect p=0.383).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
169
EK
— No-EK
100
O w U
< D
| o
H -50
a:
a -100
-150
a >
C D
| -200
O
-250
-300
-350
Ret Transter
Day 1 Day 2 Day 3
Blocks
Figure 4. Across SRT task acquisition both CB EK and No-EK groups
benefited from practice (Main Effect of Block). However, the CB EK group
decreased their RTs to a larger extent than the No-EK group (Main Effect of
Knowledge). No reliable between group differences were noted for the
retention or transfer tests. Error bars are SEM.
Testing of explicit knowledge of the CB EK group for the SRT task at
the end of day one revealed that only 50% (2/4) subjectively reported noticing
the sequence (Table 9). Recognition was just above chance (58%) and recall
was only 33%. At the end of day two (when subjects knew there was a
sequence), recognition was still only 50% and recall had not changed (33%).
Prior to practice and following study of the sequence on day three, recognition
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
170
was 66% and recall was 33%. After implicit sequence practice, recognition
had improved further to 83%, however, recall remained at 33% (Table 1).
CT Task
Continuous tracking task practice allowed both CB knowledge groups to
decrease their tracking error (Main Effect of Block F(2,10)=14.70, p=0.001;
Figure 5). There was a large advantage for the CB EK group over the CB No-
EK group (Main Effect of Knowledge F(1,5)=37.63, p=0.002). Across the
three days of practice, the CB EK group demonstrated a greater reduction in
RMSE error than did the CB No-EK group (Knowledge by Block interaction
F(1,5)=7.00, p=0.046). Interestingly, and similar to that seen for the SRT task,
this advantage was not maintained at the retention test. The learning score
was not different for the two CB knowledge groups (Knowledge Effect
p=0.594). Neither CB knowledge group showed amplitude (Sequence Effect
p=0.080) or temporal (Sequence Effect p=0.106) transfer. Similarly, the two
CB knowledge groups were not different from one another for either transfer
test (Knowledge Effect amplitude p=0.951; temporal p=0.955).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
171
L U
a:
c
< D
D )
C
c d
JO
O
5 EK
No-EK
1
-3
7
-11
-15
Day 2
Ret TxA TxT Day 1 Day 3
Blocks
Figure 5. Both cerebellar groups decreased tracking error across acquisition
(Main Effect of Block). The large difference between the groups evident in the
magnitude of this change was confirmed by a Knowledge by Block interaction,
which showed an advantage of explicit knowledge. However, this between
group difference was not maintained at retention. No group differences were
seen for transfer (TxA, TxT). Error bars are SEM.
There was no effect of practice on time lag for individuals with
cerebellar stroke as evidenced by the relatively horizontal trajectories in Figure
6 (Block Effect p=0.847). There was also no effect of explicit knowledge
(Knowledge Effect p=0.715) on time lag. By the end of acquisition on day
three, the average lag time for the two CB knowledge groups was
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
172
approximately 175 msec. This inability to anticipate the tracking pattern
persisted to the retention test where no effect of knowledge was found
(p=0.776). Indeed the average lag time at retention was approximately 165
msec.
225
200
I 175
o
O
0 >
| 150
125
100
0.65 0.72 0.79 0.86 0.93 1.00
A
■
♦
A
O
□
O
E K d1
d2
d3
re t
N o -E K d1
d2
d3
re t
R2
Figure 6. There was not effect of practice or knowledge on the CB groups’
tracking time lag or tracking accuracy. Error bars are SEM.
The CB groups were also not able to take advantage of practice to
significantly improve tracking accuracy, however there was a trend for
improvement in the correlation (r) across acquisition that primarily occurred
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
173
between the first two days of practice (Main Effect of Block F(2,10)=3.15,
p=0.087; Figure 6 x-axis). This change in tracking accuracy was not affected
by the level of explicit knowledge (Knowledge Effect (p=0.467). These
findings were identical to those at retention using the retention score
(Knowledge Effect p=0.152).
Explicit knowledge of the CT task for the CB EK group is presented in
Table 1. For day one, none of these subjects noticed the pattern and none of
them correctly guessed that the middle third was the repeating element.
Recognition at the end of day one and two were only 33%. The pre-implicit
practice study period on day three did not stimulate much explicit knowledge
for recognition of the sequence; following explicit study recognition was only
41%. Implicit practice, however, improved recognition to 50%.
Basal Ganglia Group
SRT Task
Across practice both of the BG groups were able to change repeated
sequence RT (Main Effect of Block F(2,16)=21.27, p=0.000). Across practice
those in the No-EK group decreased their error more than those in the EK
group (Main Effect of Knowledge F(1,8)=52.21, p=0.001); this was most
prominent at the end of day three. Across blocks, the BG EK group performed
significantly worse (less decrease in RT) than did the No-EK group
(Knowledge by Block Interaction F(1,8)=7.81, p=0.023). This interference
effect of explicit knowledge on the ability of individuals with basal ganglia
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
174
stroke to decrease RT was not evident at retention test (Main Effect of
Knowledge p=0.776). There was no difference between the groups at the
transfer test (Knowledge Effect p=0.124; Figure 7).
a
< D
V I
E
'*W '
h * “
V L
< D
CD
c
0 3
s z
50
T
0
-50
-100
-150
-200
-250
-300
-350
Day 1 Day 2 Day 2
Blocks
- • - EK
No-EK
Ret Transfer
Figure 7. Both the BG EK and No-EK groups changed their RT across
acquisition (Main Effect of Block). The No-EK group demonstrated more
decrease in RT than the EK group (Group by Knowledge interaction). This
between group difference disappeared by retention. No differences were
evident for transfer. Error bars are SEM.
Few in the BG EK group subjectively noticed the SRT task sequence in
the first day of practice (20%; Table 9). Recognition was 58% and recall 40%
at this time. A second day of implicit practice and awareness of the existence
of a sequence did not change SRT task recognition or recall for the BG EK
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
175
group. However, the explicit study period on day three raised recognition to
86% and recall to 46%. Implicit motor-sequence practice on day three did not
change this level of recognition or recall memory for the sequence (Table 1).
CT Task
Across CT task practice, both EK and No-EK groups again decreased tracking
error (Main Effect of Block F(2,16)=21.27, p=0.000; Figure 8).
• EK
No-EK
LU
C O
5
v'T
0 1
c
< D
O )
c
C O
-10
-15
Day 1
Day 2 Day 3
Ret TxA TxT
Blocks
Figure 8. Across practice both BG EK and No-EK showed decreased tracking
error (Main Effect of Block). Again, larger changes were demonstrated by
those in the No-EK group relative to the EK (Knowledge by Block interaction).
This between group difference was temporary and not seen at retention.
Again, there were no differences between the groups for the transfer tests
(TxA, TxT). Error bars are SEM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
176
There was a significant negative effect of explicit knowledge for the CT task
(Main Effect of Knowledge F(1,8)=52.21, p=0.001). There was also a trend for
a Knowledge by Block interaction (p=0.110). This trend was confirmed by
large effect size differences between the two groups at block 10 on day two
(ES=1.18) and at block 15 on day three (ES=1.20). Assessment of retention
test data showed that the interference of explicit knowledge was temporary, as
a large improvement from day three in tracking error was seen for the BG EK
group at retention. In addition, there was no Main Effect of Knowledge at
retention (p=0.404; Figure 8). Again no transfer learning was shown by either
group for amplitude (Sequence Effect p=0.080) or timing (Sequence Effect
p=0.106) re-scaling of the practiced sequence.
Across practice both BG groups demonstrated the ability to decrease
their time lag on the CT task from 214 to 188 msec for the BG EK group and
from 214 to 172 msec for the BG No-EK group (Main Effect of Block
F(2,16)=10.09, p=0.001). This tracking efficiency was not affected by explicit
knowledge (Main Effect of Knowledge p=0.714; Figure 9). At the retention test
there was not a difference in time lag for the two BG knowledge groups (EK
time lag = 160, No-EK =144 msec; Knowledge Effect p=0.377). Tracking
accuracy ranged from 0.86 to 0.87 for the BG EK group and 0.79 to 0.84 for
the BG No-EK group across acquisition and did not improve significantly with
practice for either group (Block Effect p=0.263; Figure 9). There was also not
a difference between the two knowledge groups during acquisition
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
177
(Knowledge Effect p=0.418). The two groups demonstrated similar tracking
accuracy at the retention test (EK 0.93, No-EK 0.89; Knowledge Effect
p=0.377).
225
200
I 175
o
o
v
| 150
125
100
0.65 0.72 0.79 0.86 0.93 1.00
▲ E K d1
•
d2
■
d3
♦
re t
A N o-E K d1
O d2
□
d3
A
re t
R2
Figure 9. Both the BG EK and No-EK groups decreased tracking time lag with
practice (Main Effect of Block). There was no effect of explicit knowledge and
no between group differences at retention. Tacking accuracy did not differ
between the groups. Error bars are SEM.
In general across the three acquisition days, explicit knowledge of the
CT task by the BG EK group was much less than that for the SRT task (Table
1). None of these subjects noticed the sequence, and only 20% (1/5) guessed
which third repeated when prompted. Recognition of the sequence on day
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
178
one was below chance 40%, and remained so, despite a second day of
practice and awareness of the existence of the sequence (day two 26%), pre-
implicit practice explicit study on day three (46%), and following implicit motor-
sequence practice on day three (40%).
Sensorimotor Cortex Group
SRT Task
Individuals in both sensorimotor groups were able to change their RT
with practice (Main Effect of Block F(2,16)=10.84, p=0.002; Figure 10). The
provision of explicit knowledge, however, slowed RTs relative to the No-EK
condition (Main Effect of Knowledge F(1,8)=19.90, p=0.002). The detrimental
effect of explicit knowledge was most evident at the end of day three when the
EK group had full explicit knowledge and statistically reliable as demonstrated
by a Knowledge by Block Interaction (F(2,16)=4.82, p=0.014; Figure 10).
These differences were maintained at retention, where the SMC No-EK group
demonstrated significantly larger RT learning scores (No-EK -261 msec, EK -
138 msec; Main Effect of Knowledge F(1,8)=2.67, p=0.030). There was not a
reliable difference between the repeating and SRT Rand2 for the transfer test
(Sequence Effect p=0.124).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
179
100
o
< D A
< / > u
— -50
5 -100
« -150
j= -200
C O
6 -250
-300
-350
Day 3 Ret Transfer Day 1 Day 2
Blocks
Figure 10. RT for the repeated sequence decreased for both groups across
practice (Main Effect of Block). Providing explicit knowledge slowed RT
relative to No-EK (Knowledge by Block interaction). At retention, the SMC
No-EK group was significantly faster (Main Effect of Knowledge). There were
no between group differences for the transfer test. Error bars are SEM.
The SMC EK group demonstrated poor subjective knowledge of the
sequence at the end of day one (20%; Table 1). Recognition was at a chance
level (46%) and recall was below chance (40%). A second day of practice
with the knowledge that a sequence existed improved recognition (53%) and
recall (47%). The pre-implicit practice explicit study period substantially
improved sequence recognition (80%) but not recall (46%). Following
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
sequence practice on day three, recognition was well above chance (73%)
and recall just at chance level (53%).
CT Task
During practice of the CT task both SMC knowledge groups were able
to decrease their tracking error with practice (Main Effect of Block
F(2,16)=17.86, p=0.011; Figure 11).
♦ EK
5
No-EK
1
3
O )
c.
7
-11
-15
Ret TxA TxT Day 1 Day 2 Day 3
Blocks
Figure 11. Both SMC groups decreased tracking error (Main Effect of Block).
Explicit knowledge decreased the amount of change shown by the SMC EK
group (Main Effect of Knowledge). There were no between group differences
for the retention or transfer tests (TxA, TxT). Error bars are SEM.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
181
Explicit knowledge interfered with implicit motor-sequence learning as shown
by a nearly significant Main Effect of Knowledge (p=0.052). There was not a
significant Knowledge by Block interaction (p=0.408), likely as a result of the
large amount of variance demonstrated by both knowledge groups. The
knowledge effect seen during acquisition was transient, as evidenced by the
lack of a knowledge effect for the learning score at the retention test
(p=0.365). Despite this finding a large effect of knowledge was demonstrated
(ES=1.3). Neither group was able to generalize their CT task learning when
the sequence was re-scaled in either amplitude (Sequence Effect p=0.080) or
duration (Sequence Effect p=0.106).
Both SMC groups were able to take advantage of practice to
significantly decrease time lag during CT tracking (Main Effect of Block
F(2,16)=6.11, p=0.011; Figure 12). This tracking time lag decreased across
acquisition from 225 to 169 msec for the SMC EK group and from 220 to 163
msec for the SMC No-EK group. During practice there was not an effect of
explicit knowledge on time lag (Knowledge Effect p=0.305). Knowledge also
did not impact time lag of tracking at the retention test (EK 130 msec, No-EK
146 msec; Knowledge Effect p=0.486). For tracking accuracy (R2) during the
CT task, there was no practice (Block) effect (p=0.525) or knowledge effect
(p=0.244; Figure 12). No knowledge group differences were found for tracking
accuracy at the retention test (R2 EK 0.93, No-EK 0.82; Knowledge Effect,
p=0.348).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
182
0 3
< 1 >
Q
o
< D
(/)
2 2 5
200
175
150
125
100
0.65 0.72
_L
0.79 0.86
R2
A E K d1
• 62
■ 63
♦ re t
A N o -E K d1
O 62
□ 63
O re t
0.93 1.00
Figure 12. Both SMC groups decreased tracking time lag across acquisition
(Main Effect of Block). There was no effect of knowledge on this decrease.
There were no practice or knowledge effects on tracking accuracy across
acquisition or at retention. Error bars are SEM.
Explicit testing found that at the end of practice on day one, none of the
SMC EK group subjectively noticed the sequence and only 20% correctly
guessed that the repeating element was in the middle third (Table 1).
Recognition, however, was 53%. Practice on day two and knowledge that a
sequence existed did not change recognition (53%). The pre-implicit practice
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
183
study period raised recognition to 73%. After implicit sequence practice on
day three recognition dropped slightly to 66%.
Discussion
The primary purpose of this chapter was to evaluate the effect of
explicit knowledge on performance of the focal stroke groups (CB, BG, and
SMC) as they practiced two implicit motor-sequence tasks each with different
motor demands - one relatively discrete and one continuous. The
performance impact of informing subjects of either the existence of a
sequence, or of all of the regularities in the stimuli for movement during task
practice has been debated. Some of the disparity in the literature has
stemmed from the use of different experimental tasks with varying goals,
demands, and degrees of complexity. In addition, only one study (Boyd &
Winstein, 2001), utilized a pre-test to ensure the veracity of explicit knowledge.
This dissertation was constructed to control for the effects of the motor
demand of tasks by having the same group of subjects practice two separate
implicit tasks. Further, serial explicit tests enabled us to assess the degree of
declarative knowledge for each subject in the EK groups, and consider this
factor when evaluating implicit motor task performance and learning. Most
importantly, this dissertation was designed to allow a controlled examination of
the differential impact of explicit knowledge for implicit motor-sequence
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
184
learning on individuals with focal stroke in neural regions considered to be
critical in the procedural motor learning network.
Evaluation of the results from the HC groups will be considered first to
demonstrate how explicit knowledge and implicit motor-sequence learning
typically interact without the additional confound of neural damage. Explicit
knowledge affected performance differently for the two tasks. It was very
beneficial for the SRT task as demonstrated by the significantly larger
decreases in RT for the HC EK group than the No-EK group by the end of
practice, and corroborating previous findings by Curran and Keele (1993).
These beneficial effects of explicit knowledge differ from those of Reber and
Squire (1998) who did not find a benefit of prior explicit knowledge for a group
who memorized the sequence before practice. There are two logical
explanations for the disparity between our work and Reber and Squire’s work.
First, no pre-test was administered in Reber and Squire’s study, and it cannot
be assumed that the subjects had full explicit knowledge of the sequence. In
fact, it may be that subjects in Reber and Squire’s (1998) work did not have
explicit knowledge prior to physical sequence practice. In this dissertation, we
controlled for this factor by using a pre-test to demonstrate that the HC EK
subjects showed a high degree of explicit knowledge prior to practice at the
beginning of day three (86% recognition; 73% recall). A second significant
difference between these two studies centered on the difference in the timing
of the provision of explicit knowledge. In this dissertation, subjects in the EK
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
185
groups had an initial 50 blocks of sequence practice without the provision of
any explicit knowledge; another 50 blocks were practiced with partial explicit
knowledge (of the existence of a sequence). Finally, at the beginning of the
third day, full explicit knowledge was provided. Thus, subjects in the EK
groups all had considerable amounts of implicit SRT task practice before they
were encouraged to incorporate explicit knowledge into their performance. In
Reber and Squire’s work (1998), subjects tried to explicitly learn the sequence
before ever attempting the SRT task and their sequence performance was
evaluated immediately following the explicit study period (subjects watched the
sequence 10 times, then practiced it physically). Previously, it has been
hypothesized that explicit and implicit memories develop in parallel, but that
awareness of explicit knowledge does not occur until a certain degree of motor
success has been achieved (Brooks et al., 1995; Willingham & Goedert-
Eschmann, 1999). If this is true, then perhaps it is not just awareness of
explicit knowledge that occurs following some motor success, but even more
importantly that explicit knowledge may only be incorporated into and benefit
the motor plan after some degree of implicit task ability has been gained.
Interestingly, complex (CT) task acquisition performance was not
significantly affected (positively or negatively) by explicit knowledge. There
was a slight performance benefit for the HC EK group, but this was not large.
