University of Southern California Dissertations and Theses

The role of the sensorimotor cortical system in skill acquisition and motor learning: a behavioral study

(USC Thesis Other)

The role of the sensorimotor cortical system in skill acquisition and motor learning: a behavioral study

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE ROLE OF THE SENSORIMOTOR CORTICAL SYSTEM IN
SKILL ACQUISITION AND MOTOR LEARNING: A BEHAVIORAL STUDY
by
Katherine Josephine Sullivan
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 1998
Copyright Katherine J. Sullivan
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
U N IV E R SIT Y OF SOUTHERN CALIFO RNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
K atherine J . S u lliv a n
under the direction of h.ex ..... Dissertation
Committee, and approved by all its m em bers,
has been presented to and accepted by Vie
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
/
“f r v J /-rSfcz:.................
Dean of Craduafe Stuaies
Date ..........
DISSERTATION COMMITTEE
C -»
c J O L
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEDICATION
To my sons, Paul, Tyler, and Kyle.
I will always be there to support your dreams and aspirations
just as you have supported mine.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iii
ACKNOWLEDGEMENTS
Life is a great adventure or it is nothing.
Helen Keller
To pursue a Doctorate of Philosophy is to embark on a great adventure. Without the
encouragement and support of my family and friends, this segment of my life’s adventure
would never have germinated and come to fruition. My gratitude is embedded in every word of
this document In particular, I would like to thank:
Dr. Carofee Winstein, my teacher and advisor, who is a charismatic, energetic individual and
an outstanding scientist and clinician. It has been an honor to work with her.
My dissertation committee, Dr. Stan Azen, Dr. Nina Bradley, Dr. Helena Chui, and Dr. Luanda
Baker, who have been committed to my growth and development and provided insight and
counsel throughout my studies.
Dr. Helen Hislop, who from the day I sat on her couch during my USC interview, has always
had unending confidence in me. I hope to live up to her expectations.
Beth Fisher, my confidant commiserator, colleague and friend...we have much work to do in
the future.
Chris Powers, my grad school “ buddy”, who marked my developmental milestones and
prodded me to completion.
My fellow USC students and labmates, Sompom Onia-or, Jody Carmack, Lara Boyd, and Pat
Pohl, with whom as much informal learning has taken place as that which has come from
classes.
The patients, therapists, and my colleagues at Sharp Rehabilitation Center, San Diego,
California, particularly Ann Koeneke, Sue Wagner, Lisa Barton, and Julie Peters. The work we
did at Sharp inspired my desire to understand how individuals with brain injury learn and
recover.
My parents, Bob and Judy Sullivan, who have always been there for me. My Dad in particular
is an inspiration. He quit his job at 50 with 3 of 5 children in college to pursue a dream. He
continually demonstrates that it is never too late to achieve something that is important to you.
My brothers and sisters. Bob Sullivan, Jr., Mary Kay Schwanke, Tim Sullivan and Beth Cimler.
As with all adventures, life has its ups and downs, but through it all I never doubted that my
siblings were there with unfailing love and support A special commendation goes to my
brother Bob, who in addition to being a great brother, is also a computer programmer. The
data analysis program he wrote for me was instrumental in revealing the key findings of this
dissertation.
Finally, my husband Tommy Causey, who came in on the tail-end of this adventure but luckily
for me, will be there for the remainder.
This dissertation was partially supported by grants from the Foundation for Physical Therapy
and the Patrida Leahy Memorial Scholarship from the Neurology Section, American Physical
Therapy Association.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
IV
List of Tables
List of Figures
Abstract
Chapter 1
Chapter 2
Chapter 3
Chapter 4
OVERVIEW AND PRESENTATION OF
EXPERIMENTS
NEURAL BASIS OF MOTOR SKILL ACQUISITION
DISSOCIABLE LEARNING SYSTEMS IN THE BRAIN
BEHAVIORAL BASIS OF SKILL ACQUISITION AND
MOTOR LEARNING
v
vi
xi
1
32
43
Chapter 5 DEFICITS IN MOTOR PROGRAM ACQUISITION AND
EXECUTION BUT NOT MOTOR LEARNING AFTER
UNILATERAL SENSORIMOTOR CORTICAL SYSTEM
DAMAGE
55
Chapter 6 DEFICITS IN PARAMETER CONTROL: SPATIAL AND
TEMPORAL COUPLING AFTER UNILATERAL
SENSORIMOTOR CORTICAL SYSTEM DAMAGE
116
Chapter 7 PREDICTING UPPER EXTREMITY IMPAIRMENT
AND PHYSICAL DISABILITY AFTER MIDDLE
CEREBRAL ARTERY STROKE
154
BIBLIOGRAPHY 176
APPENDIX 189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLES
Chapter 5:
Table 1. Clinical and neurologic characteristics of stroke participants.
Table 2. Group comparisons (excluding motor).
Table 3. Group by hand upper limb motor comparisons.
Table 4. Performance and learning effects indicated by change in block
mean RMSEs by Group between acquisition and retention phases.
Table 5. Mean (+SE) for kinematic variables by group across practice
blocks.
Table 6. Group by arm-used mean (+SE) residual RMSE (deg).
Table 7. Group by Arm-used mean (+SE) number of discontinuous
movements.
Chapter 6:
Table 1. Mean (+SE) temporal and amplitude scaling factors by Group for
the acquisition and retention phases.
Chapter 7:
Table 1. Disablement Model: Nagi Scheme
Table 2. Descriptive statistics for outcome variables.
Table 3. Univariate correlation of motor impairment (upper extremity
Fugl-Meyer motor score) with lesion extent
Table 4. Mean upper extremity Fugl-Meyer score (+SD) by lesion location.
Table 5. Stepwise multivariate analysis of motor impairment (upper
extremity Fugl-Meyer score) with pathology.
Table 6. Univariate correlations of physical disability (SF-36) with model
variables.
Table 7. Stepwise multivariate analysis of physical disability (SF-36) with
model constructs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FIGURES
Chapter 5:
Figure 1. Lesions of the stroke participants. The greatest extent of each
lesion is displayed using the functional atlas of Damasio and Oamasio
(1989). The location of the central sulcus (filled arrow) or Sylvian fissure
(unfilled arrow) is indicated. Lesion data not available for L4.
Figure 2. Experimental set-up with arm lever and feedback display. A)
Lower part shows the subject with lever and feedback monitor. Upper part
shows an overhead view of the subject with arm lever. B) An example of
the augmented visual feedback with a graphic representation of the
subject’s movement trajectory for that trial superimposed over the target
pattern and the RMSEr score, horizontal movement of the lever produced
a change in voltage corresponding to the lever's angular displacement
Figure 3. Schematic of the pre- and post-movement events. Top panel
(A) depicts the pre-movement time events. The lower panel (B) depicts the
response and post-response events for the FB trials. For the no-FB trials,
the post-response interval was displayed with no FB for 4500 ms. Note
that time is not to scale.
Figure 4. Example of movement trajectory scaling. The goal movement is
displayed as the thick black line. The subject’s actual movement is
displayed as the thin black line (tic marks indicate peak displacement for
reversals). RMSEs is the difference, in degrees, between the goal pattern
and subject’s movement The rRMSE is the difference in degrees between
the goal pattern and the subject's trajectory after the subject’s trajectory is
scaled in time (dashed line) and then amplitude (dot-dashed line).
Figure 5. Displacement-, velocity-, and acceleration-time profiles for the
goal movement The temporal locations of landmarks a (movement
onset), b (1st peak acceleration), c (2nd peak acceleration), d (3rd peak
acceleration), e (4th peak acceleration), f (5th peak acceleration), and g
(movement offset) are indicated. The times of these respective landmarks
are displayed on the x axis. The average velocity for the goal movement is
175 deg/sec.
Figure 6. Control and stroke group RMSEs block means for acquisition
(Blocks 1-20), no-FB retention (Blocks 21-22), and FB retention (Blocks
23-34) phases. A 30-min break occurred between blocks 10 and 11.
RMSEs represents overall accuracy.
Figure 7. Control and stroke group residual RMSE block means for the
acquisition (Blocks 1-20) phase. A 30-min break occurred between blocks
10 and 11. After rescaling in time and amplitude to reduce
parameterization errors, the residual RMSE represents the accuracy of the
motor program. Group effect p=.01; Block effect p<-0001; Group by block
interaction, p=. 02).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 8. Individual subject trial data. Rows show data from a control (top)
and two stroke subjects (bottom) during early (left) and late (right) practice.
The goal movement is the thick black line. The subject’s actual movement
(thin black line), time-scaled movement (dash line), and amplitude-scaled
movement (dot-dash line) are displayed. The trial error scores for the
unsealed (RMSEs) and scaled (rRMSE) trajectories are in the table.
Figure 9. Displacement-, velocity-, and acceleration-time profiles for a
control subject (JA). Each panel displays 10-trials from the A) beginning
(Trials 1-10), B) middle (Trials 91-100), and C) end (Trials 191-200) of
practice. Movement onset is synchronized at 0 ms. The vertical marker on
the right indicates the end of movement for each trial.
Figure 10. Displacement-, velocity-, and acceleration-time profiles for a
stroke subject (R4). Each panel displays 10-trials from the A) beginning
(Trials 1-10), B) middle (Trials 91-100), and C) end (Trials 191-200) of
practice. Movement onset is synchronized at 0 ms. The vertical marker on
the right indicates the end of movement for each trial.
Figure 11. Frequency histogram of the total number of discontinuous
movements for each group during 1) Trials 1-50, 2) Trials 51-100, 3) Trials
101-150, and 4) Trials 151-200. Group effect p<001; Block effect
p<.0001; Group x Block interaction, p<.01. The asterisk (*) indicates the
locus of the interaction effect
Figure 12. Averaged within-subject correlations (+SE) between the start of
movement (a) and kinematic landmarks (b through f) and the end of
movement (g) and kinematic landmarks (b through f) for the Control (left)
and Stroke (right) groups during the beginning (Trials 1-10), middle (Trials
91-100), and end (Trials 191-200) of practice. The asterisk (*) indicates
intervals between correlation pairs greater than .20. The average
correlation for each group of comparisons is indicated
Figure 13. Bar graph of the mean residual RMSE (+SE) by group for
acquisition blocks 1) Trials 1-100 and 2) Trials 101-200. For both blocks,
the group by hand interaction was significant (p<.05). Post-hoc
comparisons indicated that the locus of the effect was a statistically
significant difference between the left control and stroke groups (*, p<01)
but not between the right control and stroke groups (ns).