This finding is intriguing as those in the HC EK group did show (via explicit
knowledge testing) relatively large amounts of recognition of the sequence on
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
186
the pre-test before practice on day three (60%). Two conclusions may be
drawn from these findings. It is very likely that the critical features of complex
tasks (e.g. kinematic patterns, movement velocity and acceleration), such as
the CT, are not ones that are easily converted into explicit terms. In fact, when
the predictable elements of complex movement sequences are translated into
explicit terms they typically are represented as equations, such as the one
governing the repeated third of the CT task3, or as schematic representations
(see Chapter 4, Figure 2). Unfortunately this information cannot be easily
translated into movement and thus, may not aid implicit task performance.
Therefore, it appears that for explicit knowledge to benefit subjects as they
practice an implicit task, it must be salient to the learner. Second explicit
knowledge must contain some degree of practical information that will aid the
performance of the learner. For the CT task, it is likely that the only explicit
information that greatly assisted the HC EK group was knowledge of the
location of the repeating element; from this information subjects might have
been able to focus more directly on the middle third of the tracking pattern as
they practiced. This factor may explain the finding that explicit knowledge did
beneficially affect the time lag of tracking performance; lower time lag was
shown by the HC EK groups for practice and retention and this finding was
nearly significant (p=0.082). In fact, by the retention test, time lag for the HC
o
One form of explicit representation of the CT task sequence /(x) = b0 + a -i sin (x) + b, cos (x)
+ a2 sin (2x) + b2 cos (2x) +...+ a6 sin (6x) +b6 cos (6x)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
187
EK group was faster (109 msec) than typical reaction times recorded for
voluntary movements (150-190 msec; Ghez & Krahauer, 2000), and can
therefore be considered anticipatory or predicative. In contrast, tracking
accuracy did not change significantly across practice for either HC group. This
was due primarily to the very accurate performance of these groups (EK R2 =
0.95, No-EK = 0.92 after time lag was accounted for) and likely represented a
ceiling effect.
On the last day, when retention and transfer tests were administered,
subjects were not provided with any explicit knowledge reminders. Effect size
calculations demonstrated that there continued to be meaningful differences
between the knowledge groups at this time, however, this was not a
statistically reliable effect. The poor generalizability of implicit learning for the
SRT and CT tasks to the transfer tests suggests that learning of these tasks
was not yet at asymptote. Generalizability is commonly noted once learning is
robust and indicates that a task is represented as an abstract motor plan. It
may be that more practice would have benefited both SRT and CT task
performance and facilitated the formation of an abstract motor plan for these
sequences. Future work investigating transfer of learning for the SRT and CT
tasks (as well as other implicit motor-sequencing tasks) should endeavor to
evaluate the impact of increased practice on generalizability of implicit motor-
sequence learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
188
In many ways the CB EK group was similar in their acquisition
performance and retention learning to the HC EK group. Both SRT and CT
task performance for the CB EK group were benefited by practice, and more
importantly, by explicit knowledge. For both tasks, the CB EK group
demonstrated enhanced performance when influenced by both partial (day
two) and full (day three) explicit knowledge. Further, both CB knowledge
groups were able to show implicit learning of the SRT and CT tasks; explicit
knowledge had no effect on retention test data (the groups were not
statistically different) and transfer results mirrored those of the HC groups.
These findings directly contradict our hypothesis that implicit motor-sequence
learning in subjects with unilateral cerebellar damage would not be affected by
the nature of explicit knowledge.
It appears that the function of the cerebellum during implicit motor-
sequence learning is lateralized to the ipsilateral hemisphere. Gomez-
Beldarrain et al., (1998) documented that individuals with focal cerebellar
stroke were able to demonstrate implicit motor-sequence learning of the SRT
task only when they invoked the uninvolved hemisphere (contralateral to the
hand being used to practice the task). This is the same condition used in this
dissertation, and our data are consistent with those of Gomez-Beldarrain et al.
(1998). Further, our findings of improved practice performance for the CB EK
compared to the CB No-EK, group extend our understanding of cerebellar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
189
function during implicit motor-sequence learning to include the ability to benefit
from explicit knowledge.
Individuals in the CB EK group showed relatively good explicit
knowledge at the end of day two. Overall, their pre-test data on day three was
the poorest of the three focal stroke groups but they demonstrated the ability
to increase their explicit knowledge with implicit task practice on day three.
The CB EK group also greatly improved their performance on day three
(through faster RTs and fewer tracking errors). These parallel improvements
point to an intact capability to share implicit motor-sequence learning with the
explicit learning system, and suggest that the neural structures underpinning
such a transfer of information were not damaged by focal unilateral cerebellar
stroke; this may suggest that they reside elsewhere in the brain. Thus, even
following focal cerebellar stroke, explicit knowledge could be beneficially
incorporated into the implicit motor plan (as shown by the main effect of
knowledge) and explicit knowledge could be enhanced by implicit motor-
sequence practice. This finding points to two important conclusions. First,
transfer of information between the implicit and explicit systems most likely
occurs elsewhere (non-cerebellar) in the brain, and second, due to its
lateralized role in implicit motor-sequence learning even following unilateral
damage the cerebellum is able to take advantage of explicit knowledge of the
implicit sequence.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
190
This transfer of information between explicit and implicit systems may
take place through the prefrontal basal ganglia-thalamocortical circuit
(Selemon & Goldman-Rakic, 1985). This circuit is left intact by cerebellar
brain damage and is ideally situated to incorporate implicit motor information
from the basal ganglia with explicit knowledge which is likely mediated by the
DLPFC (see Chapter 2, Figures 3 and 4). This is likely the primary
neuroanatomic pathway for information transfer between the declarative and
procedural memory systems, however, a second neural network that may be
important has been described by Middleton and Strick (1994). These authors
identified a projection pathway from the dentate nucleus of the cerebellum and
globus pallidus of the basal ganglia to DLPFC via thalamus. It is likely that
under normal circumstances these two projections transfer information from
the implicit motor learning systems, residing in the cerebellum and basal
ganglia, to the frontal cortex where it may be translated into explicit knowledge
as the learner becomes aware of the regularities in responding. In this
experiment we only tested the contralateral or less involved cerebellar
hemisphere. However, an intriguing follow-up study might assess the impact
of explicit knowledge on implicit motor-sequence learning using the ipsilateral
cerebellum. Such an investigation might permit a behavioral analysis of the
role of the dentate-thalamic-DLPFC pathway during implicit motor-sequence
learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
191
Despite the fact that the more intact, contralateral cerebellar
hemisphere was tested, a profound impairment in tracking timing was
identified for both cerebellar knowledge groups (EK and No-EK). None of the
individuals with cerebellar damage demonstrated the ability to reliably
decrease time lag with practice. This occurred despite large improvements in
tracking accuracy for both groups (due to large between subject variability,
tracking accuracy changes did not reach statistical significance, see Figure 6).
These data support the long held conceptualization of the cerebellum as
somehow involved in the time aspect of movement (Ivry & Keele, 1989; Ivry et
al., 1988). They also demonstrate a failure to use either practice experience
or explicit knowledge to predict the target’s pathway. Cerebellar subjects,
therefore, may have been in a somewhat reactive, visually guided tracking
mode, unable to learn to anticipate the target’s path.
In strong contrast to the performance of the HC and CB subjects,
individuals with basal ganglia stroke demonstrated an inability to take
advantage of explicit knowledge for either the SRT or CT tasks. This finding
partially supports our hypothesis. We had proposed that implicit learning
would be severely disrupted by damage to the basal ganglia and that explicit
knowledge would not benefit implicit learning. Both knowledge groups (EK
and No-EK) showed improvement with implicit practice (as demonstrated by
faster RTs and decreased tracking error relative to random sequences).
However, explicit knowledge did not benefit implicit learning; in fact, it
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
192
appeared that explicit knowledge had an interference effect on implicit motor-
sequence learning following BG stroke. For both the SRT and CT tasks the
BG EK group performed significantly worse than did the BG No-EK group.
The BG EK group did show an ability to decrease time lag of tracking,
however, not to the same degree as the No EK group. This interference effect
of EK was temporary and there were no differences between the knowledge at
the retention test. The temporary nature of this interference effect is
interesting, there were not explicit knowledge reminders provided at retention
and this factor may have “released” the BG EK subjects from the detrimental
effects of instructions.
The interference effect of instructions on motor performance was first
described by Bliss (1892) and subsequently by Boder (1935). Much more
recently, Reber (1976) noted that during the practice of an artificial grammar
task, when subjects were told of the existence of rules that governed the task,
performance was negatively affected. This finding was extended to a more
pure motor task by Green and Flowers (1991) who considered two groups of
subjects (without neurologic injury) as they practiced a probabilistic catching
task. In this experiment, one group was told of the probability relationships of
the target’s path and the other was left uninstructed. The uninformed group
demonstrated superior performance (as indexed by fewer errors) compared to
the instructed group. Thus, explicit instruction has been shown to have an
interfering effect on implicit motor task performance, however, these findings
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
193
have rarely been extended into subject populations with central neurologic
damage.
The inability of individuals with basal ganglia damage to alter their
motor output based on instructions (Jennings, 1995) or conditions of task
practice (Onla-or, 2001) has been demonstrated previously. However, in both
Jennings’ and Onla-or’s work the study population was one with bilateral basal
ganglia damage as a result of Parkinson’s Disease. Apparently the finding of
an inability to alter motor output based on external information extends to
individuals with unilateral basal ganglia stroke. Also important to consider, is
that in this dissertation, individuals with basal ganglia stroke practiced with the
less involved arm, corresponding to the less damaged basal ganglia.
Therefore, it may be concluded that both contra- and ipsilateral basal ganglia
play an important role in the incorporation of explicit knowledge during implicit
motor-sequence learning.
Further, it may be that the basal ganglia represent the critical site for
integration of explicit information into the implicit motor plan. Data to support
this contention may be derived from studies of patient populations,
neuroimaging reports, and consideration of cortical-sub-cortical loops. Onla-or
(2001) demonstrated the individuals with Parkinson’s Disease were unable to
take advantage of variable practice conditions that were intended to promote
evaluation (an explicitly mediated process) of the immediate past motor
response and incorporation of this information into the upcoming movement.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
194
Onla-or considered this to be a “set-shifting deficit”, where individuals with
damaged basal ganglia were unable to alter their motor output based on
changing conditions of practice. Jennings (1995) in similar work, explicitly
informed individuals with Parkinson’s Disease of upcoming alterations in a
motor task and reported that they were unable to take advantage of this
predictive information to prepare their next movement response. These
findings are very similar to those of this dissertation. The BG EK subjects
were unable to use explicit knowledge to plan their SRT responses and predict
the tracking path of the CT sequence. In fact, explicit knowledge actually
prevented the BG EK group from benefiting from practice, particularly on the
third day when they had full knowledge of the sequences (see Figure 7 and 8).
Grafton et al. (1995) reported increased activation in the putamen (using PET)
during SRT task practice once subjects gained explicit knowledge of the
sequence. This increase in basal ganglia activation following the acquisition of
explicit knowledge during implicit motor-sequence learning may represent the
integration of information from these two memory systems into a common
motor plan.
Further evidence for the basal ganglia’s role in the incorporation of
explicit knowledge into the implicit motor plan for sequences may be derived
from the reciprocal loops connecting it with the prefrontal regions that are well
known to support declarative memory (Parent & Hazrati, 1995; Middleton &
Strick, 1994). Also critical in this process is the thalamus, a site that is likely
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
195
integral for both integrating and mediating the output of the basal ganglia as
well as the cerebellum (Chapter 2, Figure 3). Clear from our data as well as
those in the literature, is that the basal ganglia are integral for implicit motor-
sequence learning, and it is our contention that they also represent the primary
site for incorporation of explicit knowledge into the implicit motor-sequence
plan.
Similar to the BG groups, individuals with sensorimotor cortical area
damage were unable to benefit from explicit knowledge. Again, this finding
contradicted our hypothesis of a positive effect of explicit knowledge on implicit
motor-sequence learning following sensorimotor cortical damage. For both
tasks the No-EK groups demonstrated superior performance for practice,
retention and transfer tests. Very interesting, was the complete lack of change
in tracking time lag seen on day three (when explicit knowledge was at its
maximum) for the SMC EK group (see Figure 12); the next day when there
was no instruction component to the tracking task the SMC EK group showed
a large improvement in time lag of tracking. This pattern of improvement
again suggests an interference effect for explicit knowledge, and perhaps the
improvements seen on the retention test day reflect a “release” or rebound
from the instructed condition.
Regions within the sensorimotor cortex have long been considered
critical for most motor output. It has been previously demonstrated that
ipsilateral M1 is active during the execution of complex, repetitive finger
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
196
movements (Shibasaki et al., 1993). Further, scanning data (PET) has shown
that once explicit knowledge is gained for implicit tasks, bilateral PMC are
active (Seitz et al., 1994). This suggests a strong role for PMC’s regulation
over sequence production under conditions where learners have access to
explicit knowledge for the task. Interestingly, PMC has strong connections
with the prefrontal regions associated with explicit knowledge (i.e. DLPFC;
Chapter 2, Figure 3). PMC is also highly reciprocally interconnected with the
caudate nucleus of the basal ganglia. Therefore, it is quite likely that damage
to and in regions associated with the PMC has as one consequence disrupted
integration of explicit knowledge into the motor plan, and thus, poorer implicit
motor-sequence performance. Even if explicit knowledge is integrated into the
motor plan through an intact basal ganglia it may be that focal stroke affecting
the motor output pathways (sensorimotor cortical areas) can disrupt the
implicit motor plan. The role of PMC in the integration of explicit knowledge
into the implicit motor plan must also be considered. Perhaps under normal
circumstances, both the basal ganglia and PMC are highly active in the
integration of explicit knowledge into the implicit motor plan. It may be that
disruption of either of these two regions leads to altered ability to take
advantage of explicit knowledge during implicit motor-sequence practice.
One more interesting detail of neural function is evident via
consideration of the data from the SMC groups. These groups (both EK and
No-EK) demonstrated some of the largest amounts of change across practice.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
197
This finding was largely due to the very slow RTs and the large magnitude of
tracking errors made early in practice and likely reflect a disrupted motor
output. This detail is interesting for two reasons. First, remember that these
individuals were practicing using the arm ipsilateral to brain damage and thus
their less involved hemisphere. As both the SRT and CT tasks were
unimanual, this finding demonstrates the large role that both hemispheres play
in learning and executing motor plans. Second, individuals in the SMC groups
showed a large benefit from practice - they were able to decrease their
tracking errors and RTs. This finding is also interesting to consider with
respect to the fact that those in the SMC groups showed the greatest amount
of upper extremity motor impairment (lowest Fugl-Meyer scores see Chapter
4, Table 2). Thus, despite being affected by the most severe strokes,
individuals in the SMC groups showed large improvements with practice in
implicit motor-sequence function on both tasks.
Summary
Comparison of the explicit knowledge conditions across the focal stroke
groups and two tasks demonstrates several interesting findings. First, explicit
knowledge appears only to benefit implicit motor-sequence learning when
subjects have intact basal ganglia function. In groups where the basal ganglia
were damaged, explicit knowledge of the sequence appeared to interfere with
implicit motor-sequence learning. Interestingly, this was true for both tasks
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
198
(the differential effect that explicit knowledge had on each task will be
considered in detail in Chapter 7). On the contrary, explicit knowledge only
appeared to benefit the HC group for the simple (SRT) task; there was no
effect of explicit knowledge on the CT task for the HC EK group. This likely
reflects the lack of salience of the explicit instruction provided for the complex
task for the HC EK group. Finally, explicit knowledge aided the CB EK group
for both the SRT and CT tasks. In light of the HC EK data, it would seem
logical that this finding is due to the increased attention subjects were able to
pay to the middle, repeating third of the CT sequence; localizing it in the
overall pattern may have proved beneficial for the CB EK groups performance.
The same may not have been true for the HC EK group, as tracking
performance was already good, a floor effect may have existed which limited
the benefit received by explicit awareness of the location of the repeating
segment. However, despite an advantage for the CB EK group, as compared
to the CB No-EK group, for decreasing RT and tracking error, an almost
complete inability to alter tracking time lag was demonstrated for both CB
groups. This finding probably reflects a fundamental function of the
cerebellum during motor tasks and was unaltered by explicit knowledge.
In general several conclusions may be drawn from these data. First, for
healthy subjects, explicit knowledge appears to act differentially on implicit
motor-sequence learning depending on task type. Second, as long as the
basal ganglia, and PMC are intact, it seems that explicit knowledge may
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
199
benefit implicit-motor-sequence learning. However, damage to the basal
ganglia disrupts the capacity for explicit knowledge to be integrated into the
implicit motor plan, and it interferes with implicit motor-sequence learning. A
similar finding appears to be true for damage to the sensorimotor cortical
areas, and consideration of neural function and interconnections lead to
consideration of the PMC as another important region for the incorporation of
explicit knowledge into implicit motor-sequence learning.
The next chapter (7) directly evaluates the additional factor of task type.
A complete examination of the interactions between lesion location, explicit
knowledge and task type will be considered. Chapter 7 will be the final data
based chapter, and is followed by a general discussion and summary of this
dissertation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
200
CHAPTER 7
The Interactions Between Explicit Knowledge, Task Demand, and Focal
Brain Damage on Implicit Motor-Sequence Learning
Overview
It is apparent from the data already presented, that the specific task and
its relative demands on the learner is another critical factor (in addition to
location of focal brain damage and explicit knowledge) that must be
considered in the evaluation of implicit learning. This contention is not
surprising, nor is it entirely novel (see Wulf & Shea, 2001), however, it has
never been extended to individuals with stroke and it has rarely been
deliberated in the literature describing implicit motor-sequence learning.
Recently, Wulf and Shea (2001) called for a broad review of the principles
derived from motor learning studies, to account for the relative complexity of
the task used, and we believe that a similar consideration may be necessary
for the investigations of implicit motor-sequence learning. This chapter was
constructed to address our third specific aim, to determine if there was an
interaction between task type and implicit motor-sequence learning.
Some data already have described the relationships between task
complexity and implicit motor-sequence learning. For example, Stadler (1992;
1995) and Cleeremans and Jimenez (1998) have shown that as sequences
became more complex (e.g. ambiguous versus probabilistic, more versus
fewer elements), increased practice is necessary before implicit learning
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
201
occurred. Through their experimental manipulations, these authors essentially
considered alterations in the cognitive demand of the sequence. Cognitive
demand of a motor task has been manipulated experimentally in three ways,
sequence length, composition, and attentional demand. Little attention has
been paid to the other component of implicit task complexity, i.e. the motor
demands (information processing load). Motor demand is related to the
physical demands of the task (e.g. degrees of freedom, number of joints
involved). Wulf and Shea (2001) have recently presented an extensive review
of the effect of increasing motor task complexity on motor learning and came
to several interesting conclusions. First, as motor task complexity increased,
more practice was necessary for learning to occur. Wulf and Shea (2001) also
noted that many of the long held principles of motor learning (e.g. contextual
interference, feedback frequency, physical guidance) had opposing effects on
complex versus simple motor skills. As an outgrowth of Wulf and Shea’s
(2001) work, task type was manipulated in this dissertation to enable an
evaluation of implicit motor-sequence learning for two separate tasks. We
chose to utilize two tasks that could be constructed to have similar structure
(i.e. 10 elements / reversals, ambiguous composition) but with different
information processing loads (discrete versus continuous) to test the
hypothesis that:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
202
Implicit sequence learning (as indexed by change in reaction time or tracking
error) will be less for a continuous visuomotor-sequence tracking task as
compared to a discrete stimulus-response sequence task. Further,
following unilateral damage to the sensorimotor cortex, basal ganglia, or
cerebellum
deficits in implicit motor-sequence learning will be more pronounced for a
continuous visuomotor-sequence tracking task compared with that for a
discrete stimulus-response sequence task. (Hypothesis #7)
First, in this chapter the methods used to test this hypothesis will be
presented. These will be followed by results and a discussion of these
findings with respect to other relevant literature and the data previously
presented in this dissertation.
Methods
To assess the impact of task type on implicit motor-sequence learning,
two separate tasks were designed. The SRT and CT tasks were selected for
use in this study for several reasons. First, both the SRT and CT tasks had
been previously described and it had been shown that the learning that
occurred for each was implicit. Other work had shown that practice of the
repeating element of each of these tasks led to improvements in performance
relative to a random condition. Again, for both the SRT and CT tasks
improvements were made with any declarative awareness of the repetitive
nature of the task (Nissan & Bullemer, 1987; Wulf & Schmidt, 1997). Second
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
203
both tasks were unimanual in nature, allowing use of the less involved arm for
the individuals with stroke. This was an important consideration as it allows us
to rule out any bias that might have otherwise been introduced in our data via
use of the affected upper extremity. Last the general structure of each task
needed to be equated. The nature of the SRT and CT tasks allowed us to
keep both sequences the same length - 10-elements for the SRT task and 10-
direction reversals for the CT task. Further, both tasks could be constructed
as ambiguous, without any probabilistic relationships among the elements of
the sequence. Finally, equivalent amounts of practice were provided for both
tasks (5 blocks / day for three days; retention and transfer tests of day four).
In order to assess the impact of explicit knowledge on these two
different motor tasks, we provided the same type of information at the same
relative times during practice for both the SRT and CT tasks. Thus, for those
in the EK groups, explicit knowledge was not given on day one, subjects were
informed only of the existence of the sequence on day two, and then full
explicit knowledge (verbal and schematic explanations) were provided on day
three of practice. Finally, all subjects practiced both the SRT and CT tasks,
eliminating any between subject effects on the comparison of motor task type.
To our knowledge this is the first investigation that has systematically and
directly compared two tasks as well as the impact of explicit knowledge in the
same group of subjects during implicit motor-sequence learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
204
Subjects
The same 37 individuals (10 HC, 7 CB, 10 BG, and 10 SMC)
participated. The characteristics of each of these subject groups has been
previously described (please see Chapter 4, Table 2), and will not be repeated
here.
Dependent Measures
To directly compare performance for the SRT and CT tasks several
data transformations were necessary. First, to assess practice performance,
slope of change for each day of practice was calculated by task for each
subject. This enabled a direct comparison of the amount of change occurring
with practice, by day and explicit knowledge condition, between the two
different tasks.
Retention test data were compared by calculating percent change in
repeated sequence responses at this time with respect to random. Thus, for
the SRT task, we divided RT for retention test by RT random and converted
this to percent. This procedure was repeated for tracking error data, and thus
lower percent random scores reflects more change or larger magnitude of
performance change at retention.
Statistical Measures
To determine the reliability of performance differences by task type,
explicit knowledge, and group across practice a Group (HC, CB, BG, SMC) by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
205
Knowledge (EK, No-EK) by Task (SRT, CT) by Block (day 1-3) ANOVA was
calculated with repeated measures correction for block on the slope measure.
A post-hoc Scheffe test was used to assess the locus of between group
differences (p<0.05).
Retention test data were similarly compared using a Group (HC, CB,
BG, SMC) by Knowledge (EK, No-EK) by Task (SRT, CT) ANOVA with the
dependent measure consisting of percent change. Again, to determine the
locus of between group differences a Scheffe test was used (p<0.05).
Results
Practice data are summarized by group in Figure 1 A-B. Visual
inspection of the slope data revealed several interesting findings. Much larger
slopes were generally noted for the SRT than the CT task for all groups and
both explicit knowledge conditions. The HC EK group generally showed a
positive effect of explicit knowledge as evidenced by larger slopes on day two
and three as compared to day one (Figure 1A). In contrast, no clear pattern
was evident for the CB groups (Figure 1B). A clear interference effect was
noted on day three for the BG EK group, when the slope of change was
negative for both tasks (Figure 1C). This is in stark contrast to the large
magnitude slope (indicating a good degree of change or improvement for the
task) noted for the BG No-EK group on day three (particularly for the SRT
task).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 0 6
Figure 1
80
6 0
4 0
C O 2 0
SRT
I I CT
-20
EK No-EK
Day 1
EK No-EK
Day 2
Explicit Knowledge
EK No-EK
Day 3
B
80
6 0
4 0
I I CT
c o 20 -
-20
i
SRT
EK No-EK : EK No-EK i EK No-EK
Day 1 Day 2 Day 3
Explicit Knowledge
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 0 7
EK No-EK
Day 1 Day 2
Explicit Knowledge
n ct
Figure 1 (continued)
No-EK No-EK
Day 3
SRT
D
80
60
40
c o 2 0
-20
EK No-EK
Day 1
EK No-EK
Day 2
Explicit Knowledge
SRT
C Z U CT
EK No-EK
Day 3
Figure 1. Slope of change for the SRT and CT tasks across practice by day
and explicit knowledge condition. (A) HC group, (B) CB group, (C) BG group,
(D) SMC group. A Group by Task by Block interaction was found.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 0 8
The SMC groups also showed a negative response to explicit knowledge, with
smaller slopes of change for day two and three than the No-EK group (Figure
1D). The impact of explicit knowledge for the SMC EK group was not nearly
as detrimental, however, as it was to the BG EK group. Statistical analysis
determined that for the practice data there was a main effect of Block
(F(1,58)=7.56, p=0.008). There was also a Task by Block interaction
(F(1,58)=6.06, p=0.017) and a Group by Block interaction (F(3,58)=6.161,
p=0.001). A Group by Task by Block interaction was also significant
(F(3,58)=7.26, p=0.000). Finally, two interesting trends were identified. First,
a Group by Knowledge by Block interaction was nearly significant (p=0.080),
as was a Group by Knowledge by Task (p=0.100).
The retention test data demonstrated that across groups larger percent
change in performance ability was made for SRT task relative to the CT task
and that this was maintained at retention (Figure 2). This was confirmed by a
Main Effect of Task (F(1,58)=10.13, p=0.002). The absence of a Main Effect
of Knowledge (p=0.206) confirms our earlier finding (reported in Chapter 6) of
a limited effect of explicit knowledge on implicit motor-sequence learning. Due
in large part to the poorer performance of the BG and SMC groups there was
a trend towards a Group by Task interaction (p=0.060).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
209
< D
CD
C
C O
sz
O
0 )
o
k—
< L >
a.
60
45
30
15
0
□ CT
SRT
EK No-EK
CB
EK No-EK EK No-EK
HC
EK No-EK
Explicit Knowledge
Figure 2. Percent change in repeated sequence as compared to random
sequence performance for both RT (SRT) and RMSE (CT) at retention. All
groups (HC, CB, BG, SMC) demonstrated more change in performance for the
simple task (Main Effect of Task). Larger changes in performance ability for
the repeated sequence relative to random is demonstrated by larger percent
change.
Discussion
In this discussion we will consider several key findings from our
comparison of two motor tasks. First, it was apparent in our data that all
groups were able to demonstrate more change in performance for the discrete
SRT versus the continuous CT task (Task by Block interaction for practice
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
210
data) and that this change was reliable and reflected learning (Main Effect of
Task at retention). Next, providing explicit knowledge differentially impacted
the different groups in this study (HC, CB, BG, and SMC). This was not a
large effect, as indicated by only a trend for statistical differences between
slopes across practice (Group by Knowledge by Block p=0.080). Last, no
large effect of explicit knowledge was found for the retention test data (Main
Effect of Knowledge p=0.206), suggesting that the impact of this variable was
transient.
The fact that larger changes in performance were noted for the SRT
task, as compared to the CT one, regardless of the presence of a stroke and
or explicit knowledge, suggests a very robust effect of task. The exact
mechanism for this difference was not directly evident from the measures
employed in this study; however, several speculative interpretations may be
formed. Perhaps, more practice is necessary to produce comparable change
scores during the practice of the CT task. A relationship has already been
demonstrated between the length of sequences for the SRT task and practice
amount (Pascal-Leone et al., 1993; Curran & Keele, 1993; Stadler, 1995). As
much more arm control was required for the CT task relative to the SRT task,
it may be that increased exposure to the CT task would enable more change.
Intuitively, this is an appealing interpretation of these data and future work
should endeavor to test this hypothesis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
211
In addition, it may be that there exists a separate factor that
distinguishes the amount of performance change that may be documented for
these two tasks. Evaluation of the ratio of exposure to the repeated versus
random elements of each task demonstrated a difference the number of times
the repeated and random sequences were practiced. This factor may have
interfered with or slowed implicit motor-sequence learning of the repeated
sequence for the CT task. Specifically, for each day five blocks of repeating
SRT sequence was practiced for every two blocks of random sequence.
Although transitions between the repeating and random sequences were
seamless, they were grouped separately; either repeating or random
sequences were being practiced, they were not interspersed. Due to the
nature of the CT task, each repeating sequence was bracketed by random
sequences. It has been previously demonstrated that when there are long
delays or secondary tasks inserted between repetitions of the repeating
sequence more practice is necessary for implicit learning to occur (Stadler,
1995). Also, for every five blocks of repeating CT sequence practice there
were ten blocks of random sequence, or twice as much exposure to random
sequences. Despite the fact that the number of times the repeating sequence
practice was the same for the simple and complex tasks, more random
sequence exposure occurred overall for the CT task. Additionally, the impact
of bracketing the repeating sequence with random ones may have been to
slow or delay implicit motor-sequence learning. This finding would be similar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
212
to those demonstrated in studies of trace conditioning (another form of non
declarative learning). In this scenario, long delays between the conditioned
and unconditioned stimuli make it harder for associations to be formed (see
Clark & Squire, 1999). It is likely that each of the factors discussed above
played a role in the differential change demonstrated for the two tasks
employed in this study. Further, these data illustrate the need for careful
consideration of many different in the assessment of implicit motor-sequence
learning.
A second finding from this between task analysis was the differential
impact of explicit knowledge across groups and tasks. This was not a large
effect (trend for Group by Task by Block interaction across acquisition
p=0.080), but several interesting features deserve discussion. Visual
inspection of the data show that when explicit SRT task knowledge was full
(day three) the HC EK group appeared to benefit (Figure 1 A). Similarly, for
the CB EK group, explicit knowledge of the sequence’s existence (day two)
and full explicit knowledge (day three) aided SRT performance (Figure 1B). It
is likely that two factors allowed explicit knowledge to aid SRT task
performance for the HC and CB groups. It may be that for the SRT task
explicit knowledge may be easily converted into movements (i.e. colors code
for which finger needs to flex to press the correct key). The explicit knowledge
Blue-Yellow-Red can quite easily be transformed into flexion of the third, first,
and then second finger (if performing the task with the right hand). Being able
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
213
to prepare movements in advance of the stimuli, therefore allows for much
faster responses and improved performance. Therefore for the SRT task, due
to its low information processing demand, explicit knowledge was beneficial as
it was easily translated into movements. This form of explicit knowledge was
salient, easy to remember, and simple to convert into motor control
commands.
Conversely, explicit knowledge for the CT task did not greatly impact
the performance of the HC EK group and only slightly benefited the CB EK
group (as noted by comparison to the CB No-EK group, which actually
demonstrated a flat or negative performance slope at this time). Explicit
knowledge for the CT task consisted of the information that the middle third
repeated, there were ten reversals in ten seconds, and a consistent waveform
was available for study. The first two aspects of this information may have
been useful - spatial location and relative length of the sequence. The explicit
waveform data, however, is not easily converted into motor commands. Even
explicitly knowing when and where direction reversals occur did not aid
performance very much. Perhaps what would have been more important for
task success was information regarding the appropriate control of movement
(e.g. limb segment acceleration and precise timing of agonist-antagonist
muscle activation patterns).
A second major factor aiding the HC and CB groups in using salient
explicit information (for the SRT task), was the presence of intact neural
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
214
networks to support the integration of explicit information into the implicit motor
plan. As has already been discussed, it is likely that the neural pathways
mediating the integration of explicit knowledge into the implicit motor plan
include the DLPFC, basal ganglia, PMC, and thalamus. These structures are
not directly affected by focal cerebellar stroke, allowing the benefit of SRT
explicit information to be exploited by the CB EK group In a similar fashion to
that seen for the HC EK group.
In stark contrast to the HC and CB EK groups, the BG EK group did not
show a larger performance improvement for the SRT versus the CT tasks
when explicit knowledge was available on day three. In fact, when complete
explicit knowledge was provided to the BG EK group on day three, a clear
interference effect was noted for both tasks (Figure 1C). In this instance, the
slope of change for the SRT and CT tasks was negative, indicating a
decrement in performance across day three. The robustness of this
interference effect was demonstrated by the fact that this was the one instance
where task type did not matter; SRT and CT performance was similarly and
severely impacted by explicit knowledge.
The interference effect of explicit knowledge is not new, however, this is
the first instance where it has been shown for implicit motor-sequence learning
of two separate tasks in individuals with focal basal ganglia stroke. Prior data
from Reber (1976; artificial grammar) and Green and Flowers (1991;
probabilistic catching) have also shown a negative effect of explicit knowledge
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1 5
on implicit learning. However, our work extends understanding of the
interference effect phenomenon. The interference effect found for the BG EK
group was not task specific, demonstrating something fundamental about the
basal ganglia’s role in integrating explicit knowledge into implicit motor-
sequence learning. It s very likely that the ability to incorporate explicit
knowledge into the implicit motor plan centers on basal ganglia function.
Further, as our subjects practiced both tasks unimanually, it seems that
bilateral basal ganglia are necessary for the use of explicit knowledge in
implicit motor-sequence learning. In this scenario, explicit knowledge is
mediated by DLPFC and via the cortical-basal ganglia-thalamo-cortical loop is
sent to the basal ganglia where it may be used to guide implicit motor
performance. Apparently, damage to even the contralateral basal ganglia
disrupts this integration of explicit information into the motor plan. This finding
supports others (Jennings, 1995; Onla-or, 2001) who have shown that
individuals with Parkinson’s Disease cannot use explicit information to plan
ahead for upcoming movements or greatly alter motor output (i.e. shift motor
sets). Even more interesting was the finding that this interference effect was
transient for both tasks and not evident in the retention test data. Perhaps,
explicit knowledge must be held in working memory to have a disruptive effect
on implicit motor-sequence performance following basal ganglia stroke.
The SMC EK group also appeared to be negatively impacted by explicit
knowledge for both tasks, relative to the No-EK group, on both days two and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1 6
three (when explicit knowledge was partial and full). This, however, was not
nearly to the same degree as seen for the BG EK group. It may be that for the
SMC EK group devoting attention to the sequence diverted cognitive
resources from implicit motor-sequence learning. It has been proposed that
applying an overt (explicit) control strategy to implicit tasks often detracts from
performance (Green & Flowers, 1991; Doyon, 1997a). Further, when subjects’
have knowledge of the existence, but not of the particular nature of rules for a
sequence (as they did on day two), they are often tempted to invent
relationships between responses (Reber, 1976). This may be what lessened
the slope of the SMC EK groups performance on day two for both tasks.
Complete explicit knowledge on day three aided SRT task performance
relative to day two but not as compared to the SMC No-EK group. For the CT
task, full explicit knowledge on day three did not greatly alter performance as
compared to day two or the SMC No-EK group on day three.