Figure 14. Bar graph of the mean residual RMSE (+SE) during acquisition
by lesion location compared to that of controls. Group effect p < 01. The
subcorti'cal (SC) and cortical +/- SC groups were not statistically different
from each other (p=.64) but both were significantly greater than controls
(subcortical, p=. 02; cortical+/-SC, p<.01).
Figure 15. Frequency histogram of the total number of discontinuous
movements for each group by arm-used. Group effect p<-001; Group x
Hand interaction, p=.0568. Post-hoc comparisons indicate that the locus of
the effect was a statistically significant difference between the right control
and stroke groups (*, p<001) but not between the left control and stroke
groups (ns).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 16. Bar graph of the mean number (+SE) of discontinuous
movements during acquisition by lesion location compared to that of
controls. Group effect, p=.01. The subcortical(SC) and cortical+/-SC
groups were not statistically different from each other (p=.81) but both were
significantly greater than controls (p< 01). The difference between the
subject with pontine infarct and controls was not statistically significant
(p=. 67).
Figure 17. Averaged within-subject correlations (+SE) between the start
of movement (a) and kinematic landmarks (b through f) and the end of
movement (g) and kinematic landmarks (b through f) for the Right control
(left) and Right stroke (right) groups during the beginning (Trials 1-10),
middle (Trials 91-100), and end (Trials 191-200) of practice. The asterisk
O indicates intervals between correlation pairs greater than .20. The
average correlation for each group of comparisons is indicated.
Figure 18. Averaged within-subject correlations (+SE) between the start
of movement (a) and kinematic landmarks (b through f) and the end of
movement (g) and kinematic landmarks (b through f) for the Left control
(left) and Left stroke (right) groups during the beginning (Trials 1-10),
middle (Trials 91-100), and end (Trials 191-200) of practice. The asterisk
O indicates intervals between correlation pairs greater than .20. The
average correlation for each group of comparisons is indicated.
Chapter 6:
Figure 1. Example of movement trajectory scaling. The goal movement is
displayed as the thick black line. The subjects actual movement is
displayed as the thin black line (tic marks indicate peak displacement for
reversals). RMSEs is the difference, in degrees, between the goal pattern
and subjects movement TF and AF are calculated after the subjects
trajectory is scaled in time (dashed line) and then amplitude (dot-dashed
line).
Figure 2. Block means for acquisition (Blocks 1-20), no-FB retention
(Blocks 21-22), and FB retention (Blocks 23-24) for A) TF comparison
between control and stroke groups, and B) AF comparison between control
and stroke groups. Horizontal line at 1.00 indicates movement goal.
Figure 3. Individual subject trial data. Rows show data from a control (top)
and stroke subject (bottom) during early (left) and late (right) practice. The
goal movement is the thick black line. RMSEs subjects actual
movement (thin black line), TF from the time scaled movement (dash line),
and AF from the amplitude scaled movement (dot-dash line) is displayed in
the table for each trial. Note different time scales between trials.
Figure 4 . Change in block mean TF and AF scaling for the acquisition
phase (Blocks 1-20). Correlation coefficient (r) and normalized correlation
coefficient (z‘ score) for the control group (A) and stroke group (B) are
indicated.
Figure 5. Individual trial TF and AF data across the acquisition phase
(Blocks 1-20) for two representative control (left) and stroke subjects
(right). Horizontal line at 1.00 indicates movement goal.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 6. Bar graph of mean z’ scores (normalized correlation coefficient)
and SE for the temporal and spatial scaling correlations by lesion location
compared to control group. The correlation coefficient for each group is
indicated. Group effect, p<.01. The subcortical (SC) and cortical ±SC (
Cx±SC) group were not statistically different from each other (p=. 16) but
both were significantly greater than controls (SC, p=. 04; Cx±SC, p<001).
The difference between the subject with pontine infarct and controls was
not statistically significant (p=.60).
Figure 7. Scatter plot of grip strength by temporal and spatial coupling
correlation (normalized correlation coefficient) for control (triangle) and
stroke (circle) subjects.
Figure 8. Young and older group RMSEs block means for acquisition
(Blocks 1-20), no-FB retention (Blocks 21-22), and FB retention (Blocks
23-34) phases.
Figure 9. Change in block mean TF and AF scaling for the acquisition
phase (Blocks 1-20). Correlation for the young group (A) and older group
(B) are indicated. Horizontal line at 1.00 indicates movement goal.
Figure 10. Individual stroke subjects block RMSEs scores for acquistion 1
(Blocks 1-20), no-FB retention (Blocks 21-22), FB retention (Blocks 23-24),
acquisition 2 (Blocks 25-44), no-FB retention (Blocks 45-46), and FB
retention (Blocks 47-48). Regression lines have been added to illustrate
the direction of performance changes across phases.
Figure 11. Individual stroke subject TF and AF data for acquisition 1
(Blocks 1-20) and acquisition 2 (Blocks 21-40). The correlation coefficient
of time and amplitude factors for each acquistion phase is indicated.
Figure 12. Individual trial data for subject 12 during extended practice.
The goal movement is the thick black line. RMSEs for tbe subject’s actual
movement (thin black line), TF from the time scaled movement (dash line),
and AF from the amplitude scaled movement (dot-dash line) is displayed in
the table for each trial.
Figure 13. Individual trial data for subject R4 during extended practice.
The goal movement is the thick black line. RMSE for the subject's actual
movement (thin black line), TF from the time scaled movement (dash line),
and AF from the amplitude scaled movement (dot-dash line) is displayed in
the table for each trial.
Chapter 7:
Figure 1. Upper extremity (UE) Fugl-Meyer scores of the contralateral limb
by percent lesion volume (r = -.59, p <.01). Symbols designate individual
lesion location. The horizontal line indicates a score of 33 out of a
maximum score of 66.
Figure 2. Upper extremity (UE) Fugl-Meyer scores of the contralateral limb
by SF-36 score (r = 56, p=.01).
ix
134
138
141
142
144
145
147
148
166
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3. Contralateral Box & Blocks score by SF-36 score (r = .49, p =
.03).
Figure 4. Functional independence measure (FIM) by SF-36 score (r =
.72, p < 001).
Figure 5. Pyramidal tract origin and course. Adapted from: (Young &
Young, 1997).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
XI
ABSTRACT
The sensorimotor cortical system, which includes the primary sensorimotor cortex and
its first-order afferent and efferent projections, has a well established role in the planning and
execution of fast and accurate movements. It is less clear how this neuroanatomic network
contributes to the central processes that influence movement control from those that influence
motor learning. This dissertation investigated the role of the sensorimotor cortical system in
the control and learning of a rapid, upper limb movement that is controlled by a single motor
program with specified temporal and spatial parameter goals. Twenty right-handed adults with
unilateral sensorimotor cortical system stroke and 20 matched-controls practiced an upper
limb movement that consisted of 3 flexion-extension reversals and a 1000 ms duration goal for
200 trials on one day and returned the next day for retention tests. Stroke participants used
the limb ipsilateral to the lesion. Kinematic analyses quantified changes in motor control with
practice (i.e., motor program accuracy, proportion of temporal and amplitude error).
Throughout the first 100 trials of practice, the stroke group was less proficient in motor
programming compared to the control group; however, by the end of practice both groups had
acquired programmed movements and retained the movements to a comparable degree one
day later. Stroke participants benefitted from practice as indicated by faster and more
accurate movements across practice and they retained these control gains at retention.
However, subjects with stroke performed with greater parameter error in acquisition and
retention phases such that movements were slower, and hypermetric compared to control
subjects. Hemispheric differences were evident in motor program but not parameter control.
Temporal and spatial scaling were coupled in the stroke group that was not remediated with
extended practice.
The findings suggest that the sensorimotor cortical system has a role in the acquisition
and control of motor programs and the parameterization of these programs to meet explicit
environmental goals. However, retention results suggest that the sensorimotor cortical system
has a much smaller role in the longer-term learning of programmed actions.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
CHAPTER I:
Overview and Presentation of Results
Introduction
Motor skill acquisition of a rapid movement involves two distinct behavioral outcomes.
One is associated with the ability to produce fast and accurate movements. The other involves
the ability to ret3in this skill for subsequent action or to generalize this skill to novel actions. A
functional neuroanatomic model for the control of reaching and grasping proposed by Jeannerod,
Arbib, Rizzolati, and Sakata (1995) suggests that distinct neuroanatomic networks subserve these
two behavioral outcomes. The pragmatic system functions to “ extract parameters relevant to
action, and to generate the corresponding motor commands ” (Jeannerod et al., 1995, p.320).
The semantic system functions to take movement attributes related to action and produce a
unique precept In other words, semantic system processing provides meaning and relevance to
an action. This distinction suggests that separate neuroanatomic systems, with differing modes of
processing and cognitive function, can interact to subserve complex behavior.
The pragmatic and semantic system distinction provides a neuroanatomic and
behavioral basis for the study of motor control and learning processes. The overarching aim of
this dissertation is to investigate the role of the sensorimotor cortical system in skill acquisition and
motor learning. For the purposes of this dissertation, the sensorimotor cortical system is a
neuroanatomic network that includes the primary sensorimotor cortex and its first-order afferent
and efferent projections. Chapter 2 provides evidence that control program construction for
sensory-cued rapid arm movements occurs primarily in the sensorimotor cortical system. The
construction of the control program results from central processing within this neural network,
critical for the evolving sensory and motor schema that leads to skilled motor behavior. It is
hypothesized that the pragmatic system is involved in the control of rapid limb movements and
most likely includes the sensorimotor cortical areas. The sensorimotor cortical system is richly
connected with basal ganglia, thalamic, and cerebellar circuits. Chapter 3 provides evidence that
dissociable learning systems exist within the brain. The implicit or motor learning system is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
subserved by a cortical-striatal-cerebeliar network. It is hypothesized that this system can be
functionally associated with abstract rule learning; a semantic mode of processing.
The interconnections between the pragmatic system (i.e. sensorimotor cortical system
and its role in action-related movement attributes) and the semantic system (i.e., the motor
learning system and its role in unconscious movement knowledge) suggest that these two
systems are integrated to produce the greatest capability for skilled motor performance. It is not
exactly clear how these two systems interact to affect skill acquisition; however, it seems evident
that the pragmatic and semantic systems are neuroanatomically connected in such a way that
they can influence and be influenced by each other. Therefore, Chapter 4 will discuss behavioral
approaches that distinguish between changes in motor performance associated with motor control
processes from those that infer motor learning.
Presentation of Results
This dissertation includes three chapters that discuss the effects of sensorimotor cortical
system damage on motor control and learning. Chapters 5 and 6 address specific control and/or
learning processes. In addition. Chapter 7 will discuss the impact of middle cerebral artery stroke
on ipsilateral and contralateral upper limb impairments, functional limitations, and physical
disability.
Chapter 5: Deficits in motor program acquisition and execution but not motor learning after
unilateral sensorimotor cortical system damage.
Major Purpose: The purpose of this experiment is to investigate the contribution of the
sensorimotor cortical system in the control and learning of a rapid upper limb
movement with specific temporal and spatial goals that is controlled by a single
motor program.
Hypotheses: 1. The capability for motor learning will be similar between groups with and
without sensorimotor cortical system damage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
2. Individuals with sensorimotor cortical system damage will benefit from practice
in the acquisition of a goal-directed, rapid upper limb movement; however, motor
control deficits in motor program acquisition and execution will be evident
Independent variables; Group (control, stroke). Arm-used (right left), Trial Blocks (acquisition.
retention)
Dependent variables: Measures of overall accuracy (root mean square error), motor program
accuracy (residual root mean square error), movement time, average
movement velocity, number of discontinuous movements, and unit
structure.
Chapter 6: Deficits in parameter control; spatial and temporal coupling after unilateral
sensorimotor cortical system damage.
Experiment 1:
Major Purpose: The purpose of this experiment is to investigate the contribution of the
sensorimotor cortical system in the parameter control of a goal-directed rapid
upper limb movement
Hypotheses: Individuals with sensorimotor cortical system damage will benefit from practice in
terms of temporal and spatial parameter scaling; however, motor control deficits
in absolute error and parameter control strategies will be evident
Independent variables: Group (control, stroke), Arm-used (right, left), Trial Blocks (acquisition,
retention)
Dependent variables: Measures of proportionate temporal and spatial accuracy.
Correlation of temporal and scaling factors across trials.
Experiment 2:
Major Purpose: The purpose of this experiment is to further investigate if parameter control
movement strategies are related to sensorimotor cortical damage by comparing
the performance of two neurologically healthy groups.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
Hypotheses: If parameter control movement strategies are not different between neurologically
healthy young and older subjects, then there should be no difference in the
coupling of temporal and amplitude scaling parameters between the two groups.
Independent variables: Group (young, older), Trial blocks (acquisition, retention)
Dependent variables: Measures of proportionate temporal and spatial accuracy.
Correlation of temporal and scaling factors across trials.
Experiment 3:
Major Purpose: The purpose of this experiment is to determine if sensorimotor cortical system
damage results in persistent deficits in parameter control strategies or if control
strategies can be remediated with extended practice.
Hypotheses: If parameter control strategies are a reflection of relatively persistent deficits due
to sensorimotor cortical system damage, then there should be no change in
parameter control strategies with extended practice.
Independent variable: Trial Blocks
Dependent variables: Measures of proportionate temporal and spatial accuracy.
Correlation of temporal and scaling factors across trials.
Chapter 7: Predicting upper extremity motor impairment and physical disability after middle
cerebral artery stroke.
Major Purpose: To investigate the predictive validity of a disablement model that includes
pathologic, impairment and functional limitation measures on physical disability
outcome.
Hypotheses: A physical disablement model that includes measures of pathology, physical
impairment and functional limitation will explain a moderate to moderately-high
proportion of the variance in physical disability.
Independent variables: Lesion location and size, upper limb dexterity, strength, and motor control
measures, and level of functional assistance in motor self-care skills.
Dependent variable: Physical disability as measured by the SF-36.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
CHAPTER 2
Neural Basis of Motor Skill Acquisition
Introduction
Skilled performers are able to produce movements which are fast and accurate. One
generally accepted perspective is that fast movements are movements which are executed
predominately independent of sensory feedback and are controlled via a motor program
(Jeannerod, 1988). The motor program perspective is limited because it does not account for the
transformation between the sensory stimulus to move and the action plan (i.e., stimulus-response
transformation) that evolves over practice. Arbib (1981) has proposed that solutions to a
movement problem evolve as perceptual and motor schema combine to form a control program.
Perceptual schema are involved in perceptual encoding occurring prior to or during the action;
motor schema are movement control units specified during a movement or retrieved for upcoming
movements (Jeannerod et al., 1995). According to Arbib's schema theory, schema are
strengthened as the optimal solution or control program is constructed. Arbib’s schema theory
provides a perspective for examining the course of skill acquisition (i.e., improvements in
movement speed and accuracy with practice) as schema are constructed through a recursive
process until a “ foundation of neural localization is attained’ (Jeannerod et al., 1995, p. 316 ).
Neurophysiologic studies with primates and positron emission tomography (PET) studies
with humans provide support for the recursive nature of control program construction during motor
skill acquisition. It is well established through extensive study of the brain that the initiation and
execution of a sensory cued movement involves different cortical and subcortical brain areas
which become active at different times relative to movement onset (Johnson, 1992; Kalaska &
Crammond, 1992). The cortical areas in general and the sensorimotor cortical areas in particular
interact with other subcortical structures such as the cerebellum and basal ganglia in the control
and learning of motor skills (Grafton, Mazziotta, Presty et al., 1992). However, this chapter will
focus on the cortical areas activated during motor skill acquisition since it is not clear how
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
changes within cortical areas contribute to the incremental improvements in movement speed and
accuracy that occur over the course of motor skill acquisition.
The interaction of different cortical areas contributes to the construction and execution of
a control program (Grafton, Mazziotta, Presty et al., 1992). Both neurophysiologic studies with
primates and imaging studies with humans indicate that the cortical activation patterns associated
with sensory cued programmed movements actually changes over the course of skill acquisition.
During early training, there is diffuse cortical activation which is hierarchically and serially ordered
(Kuboto & Komatsu, 1985; Sasaki & Gemba, 1981,1982). As skilled performance develops in
later training, cortical activation becomes more localized (Aizawa, Inase, Mushiake, Shima, &
Tanji, 1991; Haier etal., 1992; Jenkins, Brooks, Nixon, Frackowiak & Passingham, 1994; Remy,
Zilbovicius, Leroy-Willig, Syrota, & Samson, 1994; Kuboto & Komatsu, 1985; Mushiake, Inase, &
Tanji, 1990; Sasaki & Gemba, 1981,1982; Seitz, Roland, Bohm, Greitz, & Stone-Elander, 1990).
These findings suggest that the neural system involved in motor skill acquisition is a distributed
hierarchy or heterarchy (Kalaska & Crammond, 1992) where changes in the processing of
information occur throughout the course of skill acquisition (i.e., acquiring programmed
movements). In addition, these observations support Arbib's theoretical posit of multiple
functional areas involved during schema formation, and hence, control program construction.
The neuroanatomic substrate involved in the control of fast and accurate movements is a
distributed hierarchy involving cortical, subcortical, and cerebellar structures. The emphasis of
this review is to investigate the literature concerning the cortical areas involved in movement
control since there is strong converging evidence from both neurophysiologic studies with
primates and imaging studies with humans that motor, sensory, and prefrontal cortices have a
functionally significant role in the acquisition and execution of control programs. The activation of
these anatomically disparate cerebral areas during movement has led to a growing consensus
that a distributed cortical control network participates in motor control, particularly for actions that
involve motor program development and execution (Colebatch, Deiber, Passingham, Friston, &
Frackowiak, 1991; Donoghue & Sanes, 1994; Doyon, Owen, Petiides, Sziklas, & Evans, 1996;
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
Grafton, Fagg, Woods, & Arbib, 1996; Grafton, Mazziota, Woods, & Phelps, 1992; Johnson, 1992;
Kalaska & Crammond, 1992; Sadato, Ibanez, Deiber, & Haliett, 1996; Winstein, Grafton, & Pohl,
1997).
The cortical areas activated in the motor control of movements initiated in response to a
sensory stimulus (visual or auditory) primarily include the following: the parietal association cortex
(PAC), supplementary motor area (SMA), premotor area (PM), and primary motor cortex (Ml).
Visual or auditory cues are processed in the primary and association visual and auditory cortices,
respectively. The ability to develop sensorimotor associations is supported by the function of
prefrontal areas (PF). Together these structures form a serially ordered cortical network which
ultimately terminates in Ml where the final control command is mediated. However, multiple
afferent and efferent connections exist between these spatially distinct areas supporting the
capability for parallel cortical processing; a level of processing used in the later phases of skill
acquisition (Kuboto & Komatsu, 1985; Sasaki & Gemba, 1981; Sasaki & Gemba, 1982). These
cortical structures contribute to the evolving perceptual and motor schema which subserve the
pragmatic system’s role in movement control.
There has been substantial investigation into the role of the SMA and PM areas in motor
programming particularly in the moments immediately preceding movement execution (Aizawa et
al., 1991; Brinkman & Porter, 1979; Fu, Suarez, & Ebner, 1993; Fu, Flament, Coltz, & Ebner,
1995; Halsband & Passingham, 1985; Mitz, Godschalk, & Wise, 1991; Mushiake et al., 1990;
Passingham, 1988; Shibasaki etal., 1993; Tanji, Okano, & Sato, 1988; Weinrich & Wise, 1982;
Wise, 1985; Wise & Mauritz, 1985); however, there has been little investigation into the role the
sensorimotor cortex may play in the sensorimotor associations evolving in the early stages of
control program construction and how this may affect the process of motor skill acquisition. One
recent exception, using primates with unilateral somatosensory cortex lesions, demonstrated that
the contralateral somatosensory cortex has an essential role in the acquisition of new motor skills
but not the retention of previously learned skills (Pavlides, Miyashia, & Asanuma, 1993). In
addition, several human studies have demonstrated persistent motor control deficits in speed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
accuracy, and consistency in well-practiced upper limb movements in subjects with unilateral
stroke (presumed to involve predominately the sensorimotor cortical system) performing with the
limb ipsilateral to the brain damage (Pohl & Winstein, 1998a; Winstein, Merians, & Sullivan.
1998). These studies raise questions about the role of the sensorimotor cortical system and the
specific contributions of each hemisphere in acquiring skilled movements.