It is tempting to conclude that confounding the SMC EK group’s use of
explicit knowledge was damage in and around the PMC which has been
demonstrated to be important in regulating sequences of movement using
imaging techniques when explicit knowledge is available (Grafton et al., 1995;
Jenkins et al., 1994; Kawashima et al., 1994). Premotor cortex’s role in
sequence production has been demonstrated to increase bilaterally when
subjects are explicitly aware of the sequence (Seitz et al., 1994). Further,
PMC has connections with DLPFC (Krahauer & Ghez, 2000). It has been
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1 7
hypothesized that PMC activity is directly related to the spatial working
memory demands of sequencing tasks (Grafton et al., 1995). Damage in this
region would likely disrupt the ability of individuals to accurately incorporate
explicit knowledge into the implicit motor plan. Our data suggest however, that
the sensorimotor cortical areas (e.g. PMC) are not as critical for the integration
of explicit knowledge into the implicit motor plan as are the basal ganglia.
Finally, evaluation of the retention test data demonstrated two key (one
important because it was, and one important because it was not, statistically
significant) findings. These were a between task difference and the lack of an
effect of explicit knowledge. First, confirming both our hypothesis and
previous work by Wulf and Schmidt (2001), task type had a large impact on
subjects’ ability to demonstrate changes in performance. Significantly more
change relative to random sequence performance for the SRT task was found
relative to the CT task. This finding may indicate that more practice is
necessary to produce similar levels of change across tasks, and future work
could easily test this hypothesis. Subtle differences between the two tasks
may also alter the manner in which they are learned. As implicit learning must
be indirectly indexed via comparison of repeating sequence task performance
with that of a random one, the ratio of repeating to random sequence practice
may differ and will be critical for the learning of every task. Further, it may be
that what is being learned is fundamentally different for two separate motor
tasks. The SRT task requires more discrete stimulus response mapping, while
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1 8
the CT task was continuous. Thus, these two tasks differ vastly, and in
comparison to each other as well as to more “real world” movement
sequences. Therefore, caution should be exercised in generalizing the
findings from studies using traditional experimental implicit motor-sequencing
tasks to the entire implicit motor learning system.
Finally, it is clear from these data that the effect of explicit knowledge
did not have a large overall effect across the tasks; this was indicated by the
lack of an effect of Knowledge at the retention test. No benefit was seen for
the HC or CB EK groups, and the large interference effect seen on day three
of practice for the BG EK group was lessened. This is not the first description
of a slight impact of explicit knowledge. Shea et al. (2001) very recently (in
press) have data showing that there was a small advantage at retention test to
not explicitly knowing about a repeating sequence as compared to knowing
that a sequence repeated in every trial. This was demonstrated for a whole
body stabilometer tracking task.
Both our and Shea et al.’s (2001) data illustrate the need for a retention
test to adequately assess the impact of explicit knowledge on implicit motor-
sequence learning. For example, neither, Reber (1976) or Green and Flowers
(1991) employed a retention test design and it is unknown whether or not the
interference effect that they demonstrated would extend beyond practice and
become a more permanent feature affecting implicit motor-sequence learning.
In our data, the interference effect experienced by the BG EK group largely
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
219
temporary, as shown by the retention test. Additionally, there was no real
benefit of explicit knowledge for the other groups that could be ascertained at
retention. Caution should therefore be applied to interpretation of data from
work that did not employ retention test methodology as it may be simply
illustrating temporary effects of explicit knowledge (or perhaps other variables)
on implicit learning.
Summary
The combined effects of task, explicit knowledge, and lesion location
were considered in this chapter. It is apparent that task is a critical feature of
any investigation into implicit motor-sequence learning. Subjects
demonstrated larger amounts of change in performance for the SRT task
compared to the CT task, regardless of the location of stroke and the presence
or absence of explicit knowledge. This illustrates the need for investigations
into implicit motor-sequence learning that utilize various motor tasks and tasks
that more closely approximate real world movements (for one such example
see Shea et al., 2001).
A second major finding was the differential effect of explicit knowledge
across groups of subjects during practice. The HC and CB EK groups
appeared to show a benefit of explicit knowledge during practice, particularly
on day three when it was complete. This was likely due to the salience of the
provided information and the intact ability to integrate explicit knowledge into
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
220
the implicit motor plan. Conversely, the SMC EK group demonstrated an
inability to benefit from explicit knowledge; this was likely the result of damage
to and around the PMC region (a site important for explicit control of
sequential behavior). Finally, the interference effect of explicit knowledge on
the BG group was evident in the negative slope of change on day three for
both tasks. This finding supports our contention that the basal ganglia
represent the critical site for integration of explicit knowledge into the implicit
motor plan. Further, this data supports the conclusions drawn by others
(Jennings, 1995; Onla-or, 2001) of the importance of the basal ganglia for
predicting the upcoming movement based on past experience and shifting
motor strategy based on practice (or in this case explicit knowledge).
No large across task, residual effects of explicit knowledge were
evident at the retention test. This finding illustrates the importance of the
retention test in investigations into implicit motor-sequence learning.
Consideration of this methodologic detail should be included in all future
studies of implicit motor-sequence learning and caution should be exercised in
the interpretation of past work that has omitted retention tests as a design
feature.
The last chapter of this dissertation (8) will consist of a general
discussion and summary of the findings presented in Chapters 1 to 7.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
221
CHAPTER 8
General Discussion and Summary
Introduction
The aim of this chapter is to summarize the findings from this dissertation
and draw general conclusions based on this work. Three specific aims formed
the basis of this dissertation; 1) to investigate how three brain regions
(cerebellum, basal ganglia, and sensorimotor cortical areas) distinctly
contribute to and coordinate procedural motor-sequence learning; 2) to
determine the relative impact of declarative knowledge on procedural motor-
sequence learning following focal, unilateral brain damage to the cerebellum,
basal ganglia, or sensorimotor cortical areas, and 3) to determine if there is an
interaction between procedural motor-sequence learning and task type. Our
results yielded several important and interesting findings. All three focal stroke
groups demonstrated implicit motor-sequence learning capability (indexed by
improved performance on the repeating sequence components relative to a
random condition) for both the simple and complex tasks. However, none of
the focal stroke groups reduced their response times or tracking errors to the
same degree as the healthy controls. Further, each focal group demonstrated
different forms of impaired implicit learning. The cerebellar subjects failed to
take advantage of practice to alter their time lag of tracking. The performance
of the basal ganglia subjects was negatively impacted by explicit knowledge.
Finally, the sensorimotor cortical area subjects had the slowest response
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
222
times and largest tracking errors and were unable to improve their tracking
accuracy with practice. The significance of these findings is further
heightened by the fact that each of the focal stroke groups was using the less
involved arm and thus invoking the hemisphere undamaged by stroke.
Therefore, it is apparent that during implicit motor-sequence learning, bilateral
hemispheric function must be necessary. These general findings are
discussed in detail below with reference to our earlier work, the existing
literature regarding implicit motor-sequence learning, and possible
neuroanatomic relationships that underpin the explicit and implicit neural
networks.
Implicit Motor-sequence Learning following Focal Stroke
Contrary to our hypotheses, all three No-EK focal stroke groups
demonstrated implicit motor-sequence learning. Although in most cases
individuals in the focal stroke groups were slower and made more errors than
the control group, each focal stroke group was able to show improvements
with practice that were maintained at the retention test. The CB No-EK group
performed similarly to controls; decreasing RT and tracking error across
practice to the same relative degree as the HC No-EK group. These data
support previous work that demonstrated SRT task learning when the arm
contralateral to cerebellar damage was used for practice (Gomez-Beldarrain et
al., 1998). Distinguishing the CB from the HC No-EK groups was the time
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
223
series analysis. A significantly larger time lag was identified for the CB No-EK
group compared to the HC No-EK group. Further, the CB group’s deficit in
time lag of tracking was not altered by practice, despite significant
improvements in tracking accuracy. Thus, it appeared that individuals with
unilateral cerebellar stroke were able to re-create the tracking pattern
accurately in the spatial domain, however, they maintained a relatively fixed
“distance” (approximately 200 msec) behind the target despite three days of
practice. These data suggest that during pursuit tracking, the cerebellum is
involved in the timing of movements, and may also play a role in predictive
control once the tracking pattern is learned. Following cerebellar damage, the
ability to anticipate (based on prior experience, i.e. learning) the path of the
target was disrupted and subjects had to rely more on visually guided tracking.
A functional role for the cerebellum in the timing of action is not a new
functional conceptualization (Ivry, 1989), however, it has not been considered
or demonstrated during implicit motor-sequence learning.
A role for the cerebellum in predicting the path of the target, but not for
accurately reproducing the trajectory, appears to contradict earlier work by
Timmann and Horak (1996, 1998) for standing posture control. These authors
concluded that the cerebellum was not critical for predicting the next event
based on prior experience, but that cerebellar function was important for
accurate tuning of responses (Timmann & Horak, 1996). Several features
distinguish Timmann and Horak’s (1996, 1998) work from ours. First, the task
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
224
employed by Timmann and Horak (1996, 1998) was a postural response
paradigm, in which subjects with bilateral cerebellar degeneration had, or did
not have, advance knowledge of the upcoming perturbation. Subjects were
able to used advance information to predict perturbation amplitudes based on
prior experience, but could not modify the gain of their responses. In this
(Timmann & Horak’s) work, advance knowledge of the upcoming perturbation
was provided explicitly (repeated practice in a predictable blocked order).
Further, the task was a rapid automatic postural response, and subjects were
bilaterally involved. These factors make it difficult to directly compare our data
and Timmann and Horak’s, however, it is possible that the cerebellum acts
differently during full body postural responses than during implicit learning of a
unilateral upper extremity tracking task. Data from Timmann and Horak (1996,
Figure 2) show that initiation of postural responses by individuals with
cerebellar degeneration were slower than that of healthy controls (200 msec
versus 150 msec). Our CB No-EK subjects consistently tracked approximately
200 msec behind the target and perhaps this represents a lower limit of
cerebellar predictive timing capability. Future work will have to address this
issue through the use of tasks that demand rapid responses (less than 200
msec) to determine the precise role of the cerebellum in predicting upcoming
movements based on prior experience and explicit knowledge.
The timing deficit that we identified is important as it demonstrates that
despite relatively intact implicit motor-sequence learning capability following
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
225
unilateral cerebellar stroke, deficits in timing are not attenuated with practice.
This finding points to a role for the cerebellum in predicting the tracking path of
the target based on experience, but also indicates that as task complexity
increases (e.g. a continuous task) so may the importance of the cerebellum. It
may be that following cerebellar stroke, larger deficits in tracking time lag and
implicit motor-sequence learning will be revealed as more complex, real world
tasks are investigated.
Similar to individuals with cerebellar stroke, following unilateral BG
stroke, implicit motor-sequence learning was demonstrated for both the simple
and complex task. In fact, the magnitude of learning scores for the BG No-EK
group did not differ from that shown by the HC No-EK group. Our finding of
preserved implicit SRT task learning after unilateral stroke directly contradicts
Vakil et al.’s (2000). However, we provided three times more practice over
several days and did not mix the use of involved and less involved upper
extremities as did Vakil et al. (2000). These two critical features likely explain
the differences in conclusions.
Again, our time series analysis identified a significant deficit in time lag
of tracking at the retention test when the BG No-EK group was compared to
the HC group. This significant deficit in the time lag of tracking for those with
unilateral BG stroke, shows a deficiency in the capability to predict the
sequence to the same degree as those in the HC No-EK group. Similar to the
CB group, once adjusted for time lag, tracking accuracy for those in the BG
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 2 6
No-EK group did not differ from the HC group. Distinguishing the performance
of the BG and CB groups, however, is the finding that following basal ganglia
stroke, individuals benefited from practice, decreasing lag time across blocks,
while those with cerebellar stroke did not. Previous work demonstrating a set-
shifting deficit (Onla-Or, 2001) and an inability to use predicative knowledge to
form upcoming responses (Jennings, 1995) has relied on the explicit system to
convey information to the implicit. In our work, we assumed that improved
performance on the CT task (decreased tracking error) indicated increasing
implicit knowledge of the repeated segment and that this knowledge would
form the basis for predicting upcoming movements. This supposition is
partially true; subjects in the BG No-EK group decreased time lag of tracking
with practice but not to the degree seen for those in the HC group. From
these data it appears that the basal ganglia function during implicit motor-
sequence learning to serially coordinate the individual elements of the
movement sequence. When damaged unilaterally, deficits emerge in using
prior experience and knowledge (apparently both explicit and implicit) to
predict and prepare responses in advance. It may be that subjects in the BG
No-EK group could improve their time lag (and show some predictive
capability) with practice due to the unilateral nature of their brain damage.
This suggests that bilateral basal ganglia function is important for this task.
However, to date no work has examined the ability of individuals with
bilaterally damaged basal ganglia (e.g. Parkinson’s Disease) during implicit
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 2 7
learning of complex pursuit tracking tasks. It would be tempting to
hypothesize that severely impaired time lag of tracking would be demonstrated
by individuals with bilateral basal ganglia disease such as Parkinson’s. Such
a finding would more definitively identify the basal ganglia’s role in advance
planning of movements, and coordinating responses, based on prior
experience.
Our last hypothesis proposed that implicit motor-sequence learning for
individuals with SMC area damage would be decreased relative to healthy
controls. In fact, this also proved to be incorrect; the SMC No-EK group
demonstrated implicit motor-sequence learning of both tasks. However, when
absolute RT and tracking error scores of the SMC group were compared to
those of the HC group, much slower responses and larger magnitude of errors
were noted. The SMC group made large improvements with practice of both
tasks. This finding was somewhat a reflection of their extremely poor initial
performance. Further, it is important to note that the SMC group was the most
severely impacted by stroke as demonstrated by the lowest average upper
extremity motor Fugl-Meyer scores (SMC = 26; BG = 44; CB = 60; 66= highest
possible score and least impairment). Despite this, the SMC No-EK group
was able to improve greatly with practice and demonstrate implicit learning of
both tasks.
It is well known that during early learning Ml is active, rapidly
reorganizing with task practice (Kami et al., 1998). The CT task required only
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 2 8
movements in the arm ipsilateral to brain lesion, however, the large magnitude
of errors made early in practice by the SMC group indicates that bilateral Ml
function is important for early implicit motor-sequence learning. In spite of
early difficulty tracking the target and responding to the sequence, those in the
SMC group were able to demonstrate implicit motor-sequence learning at the
retention test. Despite showing implicit learning of the tracking task, once
again time series analysis revealed a significant deficit in tracking performance
for the SMC group. Both time lag and tracking accuracy were severely
impaired following sensorimotor cortical area stroke compared to that for the
HC group. Improvements in both measures were seen across the acquisition
phase, demonstrating an ability for change, however, evaluation of the
absolute magnitude of time lag and tracking error showed that the SMC group
was significantly impaired at the retention test. It may be that even more
practice would have aided the performance of the SMC group. This
supposition is strengthened by the finding of continued improvement from day
three to the retention test on day four.
The combination of poor tracking accuracy and large time lags
distinguished the performance of the SMC groups from that of the CB and BG
groups. If accurate movements cannot be produced, then it is very difficult to
use prior experience to predict the path of the target and reduce time lag.
Individuals in the SMC group did not have the experience of practicing the
tracking task correctly to guide future performance, and thus had little basis for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
229
improvement. It is likely that the sensorimotor cortical areas are important for
the final output and production of the motor plan for each trial (Miyai, Suzuki,
Kang, Kubota, & Volpe, 1999); it appears that at least for the tasks used in this
dissertation, if the SMC areas are disrupted by unilateral stroke, inaccurate
and poorly timed movements result. Perhaps due to the unilateral nature of
brain damage following stroke, improvements can occur after SMC lesions.
Evidence in support of this supposition comes from the significantly better
performance shown by the SMC group at the end of acquisition, illustrating
benefit of practice. Future work should endeavor to determine how even more
practice (beyond three days) would affect the performance and learning of
individuals with unilateral SMC area damage on implicit motor-sequencing
tasks.
Consideration of the data from all three focal stroke groups,
demonstrates several key findings regarding the impact of focal stroke as well
as the nature of the implicit motor learning system. First, all of those with
unilateral focal stroke were able to show some degree of implicit motor-
sequence learning; and all benefited from practice. This finding has important
clinical ramifications as it adds to the growing body of work indicating the
capacity for beneficial change, and need for intensive practice, to stimulate
improved performance and learning following stroke. Improved performance
during acquisition, and the demonstration of implicit motor-sequence learning
(i.e. retention) by all groups, illustrates the highly distributed nature of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
230
implicit learning neural network. Unlike the explicit system, no single focal
lesion appears to completely abolish the ability to learn implicitly. This finding
also shows the high degree of specialized function as well as the extremely
complex nature of the information stored in the implicit learning network.
Apparently, the motor plan, may be controlled and manipulated differently by
the cerebellum (for timing), the basal ganglia (serial control of individual
elements based on experience), and the sensorimotor cortical areas (accurate
formulation of the final plan for movement). Additionally, the deficits that were
shown by the three focal stroke groups, demonstrate the importance of
bilateral neural function during implicit motor-sequence learning.
Why do these results differ from Boyd and Winstein (2001)?
Important in our discussion of the impact of focal stroke on implicit
motor-sequence learning is consideration of how the finding of preserved
implicit motor-sequence learning following focal stroke found in this
dissertation might be reconciled with our earlier work. Previously, we reported
absent implicit motor-sequence learning in individuals who had a unilateral
stroke following one day of practice (24 sequence repetitions, experiment 1),
as well as after extended (three days) of practice (72 total sequence
repetitions, experiment 2; Boyd & Winstein, 2001) of the SRT task. However
in this dissertation, all three focal stroke groups demonstrated implicit motor-
sequence learning of the SRT task to varying degrees.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
231
Several subtle, yet important experimental variables likely explain the
differences between our previous findings and the results of this dissertation.