Sensorimotor Cortical System: Activation Pattern
Changes during Skill Acquisition
The purpose of the literature review in this Chapter is to describe the sensorimotor
cortical system and the processing changes which occur within this system over the course of skill
acquisition. This section will review the following areas; 1) the primary cortical response areas in
the distributed sensorimotor cortical system, 2) cortical response pattern changes over the course
of motor learning in primates, 3) cortical response pattern changes over the course of motor
learning in humans, and 4) changes in motor learning capabilities which occur with cortical
lesions.
Anatomical Cortical Substrate for Sensorv-cued Arm Movement
The motor control of sensory-cued (visual or auditory) arm movements involves a
distributed cerebral control network (Colebatch et al., 1991; Grafton et al., 1996; Grafton,
Mazziota, Woods etal., 1992; Johnson, 1992; Kalaska & Crammond, 1992) which primarily
includes the SMA, PM, Ml, and the parietal association cortex. Recognition of visual or auditory
cues is processed in the primary and association visual and auditory cortices, respectively. The
distributed cortical control of sensory-cued movements is elucidated by examining the anatomic
location, major afferent and efferent connections, and function of each cortical area .
Furthermore, a basic understanding of these cortical structures and their functions provides a
reference for examining changes in activation patterns which occur with motor learning.
Supplementary motor area
The SMA is located within the frontal cortex (medial Brodmann's area 6), ventral and
medial to the precentral gyrus, dorsal to area 9, and extends medially to the cingulate sulcus; this
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
area is also referred to as the mesial frontal cortex. The major afferents to SMA include
subcortical connections from the ventrolateral nucleus (VL) of the thalamus, basal ganglia
(particularly the globus pallidus), and cerebellum via thalamic nuclei (X, VL). The major
corticocortical inputs are both ipsilateral (prefrontal (PF), areas 1 and 2 of the primary
somatosensory cortex, PM, Ml, area 5 of posterior parietal cortex, and contralateral PM and SMA
(Brinkman & Porter, 1979; Johnson, 1992). The SMA is somatotopically organized with its main
efferent projection to ipsilateral Ml. Other corticocortical projections include the frontal eye fields,
somatosensory association areas (SSA), and contralateral Ml (Johnson, 1992). The SMA can
influence motor responses via its direct ipsilateral projections to the medullary reticular formation,
parvocellular red nucleus, and spinal cord (Brinkman & Porter, 1979; Tanji, Taniguchi, & Saga,
1980).
The SMA is a convergent center receiving inputs ipsilaterally and contralaterally from both
cortical and subcortical structures (Brinkman & Porter, 1979). Its heavy connections to Ml make it
a gateway, whereby, substantial inputs from throughout the nervous system can influence motor
control (Johnson, 1992). Specifically, the SMA controls movement based on internal information
and has a major role in the following functions: 1) the initiation of movement based on internal
rather than external cues (Deiber el al., 1991; Johnson, 1992; Tanji etal., 1980); however, in a
recent imaging study Remy et al. (1994) demonstrated SMA activation in response to an
auditorily-cued finger movement, 2) the execution of complex rather than simple repetitive
movements (Rao et al., 1993; Remy et al., 1994; Shibasaki et al., 1993), and 3) acts in the
sequencing of movements especially movements with increased temporal complexity (Grafton,
Mazziota, Woods etal., 1992; Mushiake etal., 1990).
Premotor cortex
Premotor cortex is located in the ventral and lateral aspects of the precentral gyrus (the
most lateral portion of Brodmann’s area 6), dorsal to area 8. The PM receives its major afferents
from the deep cerebellar nuclei via nucleus X of the thalamus, the amygdaloid complex, visual
and auditory areas, and area 7 of the parietal cortex (Johnson, 1992; Wise, 1985). These
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 0
connections are noteworthy in distinguishing PM from other motor cortical areas since SMA and
Ml do not receive direct afferent input from these areas. Other inputs are received from PF areas
and the VL nucleus of the thalamus. Information flow from PF and sensory cortices is mediated
through PM to Ml (Johnson, 1992). The major efferent projections are cortical (Ml, SMA,
cingulate sulcus) and subcortical (parvocellular red nucleus, medullary reticular formation,
striatum, basilar pontine nuclei) (Wise, 1985). Unlike SMA and Ml, PM does not have direct
projections to the spinal cord.
The PM cortex is an area of the motor cortex that has exclusive connections with
structures known to be involved with motor control. The substantial afferent connections from
sensory cortices suggests that the PM has a role in the preparation for and sensory guidance of
movement (Wise, 1985). Specifically, PM has a major role in the following aspects of movement
control: 1) the generation or selection of motor commands based on visual or auditory
environmental stimuli (Johnson, 1992; Weinrich & Wise, 1982; Wise & Mauritz, 1985), 2) the
preparation for or planning of upcoming movements (i.e., motor set) (Weinrich & Wise, 1982;
Wise & Mauritz, 1985), 3) recalling and retrieving the appropriate action for a given situation
based upon a sensory cue (i.e., motor program retrieval) (Halsband & Freund, 1990; Halsband &
Passingham, 1985; Mitz etal., 1991; Passingham, 1985) and 4) the control of direction and
distance movement parameters (Fu et al., 1993; Fu et al., 1995)
Primary motor cortex
The Ml is somatotopically organized and located in the precentral gyrus and anterior
bank of the central sulcus, extending medially to the cingulate sulcus (Brodmann's area 4).
Corticocortical afferents come from SMA, PM, arcuate premotor area, cingulate sulcus, and
parietal areas 1, 2, and 5. Thalamo-cortical projections include information from the deep
cerebellar nuclei and the globus pallidus. Ml has efferent projections to all cortical areas
projecting to it and to subcortical areas (parvocellular and magnocellular red nucleus, reticular
system, pontine nuclei, spinal cord) (Johnson, 1992).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 1
Primary motor cortex is involved in the somatotopically organized execution of
movements. Being the site where the final motor command is issued. Ml has a major role in the
following functions: 1) the scaling of specific movement parameters (i.e., force, direction,
distance) (Fu et al., 1993; Fu et al., 1995; Georgopolous, Taira, & Lukashin, 1993; Jeannerod,
1988; Johnson, 1992), 2) the somatotopically organized execution of different motor plans
(Grafton, Mazziota, Woods et al., 1992; Sanes & Donoghue, 1992; Sanes, Donoghue, Thangaraj,
Edelman, & Warach, 1995; Tanji et al., 1988), and 3) bilateral Ml and primary sensory cortex
activate together in the preparation and execution of complex sequential movements (Rao et al.,
1993; Remy et al., 1994; Sadato et al., 1996; Shibasaki et al., 1993; Tanji et al.. 1988).
Parietal cortex
The posterior parietal cortex (PP) has extensive connections with the motor cortex
contributing to higher level motor control of movement. The two distinct areas of PP involved in
cortical movement control include the following: area 5, located in the posterior border of the
superior parietal lobule extending posterolaterally from the posterior border of area 2 to the inferior
aspect of the intraparietal sulcus, and area 7, comprising the entire inferior parietal lobule
(Johnson, 1992). Area 5 receives afferents from the sensory thalamic nuclei, primary sensory
cortex (SI) particularly area 2, frontal areas 4 and 6, and parietal area 7. Major efferents from
area 5 are cortical (area 7, and major projections to SMA) and subcortical (red nucleus, pons,
spinal cord). The major afferents to area 7 are from the thalamus (lateral posterior nucleus,
pulvinar). Area 7 efferents include reciprocal connections with SI, secondary sensory cortex (Sll),
area 5, SMA, PM, strong visual connections via occipital, temporal and frontal visual areas, and
subcortical projections to pons, striatum, pulvinar, and superior colliculus.
The posterior parietal areas have a role in voluntary movements which have increased
spatial complexity (Grafton, Mazziota, Woods et al., 1992), providing an analysis and
representation of external spatial relationships (Johnson, 1992). In addition, PP contributes to the
movement selection process evident by the pattern of activity in areas 5 and 7 which parallel area
6 (Deiberet al., 1991).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 2
Summary
In summary, movements are controlled through a cortical network distributed across
several cortical areas. The cortical areas (SMA, PM, Ml, PP) described have primary
responsibilities contributing to movement control. However, each of these areas has extensive
afferent and efferent connections between several cortical and subcortical sites. Kalaska and
Crammond (1992) suggest that the response properties of various cortical areas do not function
in a strict serial hierarchy but contribute to movement control in a distributed hierarchy or
heterarchy. They state, "In a heterarchy, there is no fixed chain of command or direction of
information flow, instead, the flow of information is flexible and dependent upon the context of an
event (p. 1520)." The following sections will examine changes in the flow of information which
occur throughout the course of motor learning (acquiring programmed movements).
Sinole-cell Recording and Cortical Field Potential Studies in Primates
Single-cell recording and cortical field potential studies in non-human primates provide
specific information about the relationship between cortical activation patterns and behavior. In
addition, invasive studies can provide information about the order of activation in time not
available with imaging technology. While primate studies have provided substantial information
about motor cortex function, almost all primate studies have focused on neuronal activation
patterns in monkeys who are well-trained in a motor task. Learning-dependent changes identified
in invasive studies of monkeys provide the following advantages: 1) specific spatial and temporal
detail can be measured, and 2) monkeys learn at a slower rate than humans making it easier to
identify the neurophysiologic and behavioral relationships which evolve over the course of skill
acquisition. Evidence from monkey studies which have investigated learning-dependent changes
indicate that the use of programmed movements only occurs once the monkey has developed an
understanding of the motor task’s behavioral requirements (Brooks, Kennedy, & Ross, 1983).
Behavioral performance changes as motor skill acquisition evolves from undertrained to well-
trained states are associated with changes in both cortical activation patterns (Sasaki & Gemba,
1981,1982) and neuronal activation patterns (Kubota & Komatsu, 1985).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
The relationship between behavioral understanding and the use of programmed
movements was investigated by Brooks et al. (1983). Monkeys were trained to perform a step-
tracking task which involved moving a handle back and forth in a horizontal arc to land between
two target zones. Over several weeks of training, 40,000 movements were collected between 6
monkeys. Behavioral performance indicators included the following: readiness to move within
prescribed time limits, direction selection, reaction time, and dwell time in target Programming
indicators were derived from kinematic analysis. Movements were considered programmed if
there was a continuous velocity profile with both acceleration and deceleration phases. Accuracy
(movement to the target without under- or overshoots) was also measured. Each training session
included 100-200 movements. The results of this study indicated that the monkeys reached a
steady-state for accuracy by the 5th training session but the use of programmed movements did
not increase until the behavioral indicators had been learned to about 50% accuracy. The authors
concluded that use of programmed movements only occurred when the animal had a general
understanding or schemata for the task, indicated by the improvement in behavioral performance
measures. The results of this study suggest that there is a relationship between improvements in
motor performance and the use of programmed movements. This study used kinematic analysis
and motor performance measures to investigate motor behavior and programmed movement
control but does not provide information regarding the neurophysiologic substrate which may
participate in the development of the motor program.