First, in the current work, more than twice as much practice was provided to all
of the subjects in the focal stroke groups than to the extended practice group
in our previous work. However, this factor is not the only explanation for these
disparate data, as demonstrated by evaluation of the performance of subjects
in the SMC No-EK group (the focal stroke group most analogous to those
individuals from Boyd & Winstein, 2001). Equating the amount of sequence
practice between the ST-Unaware group in our previous work and the SMC
No-EK group from the current research demonstrates equivalent amounts of
RT change: at the conclusion of 20 sequence repetitions the SMC No-EK
group showed a RT change score of 8 msec (see Chapter 5, Figure 1), while
in our previous work the ST-Unaware group from experiment 1 showed a RT
change score of -6 msec (see Chapter 3, Figure 1. However, equating the ST-
Extended Practice group from our previous work with the SMC No-EK group
halfway through day two of practice (70 sequence repetitions) showed very
different results. Subjects in the SMC No-EK group had decreased their RTs
by approximately 117 msec midway through day two of practice (Chapter 5,
Figure 1). At the equivalent time (conclusion of practice on day three) in our
previous work, subjects in the extended practice group had improved their RTs
by 8 msec (Chapter 3, Figure 2). Two additional factors likely account for this
difference. First, subjects in Boyd and Winstein (2001) had much larger (more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
232
severe) strokes than those in the current work. The extent of brain damage in
our earlier work tended to encompass a large degree of the sensorimotor
cortical areas as well as descending tracts (internal capsule) and in some
cases the striatum. Although, lesion location was not the focus of our original
work it is a critical variable that must be considered in any attempt to compare
results across studies.
A final factor that distinguishes these two studies was the method of
SRT task practice. In Boyd and Winstein (2001) an older, manual version of
the SRT task was used. In this experimental set-up the experimenter toggled
individual responses, leading to longer inter-trial intervals (approximated 1000-
1500 msec). In contrast, our current experimental set-up allows the SRT task
to run continuously from a computer program and controls the inter-trial
interval variably from 500-800 msec. It is likely that the long pauses inserted
between trials in our older experimental set-up induced an interference effect,
and reduced the magnitude of implicit motor-sequence learning. Stadler
(1995) demonstrated that inserting pauses between SRT task trials (2000
msec) severely disrupted sequence learning; under long-pause practice
conditions healthy control subjects showed very little implicit learning of the
SRT task. Over the course of SRT task practice, the long pauses in between
trials may disrupt sequence organization and subsequently sequence learning.
It has also been shown that implicit motor-sequence learning is susceptible to
the organization of the practiced sequence. When sequences are presented
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
233
in an organized fashion (i.e. no delays between individual trials or delays
between units / chunks of the sequence) implicit learning is facilitated (Stadler,
1993). In contrast when delays are randomly or consistently inserted, implicit
motor-sequence learning is diminished (Stadler, 1993, 1995).
Apparently following stroke individuals are more susceptible to the
interference effects associated with longer inter-trial delays. This is evident in
our previous work that used the older longer inter-trial interval SRT apparatus,
but demonstrated that healthy control subjects were able to show implicit
learning (RT -27 msec over 24 sequence repetitions as compared to that for
random sequences; Boyd & Winstein, 1998a). Perhaps following brain
damage it is more difficult to hold information over longer periods of time and
form associations between individual elements of the sequence.
Three factors in sum, likely explain the differences between our current
and previous findings. Practice amount, specific lesion location, and inter-trial
interval combined to alter the findings from, and distinguish, these two studies.
It is apparent that careful attention must be paid to each of these factors both
in the design and interpretation of studies investigating implicit motor-
sequence learning.
Next a discussion of the impact of explicit knowledge on implicit motor-
sequence learning will be presented with particular emphasis placed on the
potential sites and mechanisms for interaction between the declarative and
procedural systems.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
234
The Impact of Explicit Knowledge on Implicit Motor-Seauence Learning
There is little agreement in the literature concerning the impact of
explicit knowledge on implicit motor-sequence learning . Some have
concluded a benefit of explicit knowledge (Curran & Keele, 1993; Boyd &
Winstein, 2001), others a detriment (Reber 1976; Green & Flowers, 1991) and
some no effect (Reber & Squire, 1998). The most compelling explanations for
these different findings involve the different tasks used (SRT, artificial
grammar, probabilistic catching), the type and timing of explicit knowledge,
and the salience or relevance of the explicit information that was provided to
the learner. Further, no one has addressed the question of the impact of
explicit knowledge on implicit motor-sequence learning with regard to the
neural substrates that may mediate this transfer of information. This was one
of the primary questions addressed by this dissertation. How might explicit
knowledge affect implicit motor-sequence learning following focal stroke in the
neural regions that likely underpin the procedural learning system and mediate
the sharing of knowledge between the declarative and non-declarative
memory systems?
The first important finding from this investigation was the differential
impact of explicit knowledge on implicit motor-sequence acquisition
performance for the three focal stroke groups. For the HC group, explicit
knowledge aided acquisition performance for the simple (SRT) task but not for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
235
tracking accuracy of the complex (CT) task (Chapter 6, Figure 1 and 2). This
indicates that the explicit knowledge provided for the SRT task was more
relevant to the learner than that for the CT task. Likely, full explicit knowledge
of the SRT sequence’s composition was easily translated into a motor action
plan and used for advance preparation of upcoming responses. This enabled
faster RT’s for the repeating sequence for those in the HC EK group. Our
finding of a benefit of explicit knowledge for the SRT task is consistent with
Curran and Keele’s (1993), but contradicts Reber and Squire’s (1998). As has
been previously discussed in Chapter 6, this difference likely stems from two
factors. First, a pre-test documented that our subjects had explicitly learned
(to varying degrees) the sequence prior to implicit motor practice on day three
(86% accurate recognition, 73% recall). Second, subjects in this dissertation
had two days of SRT task practice (one unaware of the sequence and one
with partial explicit knowledge) before full explicit knowledge was provided.
Perhaps explicit knowledge is more useful for implicit motor-sequence
performance following some practice. This is an appealing conclusion as it is
commonly hypothesized that under normal (non-experimental) conditions,
explicit and implicit knowledge develop in parallel, yet awareness of explicit
knowledge only occurs after some degree of implicit task success has been
gained (Nissan & Bullemer, 1987; Grafton et al., 1995). Therefore, it may be
that for explicit knowledge to be beneficially incorporated into the implicit motor
plan three pre-requisites must be met; explicit knowledge must be actually
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
236
acquired by the learner (Boyd & Winstein, 2001), it must contain salient
information that may be somewhat easily translated into action (Segar, 1994),
and last it may be most helpful after some implicit motor practice and success
has already occurred (Nissan & Bullemer, 1987; Grafton et al., 1995).
For the complex task, no large effect of explicit knowledge on tracking
error was noted across acquisition performance (Chapter 6, Figure 2). This
finding lends strength to our earlier supposition that to benefit performance,
explicit knowledge must be salient and translatable into a motor action plan.
The lack of a significant difference between the groups also indicates that
contrary to Green and Flowers (1991) there was not an interference effect of
explicit knowledge. It may be that explicit knowledge acts differentially on
various task and that these effects depend on various features such as task
type (e.g. discrete versus continuous) and task goals (e.g. speed versus
accuracy). In this scenario, the SRT task is at the end of the simple / discrete /
speed goal continuum (where explicit knowledge is beneficial), the CT task lies
somewhere in the middle (neutral effect of explicit knowledge), and the
probabilistic catching task used by Green and Flowers (1991) is farther
towards the complex / continuous / accuracy goal end (where explicit
knowledge can interfere with performance).
The notion that our CT task lies somewhere in the middle of a task
complexity continuum is supported by the fact that explicit knowledge did
benefit time lag of tracking during CT task performance and learning (Chapter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
237
6, Figure 3). Those in the HC EK group significantly decreased their time lag
of tracking compared to that for the No-EK group, demonstrating an ability to
predict the path of the target during the repeating segment (HC EK ,109 msec
time lag retention test repeating segment, vs. HC No-EK, 160 msec). Explicit
knowledge of the location of the repeating segment in the middle third of the
tracking pattern may explain this difference.
The large effect size calculated for the CT task retention test data
demonstrates the importance of the retention test in accurately assessing the
impact of explicit knowledge (this issue will be discussed in detail at the
conclusion of this section, see below). There had not been differences
between the knowledge groups during acquisition. Further, there were no
explicit knowledge reminders provided at the retention test - subjects were
provided with neutral instructions to respond as quickly as possible and to
track as accurately as possible. Perhaps, the benefit of explicit knowledge
related to awareness of the location of the repeating segment for the CT task.
Apparently the cerebellum also has access to, and can benefit from,
explicit knowledge during implicit motor-sequence performance. This was
evident in the significant advantage held by the CB EK group during
acquisition (Chapter 6, Figure 4 and 5). Evidence for the preserved ability of
the explicit and implicit neural networks to share information also comes from
the finding that explicit knowledge and implicit performance of the repeating
sequence for both tasks increased in parallel on day three. Simultaneous
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
238
improvements in the explicit and implicit performance of the repeating
sequences illustrate that the neural structures interconnecting these two
systems do not reside in the cerebellum and were relatively unaffected by
focal stroke.
As has been previously discussed (Chapter 6), the most likely site of
transfer for information between explicit and implicit memory systems is the
basal ganglia-thalamo-cortical circuit (Selemon & Goldman-Rakic, 1985). This
circuit is not directly affected by cerebellar damage and is positioned to
incorporate explicit knowledge held by the DLPFC into the implicit motor plan,
which is likely, mediated by the basal ganglia. A second pathway from dentate
nucleus converges in the thalamus with information from the globus pallidus,
the output of the basal ganglia (Middleton & Strick, 1994). It is likely that
information important for the formation of the implicit motor plan converges via
these pathways in the thalamus and then is relayed in parallel to the prefrontal
cortex (i.e. DLPFC) where it may be influenced by explicit knowledge (Houk,
1997; Chapter 2, Figure 4). The unimanual nature of our tasks invoked only
undamaged cerebellar pathways, which may have allowed explicit knowledge
to be beneficially incorporated into implicit motor-sequence performance.
The impact of explicit knowledge on the CB EK group was transient; no
advantage of knowledge was identified for either task at the retention test.
Further, closer examination of the performance of the CB groups via our time
series analysis did identify a large deficit in the ability to decrease time lag of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
239
tracking regardless of the availability of explicit knowledge (Chapter 6, Figure
6). The impenetrability of the CB subjects tracking time lag supports the role
of the cerebellum in timed actions (Ivry & Keele, 1989; Ivry et al., 1988) and
demonstrates that explicit knowledge did not influence prediction and timed
tracking of the target’s path.
As the purported function of the basal ganglia is to coordinate the
elements of the motor plan, it is considered critical for implicit motor-sequence
learning. Neuroanatomic evidence suggests that it is ideally situated to
integrate explicit information into the implicit motor plan. Our data support
both of these roles for the basal ganglia. Following basal ganglia stroke,
explicit knowledge had an interference effect on implicit motor-sequence
performance; the BG EK group performed worse than the No-EK group (see
Chapter 6, Figures 7 and 8) as demonstrated by slower response times and
poorer tracking accuracy. Further, the BG EK group failed to decrease their
time lag of tracking to the same degree as the No-EK group. These data
extend previous work illustrating a failure to take advantage of explicit
predictive information in individuals with Parkinson’s Disease (Jennings, 1995;
Onla-or, 2001). Apparently, unilateral basal ganglia damage is sufficient to
induce a similar deficit in altering implicit motor output based on explicit
knowledge. It has been demonstrated that individuals with stroke confined to
the basal ganglia benefit less from rehabilitation than do those individuals with
stroke affecting the cortex or the cortex, basal ganglia, and internal capsule in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 0
combination (Miyai, Blau, Reding, & Volpe, 1997). Many facets of
rehabilitation rely on instruction and practice, and it is tempting to conclude
that the inability to beneficially utilize explicit knowledge to guide implicit motor
learning has a large impact of the efficacy of therapeutic interventions. It may
be that for individuals with basal ganglia stroke or disease therapeutic
techniques that do not invoke the declarative memory and learning systems
will enhance rehabilitation (e.g. motor skill learning) outcomes.
The basal ganglia are reciprocally connected (via thalamus) with the
prefrontal cortical areas that support declarative memory (Parent & Hazrati,
1995; Middleton & Strick, 1994). If the role of the basal ganglia is to update
and control the serial elements of the motor plan as it evolves, then it would be
very beneficial for this region to also have access to the guiding effects of
explicit knowledge. Such an integration of systems would facilitate conscious
control of motor actions and permit the incorporation of augmented feedback
into upcoming motor plans. Our data demonstrate disruptions in this process
following unilateral damage to the basal ganglia and illustrate the necessity of
bilateral circuits. Lastly, the interference effect of explicit knowledge on implicit
motor-sequence acquisition was temporary. In fact, the BG EK group seemed
by visual inspection, to improve from day three to the retention test when they
were “released” from explicit knowledge of the sequences (Chapter 6, Figures
7 and 8).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
241
Damage to the sensorimotor cortical areas clearly affected motor output
of both the EK and No-EK groups. Providing explicit knowledge did not benefit
the SMC EK group, in fact, another interference effect was noted (Chapter 6,
Figures 10 and 11). It has already been demonstrated that the PMC is active
bilaterally once explicit knowledge is gained for implicit motor tasks (Seitz et
al., 1994). A role for PMC in the mediation of information exchanged between
the explicit and implicit systems is also supported by neuroanatomic evidence
showing reciprocal interconnections with both prefrontal regions and the
caudate nucleus. It is likely that the interference effect experienced by
individuals in the SMC EK group was the result of disrupted PMC function; as
explicit knowledge was available yet the mechanism for its integration into the
motor plan was interrupted. As the function of PMC is bilateral (even during
unimanual tasks) during sequence production when external or explicit
knowledge is available (Seitz et al., 1994; Toni et al., 1998), it would be
disrupted by unilateral stroke and lead to an inability to use explicit knowledge
beneficially. However, when unilateral sequence production is implicit, SMA is
active contralaterally (Boecker et al., 1998; Toni et al., 1998) and thus would
be less affected by unilateral stroke, facilitating implicit motor-sequence
learning when subjects did not have explicit knowledge. Similar to our
previous findings, at retention there were no differences between the EK and
No-EK groups.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 2
In summary, healthy control subjects benefited from explicit knowledge
for the simple task and for time lag of tracking. Tracking error and accuracy
were not impacted by explicit knowledge likely because of the lack of salience
of the information and the already very accurate performance (floor effect).
Explicit knowledge appeared to benefit acquisition performance for the CB EK
group for both tasks; perhaps the localization of the repeating segment aided
tracking ability. However, a stable tracking time lag was identified, indicating a
role for the cerebellum in the timing of pursuit tracking. Both the BG and SMC
EK groups showed an interference effect of explicit knowledge during the
acquisition phase. This likely reflects the important role of the cortico-basal
ganglia-thalamo-cortical loops in the incorporation of explicit knowledge into
the implicit motor plan. In addition, it appears that damage in the sensorimotor
cortical areas (particularly PMC) can also disrupt the integration of explicit
knowledge. Interesting, for the CB and BG groups the impact of explicit
knowledge (both beneficial and detrimental) was temporary; no effect was
noted at the retention test (see below)
Retention Test Design
The differential impact of explicit knowledge on our retention test data is
one of the most interesting and important findings of this work. It is important
to consider how vastly different our conclusions might have been if we had not
conducted retention tests on day four. In this scenario, one might conclude
that explicit knowledge has a strong beneficial effect on the ability of cerebellar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
243
subjects to learn the SRT and CT tasks (refer to Chapter 6, Figures 1 and 2).
However, consideration of the retention test data demonstrate no statistically
significant differences between the CB EK and No-EK groups at retention as
well as no advantage for one over the other for transfer tests. Examination of
the BG groups’ practice data leads to the opposite conclusion; explicit
knowledge appears to be very interfering for this group. On the contrary,
retention test data do not show a reliable difference between the BG EK and
No-EK groups.
This temporary effect of explicit knowledge is in itself a very intriguing
finding. Perhaps explicit knowledge needs to be held in some form in working
memory during practice to have an impact on performance. Remember, that
in this work, the only information provided on day four were the instructions to
“respond as quickly as possible” (SRT task) and “track as accurately as
possible” (CT task). No explicit knowledge instructions were provided.
Apparently, for the CB and BG subjects explicit knowledge from the previous
day was either not considered to be useful, and thus purposefully not re
invoked, or forgotten.
This finding of a transient impact of explicit knowledge and a
subsequent lack of between group differences for this factor at the retention
test for some groups of subjects demonstrates the need for a re-consideration
of previous work that investigated the interaction between explicit knowledge
and implicit motor-sequence learning. For example, Curran and Keele (1993,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 4
experiment 2) showed a clear benefit of explicit knowledge for individuals
without neurologic damage practicing the SRT task. However, in this work no
retention test was employed. It is therefore unclear whether the advantage
that the instructed subjects showed over the non-instructed was a temporary
effect or if it led to a relatively permanent change in behavior.
Green and Flowers (1991) drew the opposite conclusion from Curran
and Keele (1993), and stated that in their work using probabilistic catching (a
much more complex task), explicit knowledge of the probability relationship of
the target’s path was detrimental to implicit learning. Although subjects
practiced for several days it is unclear whether or not they were re-instructed
at the beginning of each session, and again there was no retention test. One
investigation into the interaction between explicit knowledge and implicit motor
learning did employ a retention test design. At retention in Shea et al.’s (2001)
work, very small differences were noted between the explicit and implicit
conditions. Two other studies examining implicit motor-sequence learning
have used a retention test design, however, the primary purpose of each of
these was to determine the impact of a secondary tone-counting task (Frensch
et al., 1998, 1999).