In studies conducted by Sasaki and Gemba (1981, 1982), neuronal cell recordings were
used to determine the cortical areas which participate in learned behavioral associations and the
neural substrate for improvements in motor skill. By using cortical field potential recordings in 10
adult monkeys, Sasaki and Gemba (1981) found that behavioral changes associated with learning
processes were related to changes in premovement potentials. The visually initiated hand
movement used in this study was lifting a lever using wrist extension in response to a light
stimulus. Electrodes were implanted after the monkey was able to lift the lever to a juice reward
without any instructional stimulus. Chronically implanted electrodes recorded premovement
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
potentials from several cortical areas (PF association areas 8 and 10, PM, forelimb Ml, visual
association area 19, and primary visual area 17) throughout the stimulus-response learning
process. Training occurred over 60 days resulting in 4 behaviorally distinct stages.
In the beginning of training (Stage I), small potentials were present in PF association,
primary visual, and visual association areas. After a few weeks of training (Stage II), potentials in
these areas were larger in both duration and amplitude. The monkey was beginning to recognize
that the light stimulus might be something meaningful but there was no strong recognition relating
movement to reward. During Stage III, the monkey responded to the light stimulus with
movement Potentials in PM and Ml became predominate and potentials in frontal and occipital
areas were larger. The frequency of successful movements was increasing within this stage but
reaction times were still long (900 ms). Stage IV was characterized by shortened reaction times
and movements which were more skillful (fast and accurate). Premovement potentials in Stage
IV were present bilaterally in PF, PM, and occipital cortices. Interestingly, latencies for all cortical
areas were not substantially different (35-47 ms) indicating that all cortical areas responded to the
visual stimulus in parallel after the monkey had learned the relationship between sensory stimulus
and movement response.
In a continuation of this work, Sasaki and Gemba (1982) concluded that cortical field
potentials develop and change across the motor skill acquisition continuum which distinguishes
early from later stages of learning. Premovement potentials in PF cortex and visual association
areas always precede the learning of the stimulus-conditioned movement Sasaki and Gemba
observed that "visual stimuli are first conveyed only to the primary visual cortex, and are delivered
in succession to the association cortices, the premotor cortex and then to the motor cortex along
stages l-lll of the learning process (p.436)." Therefore, in the early stages of learning the PF and
visual association cortices are primarily involved in learning the meaning and relationships
between stimulus, movement and reward. Once these relationships are learned, then rapid
behavioral performance changes (speed, accuracy) occur in the later stages of learning. In the
later stages of learning, these cortical areas are activated almost simultaneously in response to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
the visual stimulus indicating a parallel processing within the cortical distribution network involved
in learned movements.
The findings of Sasaki and Gemba corroborate the observations of Brooks et al (1983);
skillful, rapid (programmed), movements depend on a behavioral understanding of the task
situation. This behavioral understanding is accomplished through a sequential activation of
cortical areas as learning occurs. The slower rate of learning in monkeys makes it easier to
observe the distinguishing features correlating behavioral performance with cortical activation
patterns during the various skill acquisition stages. The findings of Sasaki and Gemba (1981.
1982) provide support for the hypothesis that early stages of motor learning are characterized by a
hierarchical, serial process which evolves in the later stages of motor learning to a parallel
process.
Behavioral performance changes are also associated with changes in neuronal
activations patterns as the monkey's level of skill develops from undertrained to well-trained
states. Kuboto and Komatsu (1985) investigated the relationship between behavior indicators and
neuronal activation patterns in monkeys during the learning of a GO/NO-GO precued upper
extremity lever press task. Single-cell recordings of neurons in dorsolateral and ventral PF cortex
were recorded in 6 adult macaque monkeys divided into three groups: undertrained (less than
60% correct performance level), intermediately trained (60-80% correct performance level), and
well-trained (at least 85% correct performance level). Task-related neuronal firing rate was
related to the level of training. The fewest number of task-related neurons were found during the
undertrained state. Task-related neurons increased with training (highest frequency in the
intermediately trained state) but decreased by approximately 25% in the well-trained state.
Reaction time was longer and more variable for undertrained monkeys (400-800 ms) compared to
well-trained monkeys (300-400 ms). As behavioral performance measures improved (reaction
time and accuracy) neuronal activation patterns changed from low levels of activation when
undertrained, progressing to peak levels of activation during intermediate training levels. The
highest level of skilled performance was associated with a steady state in discrete neuronal
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 6
populations within PF areas. One interpretation is that the highest level of neuronal activation in
the intermediately trained state may be the stage of processing where multiple cortical areas are
activated in an attempt to produce the correct response. Therefore, as training progresses,
relationships between discrete groups of neurons develop and function as the control programs
proposed by Arbib (1981).
Very few monkey studies have examined the cortical changes which evolve with skill
acquisition; however, several studies have proposed functional roles for distinct cortical areas
associated with the learning of sensory-cued movements. Learned sensorimotor associations
appear to occur between the primary and association cortices and the PF cortex; however, SMA
and PM may have differing functional roles distinguishing one stimulus-induced movement from
another. There is consistent agreement that instruction-induced changes occur in SMA
(associated with program specification) and PM (associated with program retrieval) in the
premovement period (Tanji et al., 1988; Wise & Mauritz, 1985). SMA has a well accepted role in
the performance of sequentially organized movements particularly when the movements depend
on remembered (i.e., learned) sequences (Mushiake et al., 1990). PM is involved in retrieving
actions for a current situation based on learned information between situations and actions
(Passingham, 1988). However, the activity in PM only becomes task-related when the monkey
has learned a sensorimotor association (Mitz et al., 1991). In addition, the premovement, set-
related activity in SMA and PM may contribute to the decreases in reaction time associated with
well-learned performance. The efferent subcortical connections from SMA and PM may keep
rubrospinal and corticospinal neurons near firing thresholds resulting in fester reaction times with
learned sensory stimulus associations (Weinrich & Wise, 1982). The development of prepared
responsiveness associated with skilled performance in primates appears to be related to neuronal
activity changes within SMA and PM (Tanji et al., 1980).
A recent study by Nudo, Milliken, Jenkins, and Merznich (1996), investigated the plastic
changes in Ml that result from motor skill learning. Six adult monkeys were trained in two motor
tasks; one involving skilled digit use (pellet retrieval), the other skilled forearm movements (turning
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
a latch with forearm pronation-supination). Using intracortical microstimulation techniques. Ml
movement representations were compared before and after training. For both tasks, motor cortex
representations increased with skilled motor performance. These findings demonstrate that use-
dependent reorganization of the motor cortex occurs with motor skill learning. Clearly, motor
cortical areas have a major role in motor skill acquisition.
In summary, the primate studies provide evidence that skilled performance is associated
with changes in cortical activation. The major results from the primate studies indicate that
behavioral and neurophysiologic findings are related in the following manner 1) a behavioral
understanding of the task (i.e., association between stimulus and response) must develop before
programmed movements are used, 2) cortical processes are hierarchical and serially organized
in early training but these processes are controlled in parallel as skilled performance develops, 3)
in early stages of learning there is diffuse cortical activation across areas that becomes more
discrete in later learning stages, and 4) sensory and PF cortices appear to be the areas where
sensorimotor associations develop during early skill acquisition; SMA and PM appear to be the
areas where motor program selection and retrieval occur during highly skilled movements, once
sensorimotor associations are learned. Together these findings provide a neuroanatomic basis
supporting the behavioral observations that skill progresses from more closed-loop, feedback
driven responses (those associated with the activation of PF and sensory cortices) to more open-
ioop, programmed responses (those associated with activation of PM, SMA, and Ml). In the later
stages of motor skill learning, Ml enhancement suggests a primary role of this area in the
mediation of fast and accurate movements.
Cortical Response Patterns in Humans
Technological advances have resulted in noninvasive tools used to monitor brain activity
in humans. Synaptic activity within neuronal populations can be monitored through regional
cerebral blood flow (rCBF) measurements collected during positron emission tomography (PET)
or functional magnetic resonance imaging (fMRI). Imaging technology provides good spatial
resolution but is limited in temporal resolution since blood flow measurements are integrated over
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 8
a collection period (80-120 sec). Since onset activity cannot be measured, PET and fMRI studies
cannot provide information about the hierarchical activation of cortical areas within a collection
period. Despite the limitation in temporal resolution, serial PET and fMRI studies provide
substantial information about the anatomic structures involved in human motor task execution and
motor learning. In addition, transcranial magnetic stimulation (TMS) (another noninvasive
procedure with less risk than PET or fMRI) can determine cortical maps of motor output and
changes which occur over many practice trials. The findings from human cortical activation
studies during motor learning parallel those from primate studies: 1) cortical activation patterns
change over the stages of motor learning (Grafton, Mazziotta, Presty et al., 1992; Haier et al.,
1992; Jenkins et al., 1994; Kami et al., 1995; Remy et al., 1994; Seitz et al., 1990), 2) the
acquisition and optimization of the motor program occurs predominately in cortical structures
(Grafton, Mazziotta, Presty, etal., 1992; Haier eta!., 1992; Jenkins etal., 1994; Kawashima,
Roland, & O'Sullivan, 1994a; Remy et al., 1994) and is influenced by individual differences
(Schlaug, Knorr, & Seitz, 1994), and 3) there is an association between behavioral findings
(implicit and explicit knowledge) and cortical activation during motor learning (Pascual-Leone,
Grafman, Hallet, 1994).