The hallmark of motor learning is a “relatively permanent change” in
capability for responding. Without a retention test the long term impact of
explicit knowledge on implicit motor-sequence learning might be unclear and
easily misinterpreted. In light of our data, it would seem prudent to call for a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 5
re-evaluation of earlier conclusions regarding the benefit or detriment of
explicit knowledge on implicit motor-sequence learning. Further, we believe
that the inclusion of the retention test design into future work in this area is
critical. Such a re-assessment would be analogous to Salmoni, Schmidt, and
Walter’s (1984) call for integration of the retention test into studies of motor
learning and the re-consideration given to many principles of feedback
(knowledge of results) in motor learning that followed the routine incorporation
of this study design into most investigations. Frensh et al. (1998) made a
similar call for use of retention tests in their evaluation of the importance of
attention for implicit motor-sequence learning, and we now extend this to the
impact of explicit knowledge.
Transfer or Generalizability of Learning
None of the groups were able to demonstrate transfer of learning for
either task. Transfer of learning or generalizability, documents the extent to
which learning of a particular task contributes to other related skills. In this
dissertation, transfer learning was assessed through performance (one block
of practice) of similar sequences to those practiced. We hypothesized that the
information acquired during implicit motor-sequence learning is stored as an
abstract plan for movement. Thus, if implicit motor-sequence learning was
demonstrated then the fundamental nature of the learned pattern would
facilitate transfer to novel scalings of similar sequences. There are several
explanations for the lack of generalizability of implicit learning for the two
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 6
tasks. First, and least likely, is that implicit motor-sequence learning is not
stored as an abstract motor plan and therefore, transfer did not occur. Other
studies have demonstrated transfer of implicit motor-sequence learning of
tracking tasks (Pew, 1974; Wulf & Schmidt, 1997) in both the spatial and
temporal domains. Data from these studies seem to point to the formation of a
general motor plan developed during implicit practice of tracking tasks.
A second possibility is that explicit knowledge (or the lack of) interacted
with the generalizability of implicit motor-sequence learning. In one study Lee
and Vakoch, (1996) separated explicit and implicit learning and then assessed
transfer effects. Subjects performed an explicit task (in which they were aware
of the rules for success) and an implicit task (where they were unaware of the
regularities in their responses). When asked to generalize their learning to
new, yet similar, simple and complex tasks, subjects demonstrated positive
transfer of their explicit learning but not of their implicit learning. From these
data it appears that explicit knowledge of the simple task facilitated
generalizability while implicit knowledge may have caused a type of
interference effect; subjects were unable to adapt previously learned general
rules to the new task. Some authors have speculated that one function of
explicit knowledge is to promote the transfer and generalizability of the implicit
motor plan (Segar, 1994; Curran, 1989). However, in this dissertation half of
the subjects had explicit knowledge of the repeating nature of their responses,
yet they still failed to demonstrate generalizability.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 7
The last and most plausible explanation for the lack of transfer shown
by all groups for both tasks in this dissertation is that implicit learning was not
robust enough to permit stable formation of a generalized representation of the
motor plan. Perhaps for transfer of implicit motor-sequence learning to occur,
performance needs to be at asymptote. More practice would have benefited
SRT and CT task performance and learning. Further, it may also have
facilitated the formation of an abstract motor plan for responding. This would
have enabled the demonstration of transfer. Future work concerning the
generalizability of implicit motor-sequence learning should attempt to
disentangle the effects of explicit knowledge and the robustness of learning by
providing practice until performance is at asymptote (stable, no longer
changing significantly) and then assessing both generalizability and explicit
knowledge for the repeating sequence.
Neuroanatomic Interactions
Our data support a neuroanatomic interaction between the declarative
and procedural memory systems during implicit motor-sequence learning.
Specifically, there are two prominent pathways that likely are involved in the
mediation of information between the declarative and procedural systems -
the cortico-basal ganglia-thalamo-cortical loop (Parent & Hazrati, 1995) and
the cerebello-thalamo-cortical pathway (Middleton & Strick, 1994). In addition
it is possible that there is some interaction between these two.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 8
Projections directly connect DLPFC and basal ganglia (caudate), while
parallel projections also originate in the posterior parietal lobe and PMC, and
terminate in the caudate where they sit adjacent to one another (Selemon &
Goldman-Rakic, 1985; Chapter 2, Figure 3). Accumulation of these
projections in the basal ganglia demonstrates that it is critical position in the
network linking declarative and procedural systems. Output from the basal
ganglia pass via thalamus back to the DLPFC to partially close this loop. The
primary output pathways from the cerebellum (dentate) also project via
thalamus to the DLPFC (Chapter 2, Figure 4). Thus, regions within the basal
ganglia, cerebellum, and sensorimotor cortical regions are all associated either
directly or indirectly with the prefrontal areas supporting declarative
knowledge.
Our finding of a large interference effect of explicit knowledge following
basal ganglia stroke suggests that the prefrontal cortico-basal ganglia-
thalamo-cortical circuit represents the primary neuroanatomic pathway for
information transfer between the declarative and procedural memory systems
during motor-sequence learning. The cerebello-thalamic-cortical pathway may
be able to influence and share information with cortico-basal ganglia
connections as they both project to DLPFC; a site where information could be
entered into the basal ganglia loop from the cerebellar loop for integration into
the motor plan (Chapter 2, Figure 4). Therefore, following unilateral cerebellar
damage explicit knowledge could still be beneficially incorporated into the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
249
implicit motor plan. Projections from the posterior parietal lobe and PMC to
the caudate, as well as PMC to DLPFC augment the connections between
declarative and procedural systems. Sensorimotor cortical area damage
disrupts these connections and explicit knowledge cannot benefit the
formation of the final motor plan.
The Interactions between Explicit Knowledge. Task Type, and Implicit
Motor-Sequence Learning
Our final question concerned the impact of two different task types on
implicit motor-sequence learning. From the literature reviewed, our data and
previous discussions, it is clear that task type is a major factor to be
considered in evaluating implicit motor-sequence learning. In Chapter 6 we
showed that explicit knowledge had a differential effect on the two tasks in the
HC group, aiding HC performance on the SRT task and having little impact on
performance of the CT task. This was the only instance in which there was a
dissociation in the effect of explicit knowledge on the same group of subjects
practicing two separate tasks. This differential effect of explicit knowledge
suggests that conclusions from other work investigating the interaction
between explicit knowledge and implicit motor-sequence learning may not
generalize across tasks. Across tasks and subject groups the impact of
explicit knowledge was somewhat temporary. This topic has already been
discussed in detail (see above), however, as only one other study (Shea et al.,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 0
2001) has used a retention test, the combined long-term impacts of task type
and explicit knowledge remain unclear.
Direct comparison of the two tasks employed in this dissertation
revealed that relatively more positive change in performance occurred during
acquisition and was maintained at retention for the SRT task than the CT task
for all groups. This finding was true regardless of the presence of explicit
knowledge and or focal stroke and demonstrates the importance of task on the
change in ability seen across acquisition. What factors may explain the robust
differences in change scores for these two tasks? We constructed the SRT
and CT tasks to be relatively equal, each consisted of a 10-element
ambiguous repeating sequence. Thus, their basic structure (sequence length
and composition) was equated. Further to avoid a practice effect, we provided
the same number of repetitions, 150 trials over three days of acquisition. In
retrospect, this was likely insufficient practice for the CT task. The CT task
demanded continuous upper extremity control and it is very likely that more
exposure to (practice of) the task would lead to larger changes in
performance. A second intriguing possibility is that different practice
schedules (e.g. variable practice) and or feedback (e.g. KR in the form of
whole sequence tracking error) would maximize the benefit of practice and
enable more change in less time. At this point the relationships between
change in performance of an implicit task, practice amount, practice
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
251
schedules, and feedback remain unclear, however, future work should
address the interrelationships that exist among these factors.
We have already discussed the different ratios between repeated and
random sequence exposure for the SRT and CT tasks. This is a second factor
that likely contributed to the different amounts of change seen for the same
subjects on these two tasks. Across practice for the SRT task 15 sequence
blocks and 6 random blocks were completed. For the CT task, 15 sequence
blocks were accompanied by 30 random blocks. Thus, the ratio of repeated to
random for the SRT task across practice was 2.5:1, and for the CT task was
0.5:1. In an attempt to determine how the different ratios between the random
and repeated sequences for these two tasks impacted performance we
equated them relatively and compared their slopes. Block three on day one of
SRT task practice was the point determined to be approximately equivalent to
the end of acquisition practice for the CT task on day three. Using the HC No-
EK group as an example, we determined that the slope of change from the
beginning of practice to block three (on day one) for the SRT task was 13.8
(±10.1) as compared to the slope of change across all three days of practice
for the CT task 3.8 (±0.98). Thus, equating the ratio of repeated and random
sequence practice may elucidate some of the differences in the change seen
across practice for these two tasks, but it cannot be the sole explanation.
Another possible factor that differentiated performance on these two
tasks was the juxtaposition of the random and repeated segments during CT
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 2
practice. Each repeated CT segment was bracketed by two random
sequences. This factor may have delayed or interfered with CT task
performance. Delays (Clark & Squire, 1999) and the insertion of other
information or tasks (Shea & Kohl, 1991) into the inter-trial interval have been
previously shown to interfere with performance and learning of motor tasks. It
is likely that in this scenario it becomes harder to form associations among the
repeated aspects of the task. In contrast, SRT practice was structured such
that the random and repeated sequences were blocked separately. This
difference in practice structure was certainly another factor that likely
contributed to the change that subjects were able to effect in their ability on
these two tasks.
The results of our analysis of explicit knowledge, task type, and focal
brain damage on implicit motor-sequence learning demonstrate the difficulty in
comparing across tasks, and conversely, drawing conclusions from
performance on one task. The HC group was able to take advantage of
explicit knowledge for the SRT task, but not for the CT task. However, the
focal stroke groups were all similarly impacted by explicit knowledge on both
tasks. This suggests that careful consideration must be paid to the actual task
being practiced if performance on one is to be compared to another. Finally,
great care should be taken in making generalized conclusions and predictions
regarding performance and implicit learning ability, as well as the impact of
explicit knowledge, on that ability in different populations.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
253
Possible Clinical Applications
The findings of this dissertation may extend into the clinical arena in
several ways. First, all of the individuals with focal stroke demonstrated the
capacity for implicit motor-sequence learning. This is a new and important
finding. It appears that with sufficient practice for both the simple and complex
tasks improved performance seen during acquisition can extend into retention
and reliably be considered as learning. In addition, the most impaired
individuals in this study, the SMC group, showed the greatest amount of
improvement and the largest learning scores for both tasks.
Despite this ability to show implicit learning, all of the focal stroke
groups performed worse than the healthy control groups and demonstrated
specific deficits based on lesion location. As all participants in this study used
the arm not directly affected by stroke and thus invoked the undamaged
hemisphere, performance deficits strongly suggest that bilateral hemispheric
function is necessary during implicit motor-sequence learning. Thus, in the
clinical arena, care and perhaps therapeutic interventions may have to be paid
to the less involved upper extremity following stroke.
Another consideration was the differential impact of explicit knowledge
on the focal stroke group. Broadly speaking, the cerebellar group was the
group that benefited from explicit knowledge of the task. However, in the
clinical realm, the predominant pattern of stroke affects the middle cerebral
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 4
artery (Brust, 2000) and damages the sensorimotor cortical areas and / or
basal ganglia. Both of these groups demonstrated an interference effect of
explicit knowledge where their performance was negatively impacted. This
detrimental effect of explicit knowledge was temporary (not apparent at the
retention test), however, the larger impact of poor task performance on the
motivation and enthusiasm of patients for the motor skill being practiced
remains unknown. Presently, in clinical practice it is common for little attention
to be paid to the locus of brain damage (other than hemisphere). As
rehabilitation stays and direct contact time with health care providers
continues to decrease (as has been the recent trend) it is imperative that
therapeutic time be maximized. Thus, it is critical that healthcare providers
incorporate a thorough understanding of the locus of brain damage in
conjunction with the specific behavioral function associated with the region of
stroke into their plan of care.
It is also clear that for different implicit motor tasks increased practice
will be necessary in order to affect long term change in performance. Most
likely is that practice amount will need to be tailored to the specific skills and
goals of each individual learner. Unfortunately, financial factors dictate length
of rehabilitation stay, however intensity of therapeutic time may be
manipulated by the healthcare provider. Finally, and likely most importantly,
patients must invest in their own well being and recovery, extending their
practice into their personal time and daily life.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 5
Limitations and Suggestions for Future Work
Several factors limit the scope and expanse of this work. First, and
most obvious, is the lack of control and consistency across the size, density,
and location of focal brain damage. Best efforts were made to control for
stroke location. In the future, other methods (e.g. PET, fMRI, MEG) may be
able to more precisely evaluate the impact of focal stroke on implicit motor-
sequence learning. A second limitation was the number of participants in each
group. This factor reduced the statistical power associated with our findings.
Unfortunately, a limited pool of subjects with appropriately located stroke was
available and willing to participate. In addition, the logistical realities of
conducting a learning study (four separate days of practice at the University of
Southern California) reduced our number of subjects. Last, comparing two
motor tasks demonstrated that perhaps more practice would have benefited
performance of the CT task.
Specific recommendations for future work were incorporated into the
discussion sections of this dissertation and will not be repeated here. Rather,
several general concepts that would be interesting and important to study will
be considered. First, the role of the thalamus, the central integrative relay for
information between the cerebellum, basal ganglia, and prefrontal cortex has
largely been ignored in this study as in others (for one exception see Rauch,
Whalen, Curran, Mclnerney, Heckers et al., 1998). It is unclear what specific
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 6
contributions are made by this structure during implicit motor-sequence
learning. Second, we showed a temporary interference effect of explicit
knowledge for the BG group. It is tempting to consider whether this transient
effect of explicit knowledge might have been more permanent if it was
reintroduced and reinforced over many (more than three) days of practice and
I or at the retention test. Last, it is unknown how explicit knowledge would
impact very complex (e.g. whole body) motor tasks, and how much practice
would be necessary for their learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 7
REFERENCES
Ackermann, H., Daum. I., Schugens, M. M., & Grodd, W. (1996). Impaired
procedural learning after damage to the left supplementary motor area (SMA).
Journal of Neurology, Neurosurgery, and Psychiatry, 60, 94-97.
Alexander, G. E., & Crutcher, M. D. (1990). Preparation for movement: Neural
representations of intended direction in three motor areas of the monkey.
Journal of Neurophysiology, 64, 133-150.
Alexander, G. E., DeLong, M. R., & Strick, P. L. (1986). Parallel organization
of functionally segregated circuits linking basal ganglia and cortex. Annual
Review of Neuroscience, 9, 357-381.
Appollonio, I. M., Grafman, J., Schwartz, V., Massaquio, S., & Hallett, M.
(1993). Memory in patients with cerebellar degeneration. Neurology, 43, 1536-
1544.
Bapi, R. S., Doya, K., & Harner, A. M. (2000). Evidence for effector
independent and dependent representations and their differential time course
of acquisition during motor sequence learning. Experimental Brain Research,
DOI 10.1007/S002219900332.
Barone, P., & Joseph, J. P. (1989). Prefrontal cortex and spatial sequencing in
macaque monkey. Experimental Brain Research, 78, 447-464.
Berstein, N. (1967). The Co-ordination and Regulation of Movements. Oxford,
Permagon Press.
Berridge, K. C., & Whitshaw, I. Q. (1992). Cortex, striatum and cerebellum:
Control of serial order in a grooming sequence. Experimental Brain Research,
90, 275-290.
Berry, D. C., & Broadbent, D. E. (1988). Interactive tasks and the implicit-
explicit distinction. British Journal of Psychology, 79, 251-272.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 8
Bischoff, A. (1998). Modeling the basal ganglia in the control of arm
movements. (Doctoral Dissertation University of Southern California, 1998).
Dissertation Abstracts International, 59-08B, 3924.
Bischoff, A., Arbib, M. A., & Winstein, C. J. (1997). Modeling the role of the
basal ganglia in a reciprocal aiming task. Proceedings of the Fourth Annual
Joint Symposium on Neural Computation, University of Southern California, 7,
20-27.
Bliss, C. B. (1892). Investigations in reaction time and attention. Studies from
the Yale Psychology Laboratory, 1, 1 -55.
Boder, D. P. (1935). The influence of concomitant activity and fatigue upon
certain forms of reciprocal hand movements and its fundamental components.
Comparative Psychology Monographs, 11.
Boecker, H., Dagher, A., Ceballos-Baumann, A. O., et al. (1998). Role of the
human rostral supplementary motor area and the basal ganglia in motor
sequence control: Investigations with H2 150 PET. Journal of
Neurophysiology, 79,1070-1080.
Bower, J. M. (1997). Is the cerebellum sensory for motor’s sake, or motor for
sensory’s sake: the view from the whiskers of a rat? Progress in Brain
Research, 114, 463-495.
Boyd, L. A., & Winstein, C. J. (2001). Implicit motor-sequence learning in
humans following unilateral stroke: the impact of practice and explicit
knowledge. Neuroscience Letters, 298, 65-69.
Boyd, L. A., Baker, L. L., & Winstein, C. J. (1999). Explicit knowledge
promotes implicit learning of the serial reaction time task in individuals with
unilateral brain damage. Society for Neuroscience Abstracts, 25, 1900.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
259
Boyd, L. A., & Winstein, C. J. (1998a). Serial reaction time learning in
individuals with unilateral brain damage. Society for Neuroscience Abstracts,
24,2116.
Boyd, L. A., & Winstein, C. J. (1998b). Implicit learning is diminished in
individuals with cerebrovascular accident. Neurology Report, 22, 179.
Brooks, V., Hilperath, F., Brooks, M., Ross, H. G., & Freund, H. J. (1995).
Learning “what” and “how” in a human motor task. Learning & Memory, 2, 225-
242.
Buchner, A., & Wippich, W. (1998). Differences and commonalties between
implicit learning and implicit memory. In Stadler, M. A., & Frensch, P. A.
(Eds.), The Handbook of Implicit Learning, (pp. 3-46). Thousand Oaks: Sage.
Butters, M. A., Glisky, E. L., & Schacter, D. L. (1993). Transfer of new learning
in memory-impaired patients. Journal of Clinical and Experimental
Neuropsychology, 15, 219-230.
Brust, J. C. M. (2000). Circulation of the Brain. In Kandel, E. R., Schwartz, J.
H., & Jessel, T. M. (Eds.), Principles of Neural Science, (pp. 1306-1316). New
York: McGraw-Hill.
Chen, R., Cohen, L. G., & Hallett, M. (1997a). Role of the ipsilateral motor
cortex in voluntary movement. Canadian Journal of Neurologic Science, 24,
284-291.
Chen, R., Gerloff, C., Hallett, M., & Cohen, L. G. (1997b). Involvement of the
ipsilateral motor cortex in finger movements of different complexities. Annals
of Neurology, 41, 247-254.
Clark, G. A., McCormick, D. A., & Thompson, R. F. (1984). Effects of lesions
of the cerebellar nuclei on conditioned behavior and hippocampal neuronal
responses. Brain Research, 291,125-136.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 0
Clark, R. E., & Squire, L. R. (1998). Classical conditioning and brain systems:
The role of awareness. Science, 280, 77-81.
Clark, R. E., & Squire, L. R. (1999). Human eyeblink classical conditioning:
Effects of manipulating awareness of the stimulus contingencies.
Psychological Science, 10, 14-18.
Cleeremans, A., & McClelland, J. L. (1991). Learning the structure of event
sequences. Journal of Experimental Psychology: General, 120, 235-253.
Cleeremans, A., & Jimenez, L. (1998). Implicit sequence learning: The truth is
in the details. In Stadler, M. A., & Frensch, P. A. (Eds.), The Handbook of
Implicit Learning, (pp. 323-364). Thousand Oaks: Sage.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. (3rd
Ed). Academic Press: New York.
Cohen, N. J. & Squire, L. R. (1981). Retrograde amnesia and remote memory
impairment. Neuropsychologia, 19, 337-356.
Corkin, S. (1968). Acquisition of motor skill after bilateral medial temporal-lobe
excision. Neuropsychologia, 6, 255-265.
Curran, T. (1989). Higher-order associative learning in amnesia: Evidence
from the serial reaction time task. Journal of Cognitive Neuroscience, 9, 522-
533.
Curran, T. (1995). On the neural mechanisms of sequence learning. Psyche,
2, 1-34.
Curran, T. (1998). Implicit sequence learning from a cognitive neuroscience
perspective: What, how, where? In Stadler, M. A., & Frensch, P. A. (Eds.), The
Handbook of Implicit Learning, (pp. 365-400). Thousand Oaks: Sage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
261
Curran, T., & Keele, S. W. (1993). Attentional and nonattentional forms of
sequence learning. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 19, 189-202.
Doyon, J. (1997a). Skill learning. In Schmahmann, J. D. (Ed.), The Cerebellum
and Cognition, (pp.273-294). San Diego: Academic Press.
Doyon, J., Gadreau, D., Laforce, R., Castonguay, M., Bedard, P. J., &
Bouchard, J. P. (1997b). Role of the striatum, cerebellum and frontal lobes in
the learning of a visuomotor sequence. Brain and Cognition, 34, 218-245.
Doyon, J., Laforce, R., Bouchard, G., Gadreau, D., Roy, J., Poirier, M.,
Bedard, P. J., Bedard, F., & Bouchard, J. P. (1998). Role of the striatum,
cerebellum and frontal lobes in the automatization of a repeated visuomotor
sequence of movements. Neuropsychologia, 36, 625-641.
Doyon, J., Owen, A. M., Petrides, M., Sziklas, V., & Evans, A. C. (1996).
Functional anatomy of visuomotor skill learning in human subjects examined
with positron emission tomography. European Journal of Neuroscience, 8,
637-648.
Drepper, J., Timmann, D., Kolb, F. P., & Diener, H. C. (1999). Non-motor
associative learning in patients with isolated degenerative cerebellar disease.
Brain, 122, 87-97.
Flament, D., Ellermann, J., Ugurbil, K., & Ebner, T. J. (1994). Functional
magnetic resonance imaging of cerebellar activation while learning to correct
for visuo-motor errors. Society for Neuroscience Abstracts, 20, 20.
Frensch P. A. & Miner, C. S. (1994). Individual differences in short-term
memory capacity on an indirect measure of serial learning. Memory &
Cognition, 22, 95-110.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 2
Frensch P. A. (1998). One concept, multiple meanings: on how to define the
concept of implicit learning. In Stadler, M. A., & Frensch, P. A. (Eds.), The
Handbook of Implicit Learning, (pp. 47-106). Thousand Oaks: Sage.
Frensch, P. A., Lin, J., & Buchner, A. (1998). Learning versus behavioral
expression of the learned: The effects of a secondary tone counting task on
implicit learning in the serial reaction time task. Psychological Research, 61,
83-98.
Frensch, P. A., Wenke, D., & Runger, D. (1999). A secondary tone-counting
task suppresses expression of knowledge in the serial reaction task. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 25, 260-274.
Gabrieli, J. D. E., Brewer, J. B., Desmond., J. E., & Glover, G. H. (1997).
Separate neural bases of two fundamental memory processes in the human
medial temporal lobe. Science, 11, 264-266.
Gaffan, D. (1996). Memory, action and the corpus striatum: current
developments in the memory-habit distinction. Seminars in the Neurosciences,
8, 33-38.
Gao, J., Parsons, L. M., Bower, J. M., Xiong, X., & Fox, P. T. (1996).
Cerebellum implicated in sensory acquisition and discrimination rather than
motor control. Science, 272, 545-547.
Gentile, A. M. (1998). Implicit and explicit processes during acquisition of
functional skills. Scandinavian Journal of Occupational Therapy, 5, 7-16.
Gentilucci, M., Bertolani, L., Benuzzi, F., Et al. (2000). Impaired control of an
action after supplementary motor area lesion: a case study. Neuropsychologia,
38, 1398-1404.
Ghez, C. & Krakauer, J. (2000). The organization of movement. In Kandel, E.
R., Schwartz, J. H., & Jessel, T. M. (Eds.), Principles of Neural Science, (pp.
653-673). New York: McGraw-Hill.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
263
Glisky, E. L., & Schacter, D. L. (1987). Acquisition of domain-specific
knowledge in organic amnesia: Training for computer-related work.
Neuropsychologia, 25, 893-906.
Glisky, E. L., & Schacter, D. L. (1988). Long-term retention of computer
learning by patients with memory disorders. Neuropsychologia, 26, 173-178.
Glisky, E. L., & Schacter, D. L. (1989). Extending the limits of complex learning
in organic amnesia: computer training in a vocational domain.
Neuropsychologia, 27, 107-120.
Goldman-Rakic, P. S. (1992). Working memory and the mind. Scientific
American. 267, 111-117.
Gomez-Beldarrain, M., Garcia-Monco, J. C., Rubio, B., & Pascual-Leone, A.
(1998). Effect of focal cerebellar lesions on procedural learning in the serial
reaction time task. Experimental Brain Research, 120,25-30.
Gomez-Beldarrain, M., Grafman, J., Pascual-Leone, A., & Garcia-Monco, J. C.
(1999). Procedural learning is impaired in patients with prefrontal lesions.
Neurology, 52, 1853-1860.
Grafton, S. T., Hazeltine, E., & Ivry, R. (1995). Functional mapping of
sequence learning in normal humans. Journal of Cognitive Neuroscience, 7,
497-510.
Grafton, S. T., Mazziotta, J. C., Presty, S., Friston, K. J., Frackowiak, S. J., &
Pjelps, M. E. (1992a). Functional anatomy of human procedural learning
determined with regional cerebral blood flow and PET. Journal of
Neuroscience, 12, 2542-2548.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 4
Grafton, S. T., Mazziotta, J. C., Woods, R. P., & Phelps, M. E. (1992b).
Human functional anatomy of visually guided finger movements. Brain, 115,
565-587.
Green, T. D., & Flowers, J. H. (1991). Implicit versus explicit learning
processes in a probabilistic, continuous fine-motor catching task. Journal of
Motor Behavior, 23, 293-300.
Haaland, K. Y., Harrington, D. L., & Grice, J. W. (1993). Effects of aging on
planning and implementing arm movements. Psychology and Aging, 8, 617-
632.
Hanlon, R. E. (1996). Motor learning following unilateral stroke. Archives of
Physical Medicine and Rehabilitation, 77, 811-815.
Hazeltine, E., Grafton, S. T., & Ivry, R. (1997). Attention and stimulus
characteristics determine the locus of motor-sequence encoding: A PET study.
Brain, 120, 123-140.
Honda, M., Deiber, M. P., Ibanez, V., Pascal-Leone, A., Zhuang, P., & Hallett,
M. (1998). Dynamic cortical involvement in implicit and explicit motor
sequence learning. Brain, 121, 2159-2173.
Hoover, J. E. & Strick, P. L. (1999). The organization of cerebellar and basal
ganglia outputs to primary motor cortex as revealed by retrograde
transneuronal transport of herpes simplex virus type 1. The Journal of
Neuroscience, 19, 1446-1463.
Houk, J. C. (1997). On the role of the cerebellum and basal ganglia in
cognitive signal processing. Progress in Brain Research, 114, 543-552.
Hsiao, A. T., & Reber, R. S. (1996). Short (50ms) distracter-target
asynchronies disrupt SRT learning. Paper presented at the 37th Psychonomic
Society Annual Meeting, Chicago.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 5
Ivry, R. B., & Keele, S. W. (1989). Timing functions of the cerebellum. Journal
of Cognitive Neuroscience, 1, 136-152.
Ivry, R. B., Keele, S. W., & Diener, H. C. (1988). Dissociation of the lateral and
medial cerebellum in movement timing and movement execution.
Experimental Brain Research, 73, 167-180.
Jackson, G. M., Jackson, S. R., Harrison, J., Henderson, L., & Kennard, C.
(1995). Serial reaction time learning and Parkinson’s Disease: Evidence fora
procedural learning deficit. Neuropsychologia, 33, 577-593.
Jenkins, I. H., Brooks, D. J., Nixon P. D., Frackowiak, S. J., & Passingham, R.
E. (1994). Motor sequence learning: a study with positron emission
tomography. Journal of Neuroscience, 14, 3775-3790.
Jennings, P. J. (1995). Evidence of incomplete motor programming in
Parkinson’s Disease. Journal of Motor Behavior, 27, 310-324.
Jimenez, L., & Mendez, C. (1999). Which attention is needed for implicit
sequence learning? Journal of Experimental Psychology: Learning, Memory,
and Cognition, 25, 236-259.
Jimenez, L., Mendez, C., & Cleermans, A. (1996). Comparing direct and
indirect measures of implicit learning. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 22, 948-969.
Jueptner, M., Rijntjes, M., Weiller, C., Faiss, J. H., Timmann, D., Mueller, S.
P., & Diener, H. C. (1995). Localization of a cerebellar timing process using
PET. Neurology, 45, 1540-1545.
Jueptner, M., & Weiller, C. (1998). A review of differences between basal
ganglia and cerebellar control of movements as revealed by functional imaging
studies. Brain, 121, 1437-1449.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 6
Kandel, E. R., Kupfermann, I., & Iverson, S. (2000). Learning and Memory. In
Kandel, E.R., Schwartz, J. H., & Jessel, T. M. (Eds.), Principles of Neural
Science, (pp. 1227-1246). New York: McGraw-Hill.
Kakei, S., Hoffman, D.S., & Strick, P.L. (1999). Muscle and movement
representations in the primary motor cortex. Science. 285, 2136-2139.
Kami, A., Meyer, G., Jezzard, P., Adams., Turner, R., & Ungerleider, L.
(1995). Functional MRI evidence for adult motor cortex plasticity during motor
skill learning. Nature, 377,155-158.
Kami, A., Meyer, G., Rey-Hipolito, C., Jezzard, P., Adams, M. M., Turner, R.,
& Ungerleider, L. G. (1998). The acquisition of skilled motor-performance: Fast
and slow experience driven changes in primary motor cortex. Proceedings of
the National Academy of Sciences, 95, 861-868.
Kawashima, R., Roland, P. E., & O’Sullivan, B. T. (1994). Fields in human
motor areas involved in preparation for reaching, actual reaching, and
visuomotor learning: A positron emission tomography study. Journal of
Neuroscience, 14, 3462-3474.
Kermandi, I., Jurquet, Y., Arzi, M., & Joseph, J. P. (1993). Neural activity in the
caudate nucleus of monkeys during spatial sequencing. Experimental Brain
Research, 94, 352-356.
Knopmann, D. S., & Nissan, M. J. (1991). Procedural leaning is impaired in
Huntington’s disease: Evidence from the serial reaction time task.
Neuropsychologia, 29, 245-254.
Knowlton, B. J., Mangles, J. A., & Squire, L. R. (1996). A neostriatial habit
learning system in humans. Science, 273,1399-1402.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 7
Knowlton, B. J., & Squire, L. R. (1995). Remembering and knowing: Two
different expressions of declarative memory. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 21, 699-710.
Lavond, D. G., Steinmetz, J. E., Yokaitis, M., & Thompson, R. F. (1987).
Reacquisition of classical conditioning after removal of cerebral cortex.
Experimental Brain Research, 67, 569-593.
Lee, Y., & Vakoch, D. (1996). Transfer and retention of implicit and explicit
learning. British Journal of Psychology, 87, 637-651.
Lewicki, P., Czyzewska, M., & Hoffman, H. (1987). Unconscious acquisition of
complex procedural knowledge. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 13, 523-530.
Lewicki, P., Hill, T., & Bizot, E. (1988). Acquisition of procedural knowledge
about a pattern of stimuli that cannot be articulated. Cognitive Psychology, 20,
24-37.
Lewis, R. F., & Tamargo, R. J. (2001). Cerebellar lesions impair context-
dependent adaptation of reaching movements in primates. Experimental Brain
Research, DOI 10.1007/s002210100719.
Lu, X., Hikosaka, O., & Miyachi, S. (1998). Role of monkey cerebellar nuclei in
skill for sequential movement. Journal of Neurophysiology, 79, 2245-2254.
Magill, R. A. (1998). Knowledge is more than we can talk about: Implicit
learning in motor skill acquisition. Research Quarterly for Exercise and Sport,
69,104-110.
Matsumoto, N., Hankawa, T., Maki, S., & Graybiel, A. M. (1999). Nigrostriatal
dopamine system in learning to perform sequential motor tasks in a predictive
manner. Journal of Neurophysiology, 82, 978-998.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6 8
Masters, R. S. W. (1992). Knowledge, knerves and know-how: The role of
explicit versus implicit knowledge in the breakdown of a complex motor skill
under pressure. British Journal of Psychology, 83, 343-358.
Mauk, M.D., & Thompson, R. F. (1987). Retention of classically conditioned
eyelid responses following acute decerebration. Brain Research, 403, 89-95.
Mayor, E. A., & Rabbitt, P. M. (1995). Investigating individual differences in a
serial choice reaction time task: Use of auditory feedback and analysis of
responses surrounding errors. Journal of Motor Behavior, 27, 325-332.
McCormick, D. A., Lavond, D. G., Clark, G. A., Kettner, R. E., Rising, C. E., &
Thompson, R. F. (1981). The engram found? Role of the cerebellum in
classical conditioning of the nictitating membrane and eyelid response.
Bulletin Psychonomic Society, 18, 103-105.
Middleton, F. A. & Strick, P. L. (1994). Anatomical evidence for cerebellar and
basal ganglia involvement in higher cognitive function. Science, 266, 458-461.
Middleton, F. A. & Strick, P. L. (2001). Cerebellar projections to the prefrontal
cortex of the primate. The Journal of Neuroscience, 15, 700-712.
Milner, B. (1966). Amnesia following operation on the temporal lobes. In
Whitty, C. M., & Zagwell, O. L. (Eds.), Amnesia, (pp. 103-133). London:
Butterworths.
Milner, B., Corkin, S., &Teuber, H.L. (1968). Further analysis of the
hippocampal amnesic syndrome: 14-year follow-up study of H. M.
Neuropsychologia, 6, 215-234.
Miyachi, S., Hikosaka, O., Miyashita, K., Karadi, Z., & Rand, M. K. (1997).
Differential roles of monkey striatum in learning of sequential hand movement.
Experimental Brain Research, 115,1 -5.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
269
Miyai, I., Blau, A. D., Reding, M.J., & Volpe, B. T. (1997). Patients with stroke
confined to basal ganglia have diminished response to rehabilitation efforts.
Neurology, 48, 95-101.
Miyai, I., Suzuki, T., Kang, J., Kubota, K., & Volpe, B. (1999). Middle cerebral
artery stroke that includes the premotor cortex reduces mobility outcome.
Stroke, 30, 1380-1383.
Molinari, M., Leggio, M. G., Solida, A., Cirra, R., Misciagna, S., Silveri, M. C. et
al. (1997). Cerebellum and procedural learning: Evidence from focal cerebellar
lesions. Brain, 120, 1753-1762.
Mushiake, H., Inase, M., & Tanji, J. (1990). Selective coding of motor
sequence in the supplementary motor area of the monkey cerebral cortex.
Experimental Brain Research, 82, 208-210.
Mushiake, H., Inase, M., & Tanji, J. (1991). Neuronal activity in the primate
premotor, supplementary, and precentral motor cortex during visually guided
and internally determined sequential movements. Journal of Neurophysiology,
66, 705-718.