Several imaging studies have been conducted distinguishing the anatomic structures and
activation patterns associated with motor learning. In a pursuit-tracking task, Grafton, Mazziotta,
Presty et al. (1992) found movement execution occurred through activation of a widely distributed
neural network including bilateral Ml and SMA, contralateral putamen and midbrain, and ipsilateral
cerebellum and visual association areas. After twenty-eight 20 sec practice trials, movements
became more continuous with fewer corrections and increased time on target. These behavioral
changes were accompanied by increases in rCBF of the contralateral SMA, Ml and pulvinar. Due
to the limited amount of practice which occurred between serial scans in this PET study, the
cortical activation responses can only be associated with the early stages of skill acquisition. In
the early stages of skill acquisition, there is increased cortical activity within a subset of the
distributed cortical motor network.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
The effects of more extended training have been investigated in a thumb-finger
movement sequence comparing initial learning after instruction and performance after 100
minutes of practice (Seitz et al., 1990), a key press sequence comparing a 90 min learned
sequence and the acquisition of a new sequence (Jenkins et al.. 1994), and a complex
visuospatial task (the video game Tetris) comparing initial learning and performance after
practicing 30-45 min/day. 5 times per week, for 10 weeks (Haier et al., 1992). In all of these
studies, there was marked improvement in performance (decreased errors, decreased response
time, increased speed, and/or increased scores). Early stages of learning were related to diffuse
areas of activation. Significant decreases in regional cerebral blood flow with practice were
present in each of these studies and consistently included the following areas: prefrontal cortex,
primary somatosensory areas, and primary and association visual cortices. Haier et al. (1992)
found a decrease in Ml with extended practice in contrast to Seitz et al. (1990) who found an
increase in Ml with practice. However, the study conducted by Seitz et al. (1990) had substantially
less practice and did not control for the increase in movement rate which could have resulted in
higher Ml activity (Grafton et al., 1996). The behavioral performance indicator for the Haier et al.
(1992) study was an average score over the 50 days of practice. It is interesting to note that
scoring higher in the Tetris game requires becoming more accurate and fasten therefore,
movement rate should have been increasing with increasing scores. The decrease in Ml activity
over extended practice, despite expected increases in movement rate, is most likely related to
more efficient and discrete activation of specific neuronal populations.
In summary, findings from the human imaging studies presented are consistent with
those from primate studies. During early stages of learning, there is diffuse activation within the
cortical motor network which becomes more discreet with later stages of learning. The early
stage of learning in humans includes activation in the prefrontal and sensory cortices. These
activation patterns are consistent with those observed in primates during the learning of
visuomotor associations. As movement becomes more skilled (fast, accurate), discrete neuronal
populations control the selection and execution of the control program.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
Human studies provide insight into functional areas involved in the development
selection, modification, and execution of the control program. Serial PET and fMRI studies
indicate that the acquisition and modification of the control program (behaviorally associated with
improvement in motor performance) is related to activity in cortical and subcortical regions
(Grafton, Mazziotta, Presty et al., 1992). The diffuse activation in the motor cortical network,
particularly in prefrontal and sensory association areas, indicate that these areas are involved in
generating new responses for trial and error analysis. Jenkins et al. (1994) observed that
activation in prefrontal cortex and anterior cingulate was only present in early learning stages.
These authors propose that the prefrontal areas have a role in generating new responses while
the anterior cingulate area is involved in selective attention. During the first attempts at
movement, the subject tries many different cognitive strategies until a set strategy associated with
improved performance develops. This preferred strategy results in fewer neurons and/or neuronal
brain circuits being used (Haier et al., 1992). It appears that prefrontal areas may have a key role
in the generation of new responses (control program development).
The SMA has an important role in programming movements particularly complex
movement sequences. However, the role of SMA in the learning process is less understood.
Remy et al. (1994) investigated the relationship between activation of motor cortical areas during
simple hand movements with different movement or task-related dimensions. PET images were
collected from 10 healthy males during the following conditions: rest, self-paced all finger flexion-
extension, auditorily-cued all finger flexion-extension, and self-paced one finger (ring finger)
flexion-extension. Instructions about the motor task were given prior to collection, therefore, there
were no practice trials before initial movement attempts. Movements were paced to the subjects
own "self-paced" frequency so that complexity and number of movements were consistent in all
conditions. Cortical areas studied were Ml, SMA, PM, upper anterior cingulate, and area 7.
Compared to rest, significant increases in activation included contralateral Ml for the self-paced all
finger condition, contralateral Ml and SMA for the auditorily-cued all finger condition, and bilateral
Ml and SMA for the self-paced one finger condition. The results of this study indicated the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
following: 1) SMA is involved in initiating movements to external sensory cues, and 2) SMA and
Ml participate in the early stages of motor learning. The all finger movement is an overleamed
task compared to the more complex and unusual movement involved in isolating the middle
finger. Activation of SMA in this new learning phase is consistent with the findings of Grafton,
Mazziotta, Presty et al. (1992) who also found increased SMA activation across all trials in early
skill acquisition. SMA appears to have a role in motor planning (i.e., acquisition of the control
program) particularly in the early stages of motor learning. Remy et al. (1994) did not find
significant activation of PM or cingulate cortex; however, the motor task did not involve any
performance goals related to speed or accuracy. This observation can be substantiated by noting
differences in afferent input to PM and SMA. Unlike SMA and Ml, PM is the only frontal motor
area receiving direct projections from the amygdala (with known function in learning, memory, and
motivational state). Movement tasks with specified behavioral performance outcomes may
involve PM more than movements without specified performance goals.
Two studies (Jenkins etal., 1994; Kawashima , Rowland, & O’Sullivan, 1994b) present
evidence that PM and Ml may be the sites for initial program acquisition, with SMA playing more
of a role in modification or program execution as learning progresses. Jenkins et al. (1994)
collected PET measurements of rCBF in 12 male subjects during each of the following conditions:
rest, performing a preleamed (approximately 90 min practice) 8-sequence key press task, and
learning a new sequence of key presses. Six subjects were scanned high in order to ensure
coverage of all motor cortical areas. Six subjects were scanned low in order to ensure coverage
of subcortical motor areas (e.g., basal ganglia, thalamus, cerebellum). The subjects had
practiced the preleamed sequence until they could perform with no errors and repeated that
performance for ten 35 min trials. For each new sequence scan, a different sequence was
presented. The subjects learned the new sequences by the same process they had learned the
preleamed sequence. The subjects pressed a key to a 3 sec pacing tone. If the key press was
correct they received a high-pitch tone; if incorrect a low-pitched tone. There was greater
activation in PM during new sequence learning compared to the preleamed sequence. In
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
contrast, SMA activation was greater during the preleamed sequence than new learning. Ml was
activated contralaterally for both new and preleamed sequences. The new learning condition in
this study involved learning the association between the movement and feedback tone. The
higher activation of PM during this phase suggests a role for PM in establishing 'what* to do. The
higher activation of SMA during the preleamed or automatic phase indicates a role for SMA in the
execution of a programmed sequence.
These findings were corroborated in a study by Kawashima et ai. (1994b). This study was
designed to get temporal activation information from PET by collecting measurements in
preparation and reaching phases. The task involved looking at seven circular targets on a screen
and memorizing their location by size. After a 90 sec delay period, the subjects were given a point
command and repeated the movement sequence for 180 sec. Twenty male subjects were
divided into two groups. The preparation group had PET collection during the 90 sec delay
period. The reaching group had PET collection in the 20 seconds before the reach and continued
for 170 sec during the reach. Each group practiced the target movement pattern 25 times.
Regional cerebral blood flow measurements were obtained at an initial learning phase (3rd trial)
and later learning phase (25th trial). PM and Ml were active in both preparation and reaching
phases. There were learning related increases in premovement activity in both PM and Ml
(activity was greater in the 25th trial compared to the 3rd). SMA activity was recorded in the
premovement phase but there were no learning-dependent changes. The activation patterns in
PM and Ml in the delay period indicate that these areas may be involved in the planning of the
upcoming reach. These results need to be interpreted with caution since the time from stimulus
(target pattern) to response (replicating the reaching pattern) occurred over a 90 sec delay.
Obviously, memorization processes were involved. However, it is interesting to note that the
learning dependent changes occurred in both PM and Ml.
A study by Schlaug et al. (1994) investigated the relationship between individual
differences in task performance and cortical activation patterns. These authors present an
important observation: PET studies usually involve group analysis which biases rCBF changes to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
those areas with high magnitude or large spatial overlap among subjects. Therefore, individual
differences during skill acquisition can be obscured. The movement task in this study involved
learning a complicated thumb-finger movement sequence (used in previous PET studies),
however, a new PET analysis method (individual response identification statistic) was used to
contrast individual from group differences. Regional CBF during PET was collected in nine
healthy males after initial instruction (subjects carefully instructed in what to do) and following 1 hr
of training (performance phase). Group analysis indicated that subjects improved performance
with training (increased rate, decreased finger switch latency, decreased faults per sequence).
Significant rCBF increases for left Ml and left PM were present for both initial and practiced
phases. Group learning-dependent cortical activation changes included increases (left
putamen/globus pallidus, left hand area) and decreases (right superior and anterior parietal, right
Broca's areas). Behaviorally, individual differences were very disparate for both speed (%
increase in rate: range = 8-125%, mean = 89%) and accuracy (fault change %: range = 67%
decrease in faults to 1400% increase in faults). Contralateral Ml and PM were activated in all
subjects in the performance phase, and in 3 of 9 subjects in the initial phase. However, there was
substantial inter-subject variability in cortical activation areas that was related to an individual's
behavioral performance. Subjects with more accurate performance (fewer faults per sequence)
had greater activation in SMA, cingulate areas, Broca's area, and somatosensory areas. Subjects
with the greatest improvement in speed were associated with predominant activation increases in
primary and secondary motor areas. The results from this study indicate that Ml and PM are
common essential cerebral activation areas for learning a motor skill, however, individual task
performance is related to patterns of activation depending on whether the individual chooses a
strategy emphasizing speed and/or accuracy. This study contributes two significant findings: 1)
there are substantial individual differences between behavioral motor performance and associated
cortical activation patterns, and 2) despite these differences, Ml and PM appear to be consistent
cortical activation areas for establishing the control program during early motor learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
All of the human imaging studies have relatively low numbers of practice trials; therefore,
the cortical activation patterns described are associated with early stages of learning. Learning
related changes in Ml appear to contribute to program acquisition and modification in concert with
either SMA (Grafton, Mazziotta, Presty et al., 1992; Remy et al., 1994) or PM (Jenkins et al.,
1994; Kawashima et al., 1994). It appears that both PM and SMA have consistently varying
roles for motor program acquisition and modification during early skill acquisition. PM contributes
to motor program development related to learning a specified performance goal. SMA appears to
be involved in the execution of learned motor sequences once sensorimotor associations are
established.
The low number of practice trials is one of the major limitations in the PET and fMRI
human motor skill acquisition studies. During early stages of skill acquisition, patterns of cortical
activation in humans parallels those observed in primates. Little comparison can be made
regarding the performance of humans compared to primates in later stages of skill acquisition.
However, two studies specifically examined Ml activation patterns after extended practice: 1)
Pascual-Leone et al. (1994) using TMS, and 2) Kami et al., 1995 using fMRI.
Pascual-Leone et al. (1994) investigated the learning of a stimulus-response motor task
and changes in cortical motor output maps. In particular, these authors were interested in the
relationship between implicit and explicit knowledge, motor performance, and motor cortical
activation. A serial reaction time task (SRTT) was used. Subjects sat in front of a computer with
the right fingers resting on a 4 button response pad. The 'go' signal consisted of a number display
(1-4) instructing the subject to push the corresponding finger button. Ten healthy subjects were
assigned to one of two groups. The control group received randomly presented go signals. The
test group received a sequence of 12 cues repeated 10 times within a practice block (120 trials).
At the end of each block, subjects were asked if the sequence was random or repeating. If the
subject answered repeating, they had to report the sequence. Correctly reporting the sequence
indicated that the subjects had acquired explicit knowledge of the task. TMS was used to monitor
the cortical motor outputs of the muscles involved in the task. Subjects completed 12 blocks
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
resulting in 1440 practice trials. Groups did not differ during the first practice block for baseline
cortical motor output or the performance indicator (RT: control = 329.4, test = 327.5). The control
group remained stable in cortical output and performance across blocks. In contrast the test
group demonstrated progressively shorter RTs and larger areas of cortical motor output Explicit
knowledge (verbally reporting the entire sequence accurately) was acquired in all subjects
between blocks 6 to 9 with continued enlargement of the cortical output maps up to this time.
After acquiring explicit knowledge, the area of cortical output maps decreased to baseline but
RTs continued to shorten (at block 4, RT = 120.6 ms; at block 12, RT = 55.1 ms). The results of
this study support findings observed in the later learning stages with primates: 1) acquiring
explicit knowledge resulted in the most efficient and effective use of programmed movements,
and 2) early training corresponded to diffuse cortical activation which became more localized with
extended practice.
Kami et al. (1995) found changes in Ml cortical activation patterns during fMRI between
early and late practice, attributed to a switch in Ml processing modes. According to these authors,
repeating a finger-to-thumb opposition movement sequence resulted in a within session change in
activation from low to high levels. These authors suggest that this mode of Ml processing is
associated with “ fast-learning” related to the acquisition of task-relevant motor routines. After 4
weeks of 15-20 minutes of daily practice, subjects demonstrated between session changes. This
“ slow-leaming” was associated with a more extensive representation in Ml and suggests that
experience-dependent reorganization of MI cortex occurs related to long-term motor skill memory.
Behaviorally, these cortical changes were concurrent with asymptotic performance in speed and
accuracy. In this study, analysis was restricted to Ml; therefore, overall changes in the
sensorimotor cortical system outside this area were not reported. However, consistent with
primate work (Nudo et al., 1996), skilled motor performance in humans is accompanied b y
dynamic, experience-dependent changes in motor cortex.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 6
Lesion Induced Changes in Cortical Activation
Lesion studies provide another line of evidence relating cortical activation patterns with
changes in behavioral performance. After cortical lesions, there are changes in functional
performance (Halsband & Freund, 1990; Halsband & Passingham, 1985) and cortical activation
patterns (Aizawa et al., 1991; Chollett et al., 1991) associated with learned motor skills. In
addition, a recent study has demonstrated that the somatosensory projections to Ml play an
important role in teaming new motor skills but a lesion to these projections does not interfere with
the retention of previously learned motor skills (Pavlides et al., 1993).
Halsband and Passingham (1985) trained six monkeys in a visual conditional motor task
(blue panel, pull handle; yellow panel, turn handle) to receive a food reward. Animals were trained
for 50 trials/dy, 7 days /wk to reach 180 correct responses out of 200 trials. Four animals were
lesioned (2 in PM, 2 in PF areas) and 2 were not lesioned. Testing was restarted within 5 wks of
reaching criterion for lesioned and non-lesioned animals. All animals learned the stimulus-
movement response within 1500 trials (mean = 929, range = 667-1503). After the 5 wks of no
practice, the non-lesioned monkeys relearned the task to criterion within 98 trials. The PF
lesioned monkeys relearned the task to criterion in 73 and 105 trials. The PM lesioned monkeys
did not relearn the task after 1000 trials of additional practice. The authors concluded that PM is
involved in selecting movements based on a conditional stimulus. The monkeys were not
impaired in vision or movement but movement selection based on visual cues. It is interesting to
note the PF lesioned animals were not affected in their capability to relearn the task. Since PF
areas have a role in developing sensorimotor associations, it appears that the learned
associations were already present and incorporated into a control program. When PM is lesioned
there is a disruption in the selection of the control program for correct behavioral performance.
The deficiencies in control program selection observed by Halsband and Passingham
(1985) was demonstrated in a human study involving individuals with PM/SMA (n =12), Ml (n = 2),
and parietal (n = 17) lesions and healthy controls (n = 30) (Halsband & Freund, 1990). All
subjects were able to discriminate primary visual, tactile, and auditory stimuli in a pretest.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
Subjects were trained to make 6 arm movements in response to 6 different sensory cues.
Passing was making 18/18 correct stimulus-movement identifications. Subjects failed if they
could not reach criterion performance within 250 trials. Individuals with PM/SMA lesions made
more errors than the other groups in selecting a movement based on a visual, auditory, or tactile
cues. In addition, the number of trials to reach criterion performance was significantly higher in
the PM/SMA lesion group. Subjects in this group did not reach criterion within 250 practice trials
and could not select correct movements to sensory cues (6/12 for visual, 5/12 for tactile, 7/12 for
auditory). The results from this study provide additional evidence that PM/SMA is involved in the
selection of movements based on sensory cues. There is also a distinct motor learning function
for PM/SMA since half of the PM/SMA lesioned subjects never reached criterion performance.
The pattern of increased cortical activation was noted after cortical lesions in both
primates (Aizawa et al., 1991) and humans (Chollett et al., 1991). Aizawa et al. (1991) trained a
monkey to perform a simple key press to a visual-trigger signal using both right and left digits.
The animal had learned the task within 2 months but training continued for 12 months to ensure
that the monkey had overleamed the task. Single-cell activity was recorded in SMA and Ml. SMA
activity is usually present in the premovement period, however, after overlearning this simple task
there was no activity recorded in SMA. Task-related activity was present in Ml. At this point right
Ml was lesioned. Left digit paresis resulted but the monkey returned to pre-lesion performance
levels after 21 days recovery. SMA neurons were very active in the premovement period after Ml
lesion. This study indicates two important findings: 1) in an overleamed simple movement
premovement SMA activity is diminished; overtraining resulted in a direct activation of Ml in
response to the visual stimulus, and 2) SMA must have a role in the development and execution
of a control program in early and intermediate motor skill acquisition stages but once the program
is developed and overleamed SMA activity is not needed. The change in SMA activity over the
course of motor learning is similar to the changes observed in PF areas. PF areas are active in
early skill acquisition contributing to sensorimotor associations. Once sensorimotor associations
are learned PF activity ceases. Concordantly, SMA areas are active in program development and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 8
execution. Once programs are well-established, the sensory cue activates the program and Ml
directly making SMA activity redundant Conceivably, SMA is required to reestablish motor
programs damaged after cortical lesions since this area appears to have a function in the
development and execution of programs in normal skill acquisition processes.
Increased cortical activation occurs when fingers of the recovered hand are moved in
individuals with stroke (Chollett et at, 1991). Regional CBF measurements with PET, were
collected in 6 individuals with unilateral stroke who had contralateral upper extremity paresis but
now demonstrated full motoric recovery. Comparisons were made between rest, finger
movements of the normal hand, and finger movements of the previously paralyzed hand. The
task was a sequential thumb-finger opposition movement Normal finger movement resulted in
activation of Ml, SMA, PM, and PP (contralaterally), and cerebellum (ipsilaterally). Movements of
the recovered hand resulted in bilateral activation of all areas. This study does not provide any
specific role delineation for the cortical areas activated. Obviously, damage to functionally
significant areas in the motor cortical network would disrupt the execution and reestablishment of
efficient and effective motor programs. Diffuse activation after cortical lesion indicates that the
recovery process involves a more distributed cortical network functioning to reestablish deficient
control programs. This diffuse activation is similar to the pattern of activation described for early
stages of learning.
A study by Pavlides et al. (1993) specifically examined the role of projections from
somatosensory cortex to Ml for new motor skill learning and retention. Four monkeys were
lesioned in the hand area, somatosensory area of one hemisphere. For 25-45 days post-lesion,
the monkeys were trained to learn a novel motor task (2 monkeys learned to catch a falling food
pellet, 2 monkeys learned to catch a pellet by manipulating a lever for task success). The
monkeys had greater difficulty learning the new motor skill with the contralateral limb indicated by
poorer performance compared to the ipsilateral limb throughout practice. After this training
period, the other hemisphere was lesioned in the same area. This lesion did not abolish the
previously learned motor skill suggesting that projections from somatosensory cortex to Ml have a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
role in the learning of new motor skills but not the retention of previously learned motor tasks.
This study provides evidence that the connections from somatosensory cortex to the motor cortex
have a crucial role in movement execution and new motor skill learning; however, other
neuroanatomic systems participate with the sensorimotor cortical system in motor learning
particularly those functions associated with long-term retention.