Mushiake, H., & Strick, P. L. (1995). Pallidal neuron activity during sequential
arm movements. Journal of Neurophysiology. 74, 2754-2758.
Mullebacher, W., Ziemann, U., Boroojerdi, B., Cohen, L., & Hallet, M. (2001).
Role of the human motor cortex in rapid motor learning. Experimental Brain
Research, 136,431-438.
Nakamura, K., Sakai, K., & Hikosaka, O. (1998). Neuronal activity in medial
frontal cortex during learning of sequential procedures. Journal of
Neurophysiology, 80,2671-2687.
Nissan, M. J., & Bullemer, P. (1987). Attentional requirements of learning:
Evidence from performance measures. Cognitive Psychology, 19, 1-32.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 7 0
Nixon, P. D., & Passingham, R. E. (2000). The cerebellum and cognition:
cerebellar lesions impair sequence learning but not conditional visuomotor
learning in monkeys. Neuropsychologia, 38, 1054-1072.
Nixon, P. D., & Passingham, R. E. (2001). Predicting sensory events: the role
of the cerebellum in motor learning. Experimental Brain Research, DOI
10.1007s002210100702.
Onla-or, S. (2001). Motor Skill Learning in Individuals with Parkinson’s
Disease: Considerations of Cognitive and Motor Demands. Unpublished
Doctoral Dissertation, University of Southern California, Los Angeles.
Packard, M. G., & McGaugh J. L. (1992). Double dissociation of fornix and
caudate nucleus lesions on acquisition of two water maze tasks: further
evidence for multiple memory systems. Behavioral Neuroscience, 106, 439-
46.
Parent, A., & Hazarati, L. (1995). Functional anatomy of the basal ganglia.
The cortico-basal ganglia-thalamo-cortical loop. Brain Research Reviews. 20,
91-127.
Pascal-Leone, A., Grafman, J., Clark, K., Stewart, M., Massaquoi, S., Lou, J.
S., & Hallett, M. (1993). Procedural learning in Parkinson’s Disease and
cerebellar degeneration. Annals of Neurology, 34, 594-602.
Pascal-Leone, A., Grafman, J., & Hallett, M. (1994). Modulation of cortical
motor output maps during development of implicit and explicit knowledge.
Science, 263, 1286-1289.
Perruchet, P., & Amorim, M. A. (1992). Conscious knowledge and changes in
performance in sequence learning: Evidence against dissociation. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 18, 785-800.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
271
Pew, R. W. (1974). Levels of analysis in motor control. Brain Research, 71,
393-400.
Platz, T., Denzler, P., Kaden, B., & Mauritz, K. H. (1994). Motor learning after
recovery from hemiparesis. Neuropsychologia, 32, 1209-1223.
Pochon, J., Levy, R., Poline, J., Crazier, S., Lehericy, S., et al. (2001). The
role of dorsolateral prefrontal cortex in the preparation of forthcoming actions:
A fMRI study. Cerebral Cortex, 11, 260-266.
Pohl, P. S., & Winstein, C. J. (1998). Age-related effects on temporal
strategies to speed motor performance. Journal of Aging and Physical Activity,
6, 45-61.
Pohl, P. S., & Winstein, C. J. (1999). Practice effect on the less-affected upper
extremity after stroke. Archives of Physical Medicine and Rehabilitation, 80,
668-
675.
Pratt, J., Chasteen, A. L., & Abrams, R. A. (1994). Rapid aimed limb
movements: Age differences and practice effects in component
submovements. Psychology and Aging, 9, 325-334
Rauch, S. L., Whalen, P. J., Savage, C. R., Curran, T., Kendrick, A., et al.
(1997). Striatal recruitment during an implicit sequence learning task as
measured by functional magnetic resonance imaging. Human Brain Mapping,
5, 124-132.
Rauch, S. L., Whalen, P. J., Curran, T., Mclnernery, S., Heckers, S., &
Savage, C. R. (1998). Thalamic deactivation during early implicit sequence
learning: A functional MRI study. NeuroReport, 9, 865-870.
Reber, A. S. (1976). Implicit learning of synthetic languages: The role of
instructional set. Journal of Experimental Psychology: Human Learning and
Memory, 2, 88-94.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 7 2
Reber, A. S. (1989). Implicit learning and tacit knowledge. Journal of
Experimental Psychology: General, 118, 219-235.
Reber, P. J., & Squire, L. R. (1998). Encapsulation of implicit and explicit
memory in sequence learning. Journal of Cognitive Neuroscience, 10, 248-
263.
Reed, J., & Johnson, P. (1994). Assessing implicit learning with indirect tests:
Determining what is learned about sequence structure. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 20, 585-594.
Rosner, B. (2000). Fundamentals of Biostatistics. United States: Duxbury.
Sakai, K., Hikosaka, H., Miyachi, S., Tajino, R., Sasakai, Y., & Puta, B. (1998).
Transition of brain activation from frontal to parietal areas in visuomotor
sequence learning. Journal of Neuroscience, 18, 1827-1840.
Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984). Knowledge of results
and motor learning: a review and critical appraisal. Psychological Bulletin, 95,
355-386.
Saper, C. B., Iverson, S., & Frackowiak, R. (2000). Integration of sensory and
motor function: The association areas of the cerebral cortex and the cognitive
capabilities of the brain. In Kandel, E. R., Schwartz, J. H., & Jessel, T. M.
(Eds.), Principles of Neural Science, (pp. 349-380). New York: McGraw-Hill.
Schmahmann, J. D., & Sherman, J. C. (1998). The cerebellar cognitive
affective syndrome. Brain, 121, 561-579.
Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral
hippocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 20,
11-21 .
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
273
Segar, C. A. (1994). Implicit learning. Psychological Bulletin, 115, 163-196.
Seitz, R. J., Canavan, A. G. M., Yaguez, L., Hertzog, H., Tellmann, L., Knorr,
U., Huang, Y., Homberg, V. (1994). Successive roles of the cerebellum and
premotor cortices in trajectorial learning. NeuroReport, 5, 2514-2544.
Seitz, R. J., & Roland, P. E. (1992). Learning of sequential finger movements
in man: a combined kinematic and positron emission tomography (PET) study.
European Journal of Neuroscience, 4,154-165.
Selemon, L. D., & Goldman-Rakic, P.S. (1985). Longitudinal topography and
interdigitation of corticostriatal projections in the rhesus monkey. Journal of
Neuroscience, 5, 776-794.
Schmidt, R. A., & Lee, T. D. (1999). Motor Control and Learning: A Behavioral
Emphasis. Champaign: Human Kinetics.
Shea, J. B., Kohl, R., & Indermill, C. (1990). Contextual interference:
Contributions of practice. Acta Psychologica, 73, 145-157.
Shea, J. B., & Kohl, R. (1991). Composition of Practice: Influence on the
retention of motor skills. Research Quarterly for Exercise and Sport. 62, 187-
195.
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, 5,179-187.
Shea, C. H., Wulf, G., Whitacre, C.A., & Park, J. (2001). Surfing the implicit
wave. Quarterly Journal of Experimental Psychology, (in press).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 7 4
Shibasaki, H., Sadato, N., Lyshkow, H., Yonekura, Y., Honda, M. et al. (1993).
Both primary motor cortex and supplementary motor areas play an important
role in complex finger movement. Brain, 116, 1387-1398.
Shimamura, A. P., & Squire, L. R. (1988). Long-term memory in amnesia:
Cued recall, recognition memory, and confidence ratings. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 14, 763-770.
Stadler, M. A. (1989). On learning complex procedural knowledge. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 15, 1061-1069.
Stadler, M. A. (1992). Statistical structure and implicit serial learning. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 18, 318-327.
Stadler, M. A. (1995). Role of attention in implicit learning. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 21, 674-685.
Stadler, M. A., & Roediger, H. L. (1998). The question of awareness in
research on implicit learning. In Stadler, M. A., & Frensch, P. A. (Eds.), The
Handbook of Implicit Learning, (pp. 105-134). Thousand Oaks: Sage.
Stark, C. E., & Squire, L. R. (2000). Functional magnetic resonance imaging
(fMRI) activity in the hippocampal region during recognition memory. Journal
of Neuroscience, 20,7776-7781.
Steinmetz, J. E., Lavond, D. G., Ivkovick, D., Logan, C. G., & Thompson, R. F.
(1992). Disruption of classical eyelid conditioned response after cerebellar
lesion: Damage to a memory trace system or a simple performance deficit.
Journal of Neuroscience, 12, 4403-4426.
Sullivan, K. J. (1998). The role of the sensorimotor cortical system in skill
acquisition and motor learning: A behavioral study. (Doctoral Dissertation,
University of Southern California, 1998). Dissertation Abstracts International,
60-02B, 0815.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 7 5
Suzuki, W. A., Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1993). Lesions
of the perirhinal and parahippocampal cortices in the monkey produce long-
lasting memory impairments in the visual and tactual modalities. Journal of
Neuroscience, 13, 2430-2451.
Squire, L. R. (1986). Mechanisms of memory. Science, 232, 1612-1619.
Squire, L. R. (1987). Memory and Brain. New York: Oxford University Press.
Squire, L. R. (1992). Declarative and nondeclarative memory: Multiple brain
systems supporting learning. Journal of Cognitive Neuroscience, 4, 232-243.
Squire, L. R., & Zola-Morgan, S. (1991). The medial temporal lobe memory
system. Science, 253, 1380-1385.
Suzuki, W. A., Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1993). Lesions
of the perirhinal and parahippocampal cortices in the monkey produce long-
lasting memory impairment in the visual and tactual modalities. Journal of
Neuroscience, 13, 2430-2451.
Thomas, J. R., Salazar, W., & Landers, D. M. (1991). What is missing in
p<.05? Effect size. Research Quarterly for Exercise and Sport, 62, 344-348.
Timmann, D., & Horak, F. B. (1996). Prediction and set-dependent scaling of
early postural responses in cerebellar patients. Brain, 119, 101-111.
Timmann, D., & Horak, F. B. (1998). Perturbed step initiation in cerebellar
subjects: 1. Modifications of postural responses. Experimental Brain
Research, 119, 73-84.
Tinazzi, M., & Zanette, G. (1998). Modulation of ipsilateral motor cortex in man
during unimanual finger movements of different complexities. Neuroscience
Letters, 244, 121-124.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 7 6
Toni, I., Krams, R., Turner, R., & Passingham, R. E. (1998). The time course
of changes during motor sequence learning: a whole-brain fMRI study.
Neuroimage, 8, 50-61.
Vakil, E., Kahan, S., Huberman, M., & Osimani, A. (2000). Motor and non
motor sequence learning in patients with basal ganglia lesions: the case of
serial reaction time (SRT). Neuropsychologia, 38, 1-10.
vanMier, H., Petersen, S. E., Temple, L. W., Permutter, J. S., Synder, A. Z., &
Raichle, M. E. (1994). Practice related changes in a continuous motor task
measured by PET. Society for Neuroscience Abstracts, 20, 361.
vanMier, H., Temple, L. W., Permutter, J. S., Raichle, M. E., & Petersen, S. E.
(1995). Generalization of practice-related effects in motor learning using the
dominant and non-dominant hand measured by PET. Society for
Neuroscience Abstracts, 21, 1441.
vanMier, H., Tempel, L. W., Perlmutter, J. S., Raichle, M. E., & Petersen, S. E.
(1998). Changes in brain activity during motor learning measured with PET:
effects of hand performance and practice. Journal of Neurophysiology, 80,
2177-2199.
Wagner, A. D., Gabrieli, J. D. E., & Verfallie, M. (1997). Building memories:
Remembering and forgetting of verbal experiences as predicted by brain
activity. Science, 281, 1188-1191.
Weiss, P. (1941). Self-differentiation of the basic patterns of coordination.
Comparative Psychology Monographs, 17, 4.
Willingham, D. B., Nissan, M. J., & Bullemer, P. (1989). On the development of
procedural knowledge. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 15, 1047-1060.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 7 7
Willingham, D. B. (1998). Implicit learning and motor skill in older subjects: an
extension of the processing speed theory. In Stadler, M. A., & Frensch, P. A.
(Eds.), The Handbook of Implicit Learning, (pp. 573-594). Thousand Oaks:
Sage.
Willingham, D. B. (1999). The neural basis of motor-skill learning. Current
Directions in Psychological Science, 8, 178-182.
Willingham, D. B., & Goedert-Eschmann, K. (1999). The relationship between
implicit and explicit learning: Evidence for parallel development. Psychological
Science, 6, 531-534.
Winstein, C. J., Merians, A. S., & Sullivan, K. J. (1999). Motor learning after
unilateral brain damage. Neuropsychologia, 37, 975-987.
Wise, S. P. (1996). The role of the basal ganglia in procedural memory.
Seminars in the Neurosciences, 8, 39-46.
Woodruff-Pak, D. S., Lavond, D. G., & Thompson, R. F. (1985). Trace
condition abolished by cerebellar nuclear lesions but not lateral cerebellar
cortex aspirations. Brain Research, 348, 249-260.
Wulf, G., & Schmidt, R. A. (1997). Variability of practice and implicit motor skill
learning. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 23, 987-1006.
Wulf, G., & Weigelt, C. (1997). Instructions about physical principles in
learning a complex motor skill: To tell or not to tell.... Research Quarterly for
Exercise and Sport, 68, 362-367.
Wulf, G., Hob, M., & Prinz, W. (1998). Instructions for motor learning:
Differential effects of internal versus external focus of attention. Journal of
Motor Behavior, 30, 169-179.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
278
Wulf, G., Shea, C. H., & Matschiner, S. (1998). Frequent feedback enhances
complex motor skill learning. Journal of Motor Behavior, 30, 180-192.
Wulf, G., & Shea, C. H. (2001). Principles derived from the study of simple
skills do not generalize to complex skill learning. Psychonomic Bulletin &
Review, (in press).
Zola-Morgan, S., Squire, L. R., Ararmal, D. G., & Suzuki, W. A. (1989).
Lesions of the perirhinal and parahippocampal cortex that spare the amygdala
and hippocampal formation produce severe memory impairment. Journal of
Neuroscience, 12,4355-4370.
Zola-Morgan, S., Squire, R. L., Clower, R. P., & Rempel, N. L. (1993).
Damage to the perirhinal cortex exacerbates memory impairment following
lesions to the hippocampal formation. Journal of Neuroscience, 13, 251-265.
Zola-Morgan, S., Squire, R. L., & Ramus, S. J. (1995). Severity of memory
impairment in monkeys as a function of locus and extent of damage within the
medial temporal lobe memory system. Hippocampus, 4, 483-495.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX 1
Cerebellar Stroke
Explicit Knowledge Group
CB1
CB5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission
280
CB7
No-Explicit Knowledge Group
CB3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
281
APPENDIX 2
Basal Ganglia Stroke
Explicit Knowledge Group
BG4
BG7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
No-Explicit Knowledge Group
BG1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BG10
Reproduced will, permission of the copyright owner. Further reproduction prohibited without permission
APPENDIX 3
Sensorimotor Cortical Area Stroke
Explicit Knowledge Group
SMC1
SMC3
SMC5
SMC6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
285
SMC7
No-Explicit Knowledge Group
SMC2
SMC4
SMC8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The influence of patella alta on knee extensor mechanics and patellofemoral joint stress
PDF
Motor skill learning in individuals with Parkinson's disease: Consideration of cognitive and motor demands
PDF
Quantifying musculoskeletal load and adaptation: Biomechanical consideration
PDF
At what age are gait characteristics mature? Evaluation of gait kinematics, kinetics, and intersegmental dynamics in 7 year-old children
PDF
Bimanual coordination of goal-directed multi-joint rapid aiming movements following stroke
PDF
Biomechanical and neuromuscular aspects of non-contact ACL injuries: The influence of gender, experience and training
PDF
Satellite cell proliferation: two models and two anabolic stimuli
PDF
Mothering occupations: An ethnographic study of Thai women with physical disabilities in the north of Thailand
PDF
Therapeutics and science in the history of occupational therapy
PDF
Cardiovascular fitness, exercise, and metabolic disease risk in overweight Latino youth
PDF
The influence of body weight and composition, pubertal status, tobacco use and exposure, physical activity and muscle strength on bone mass of Chinese adolescents
PDF
Mechanical demand distribution during multi-joint tasks
PDF
Factors contributing to patellofemoral joint stress: a comparison of persons with and without patellofemoral pain
PDF
The role of the sensorimotor cortical system in skill acquisition and motor learning: a behavioral study
PDF
Strategic moral diplomacy in theory and practice: The Lockerbie negotiations, the exile of the Shah of Iran, and Haiti's generals
PDF
The physiological effects of a single bout of eccentric versus concentric resistance exercise
PDF
Empirical evaluation of neuromusculoskeletal models for functional electrical stimulation of the lower limb
PDF
The relationship between family routines and perceived stress in mothers with children with and without a disability
PDF
Control and dynamics during angular impulse generation
PDF
No words necessary: An ethnography of daily activities with young children who don't talk
Asset Metadata
Creator
Boyd, Lara A.
(author)
Core Title
The interaction between explicit knowledge and implicit motor -sequence learning following focal brain damage
Degree
Doctor of Philosophy
Degree Program
Biokinesiology
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
health sciences, rehabilitation and therapy,OAI-PMH Harvest
Language
English
Contributor
Digitized by ProQuest
(provenance)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c16-239079
Unique identifier
UC11339069
Identifier
3093413.pdf (filename),usctheses-c16-239079 (legacy record id)
Legacy Identifier
3093413.pdf
Dmrecord
239079
Document Type
Dissertation
Rights
Boyd, Lara A.
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 au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
health sciences, rehabilitation and therapy