Summary
The results of neurophysiologic studies in both primates and humans provide evidence
that early motor skill acquisition occurs by a hierarchically organized, serial process which evolves
through recursive cortical activation during motor learning to a more localized, predominately
parallel process. With practice, primates and humans improve in motor performance. Behavioral
indicators of improved motor performance include movements that are faster (decreased reaction
time and movement time) and more accurate (increased time on target, improved outcome
scores). Improvements in performance are associated with changes in cortical activation patterns
that include the following: 1) diffuse cortical activation which becomes selectively more discrete
during skill acquisition, and 2) a hierarchical, serial activation in early learning stages that evolves
to an activation pattern that suggests that parallel processing occurs in later learning stages;
particularly evident in non-human primate single-cell recording studies. Acquiring motor skill in
early learning stages is accompanied by characteristically common cortical activation patterns
(i.e., primary and association sensory cortices, PF, PM, SMA, Ml). Changes in cortical temporal
firing patterns and brain area activation patterns which occur with motor learning provide
neurophysiologic evidence supporting the hypothesis that a control program is not innate but
evolves over the course of skill acquisition. Therefore, it appears that the primary and secondary
sensorimotor cortical areas contribute to the development of control programs during motor skill
acquisition and most likely comprise the cortical system components of the pragmatic system.
The literature from neurophysiologic studies with non-human primates and imaging
studies with humans, focused on motor skill acquisition, provides strong converging evidence that
the acquisition of control programs leading towards skilled motor behavior occurs in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
sensorimotor cortical system (i.e., pragmatic system). Sensory association areas and PF areas
are essential cortical regions for establishing sensorimotor relationships. The PF area, in
particular, has a role in generating programmed movement alternatives for trial and error analysis.
The slow rate of learning in primates helps to illustrate that the sensory cue is first registered in
primary sensory cortex and given meaning in sensory association and PF cortices. In addition,
recent evidence suggests that primary sensory areas interact with Ml and have an important role
in new motor skill learning.
Once sensorimotor relationships are learned, SMA, PM, and Ml have varying roles in the
acquisition, modification, and execution of the control program. The specific contribution to motor
learning between these discrete areas is not clearly understood; however, these areas are
involved with the acquisition of motor skill and differences in their response properties related to
motor learning can be summarized as follows. PM has a role in selecting movements in response
to a learned sensory association. Of special interest, is the tendency for PM to have a role in
selecting movements with specified performance goals. The exact role of SMA in the motor
program acquisition process is not as well understood. SMA appears to process information
related to motor planning aspects of programmed movements. Clearly, SMA is involved in motor
program execution in early and intermediate learning stages and in the execution of complex
movement sequences. However, it appears that SMA activity may not be needed for overleamed
tasks. Ml functions in concert with PM and SMA in early and intermediate learning stages;
however, in overleamed tasks Ml can execute the sensory-cued motor program independently.
The response characteristic of each area in the sensorimotor cortical system contributes
to the acquisition, modification, and execution of control programs. These areas process
information hierarchically and serially in early skill acquisition stages when movements are
typically considered as discontinuous or not programmed. However, programmed movements
evolve as the performer learns the associations between sensory stimulus, movement response,
and behavioral performance outcomes. Diffuse cortical activation in early learning stages
indicates that the cortex may be generating multiple response alternatives. As learning
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
progresses, cortical activation patterns become more discrete (i.e., fewer activation areas) yet
behavioral performance becomes more skilled (often associated with higher movement speed)
As programmed movements are overleamed, the control program is executed directly from the
sensory-cued association areas to Ml.
It appears that the control program for highly skilled motor performance contains the
information components which contribute to the most optimal behavioral response. The
neurophysiologic findings provide evidence that each area of the sensorimotor cortical system has
response properties which contribute essential information defining control program attributes.
Once this information is learned it becomes embedded into the neuronal network and used by
subsequent areas further along the learning continuum. For example, sensorimotor associations
are learned through sensory association and PF activations. Once these sensorimotor
associations are learned, activation of these areas ceases. Lesions to the PF area do not
interfere with previously learned associations, only with the capability to learn new associations.
In addition, it appears that the sensorimotor cortical system does not function alone during motor
skill acquisition but interacts with other neuroanatomic systems that have a role in the retention of
previously learned motor skills.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
CHAPTER 3
Dissociable Learning Systems within the Brain
Introduction
Historically, the early investigations of H.M. (Corkin. 1968), an individual who had bilateral
medial-temporal lobe resection, provided some of the first evidence that two dissociable forms of
memory that affect learning capability were present in the brain. H.M. had severe declarative
memory deficits but demonstrated improvements in motor performance with practice and
persistence of these changes over days. It has only been since the early 1980's that a substantial
amount of empirical work has been devoted to understanding these different forms of memory
and learning, thus, leading to a general consensus that multiple learning systems exist within the
brain (Schacter. 1987; Seger, 1994a; Shanks & S t John, 1994; Squire, 1986; Squire, 1987). The
purpose of this chapter is to describe the different learning systems of the brain. Specifically, this
chapter will review the following areas: 1) the behavioral distinction for functionally dissociable
systems of learning with particular emphasis on the implicit learning system, 2) neuroanatomic
evidence for dissociable learning systems from brain-damaged populations and PET studies, and
3) implications of a systems view of learning for skill acquisition.
Explicit and Implicit Learning Systems
The explicit learning system is supported by declarative memory. Declarative memory is
conscious recollection of facts and events and is assessed by memory tests such as recall and
recognition tasks that require explicit or conscious recollection of a prior episode. In contrast the
implicit learning system is supported in part by procedural memory. Procedural memory is
unconscious; therefore, procedural memory is assessed by observing changes in performance as
a result of prior experience. Performance changes that reflect implicit learning include skill
learning (motor, perceptual, or cognitive), habit formation, classical conditioning, and priming
(Squire, 1992). For the purposes of this chapter, the discussion of implicit learning will focus on
the processes associated with skill learning and priming since this aspect of implicit learning
pertains most to the learning processes studied in this dissertation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
The procedural-declarative distinction has led to substantial investigation into the
neuroanatomic and behavioral differences between the implicit and explicit learning systems.
Specific tasks have been identified that behavioraily represent the construct and subtypes of
implicit learning. One delineation, developed by Moscovitch (1992), describes tests of implicit
learning as being procedural or item-specific. Procedural implicit learning involves the acquisition
and retention of general skills, procedures, or rules. Item-specific implicit learning involves the
acquisition and retention of types of information such as words, faces, or objects. However, it
appears that implicit learning, in general, involves both perceptual and motor processes such that
what is learned is a series of condition-action events involving mapping of stimuli onto responses
(Nissen.1992; Willingham, Nissen, & Bullemer.1989). Repetition or experience results in a
strengthening of these associations that are behavioraily represented as performance
improvements. Investigations of implicit learning have commonly used three categories of tasks:
1) skill acquisition in tasks such as pursuit rotor, maze drawing, or mirror reading, 2) serial
learning (i.e. sequence acquisition), and 3) repetition priming in tasks such as word-stern
completion, word-association, or picture-completion.
From a behavioral perspective, it has been difficult to develop methods which
unequivocally distinguish between the explicit and implicit learning systems (Green & Shanks,
1993; Shanks & S t John, 1994). One paradigm which deserves special mention is the serial
reaction time task (SRTT). Nissen and Bullemer (1987) developed the SRTT to dissociate
declarative and performance measures of memory. The SRTT involves the acquisition of
sequences that can range from 4-15 elements. During the SRTT, the subject is seated in front of
a display upon which a light appears in one of four positions. The subject has four response
buttons aligned below four lights. When the lights are displayed in a repeating 10-set sequence,
reaction time (RT) rapidly decreases. The decrease in RT is due to anticipation of the expected
stimulus but explicit awareness of the sequence is not usually apparent In contrast RT
significantly increases relative to the repeating sequence if a random sequence is presented.
Therefore, the performance changes in RT during the repeating sequence are considered an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
indicator of implicit learning. Subjects have no conscious awareness of the repeating sequence
as indicated by the inability to generate the sequence when explicitly asked to do so. either
verbally or in demonstration. To determine if attention was necessary for this kind of implicit
learning, a variation of the SRTT was conducted under single-task and dual-task conditions
(Nissen & Bullemer, 1987). The single-task group performed the SRTT with no other task. For
the dual-task group, participants performed the SRTT simultaneously with a tone-counting task
where subjects had to count the number of times they heard a low pitch tone at the end of each
trial block. By conducting the SRTT under 1) a single-task and dual-task condition and 2) in
subjects with and without severe declarative memory impairments, these investigators were able
to dissociate the contributions of the implicit and explicit systems. In addition, they were able to
demonstrate that implicit learning requires attention but not awareness for learning.
Neuroanatomic Evidence for Dissociable Learning Systems
Learning System Distinction within Brain-damaoed Populations
Behavioral indicators such as performance improvements during skill acquisition, serial
learning, and repetition priming have been studied in brain-damaged populations. These studies
have had substantial influence because it has been through the study of individuals with brain-
damage that the distinction between explicit and implicit learning systems has been confirmed. In
addition, the neuroanatomic substrate which supports these learning capabilities has been
identified. Consistently, brain damage within specific cortical and subcortical structures results in
normal learning of certain motor, perceptual, and cognitive tasks even though severe declarative
memory deficits such as impaired performance on recall and recognition tests and, in severe
amnesics, an inability to recollect the learning experience are present
The preservation of procedural memory with impaired declarative memory has been
demonstrated in individuals with Alzheimer’s Disease (AD) (Bondi, Kaszniak, Rapasak, & Butters,
1993; Deweer etal., 1994; Grober, Ausubel, Sliwinski, & Gordon, 1992; Heindel, Salmon, Shults,
Walicke, & Butters, 1989), Huntington’s Disease (HD) (Heindel etal., 1989; Saint-Cyr, Taylor, &
Lang, 1988), Korsakoff’s syndrome (Nissen & Bullemer, 1987), Parkinson’s Disease (PD) (Doyon
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
et al., 1997; Heindel et al. 1989; Pascual-Leone et al., 1993; Saint-Cyr et al., 1988), stroke (Bondi
et al., 1993; Cushman & Caplan, 1987; Platz, Denzler, Kaden, & Mauritz, 1994; Winstein et al.,
1998), temporal lobe disorders (Corkin, 1968; Saint-Cyr et al., 1988; Yamashita, 1993), and
traumatic brain injury (Ewert, Levin, Watson, & Kalisky, 1989; Mutter, Howard, & Howard, 1994).
Collectively, these studies contribute to the general agreement that different brain areas subserve
the explicit and implicit learning systems. The explicit memory and learning system includes the
temporal lobe neocortex, hippocampus, amygdala, and midline diencephalic structures including
the mammillary bodies, and the anterior and mediodorsal thalamic nuclei (Saint-Cyr & Taylor,
1992; Squire, 1987). The neural substrate for the implicit learning system is not as clear but
evidence suggests that the brain areas involved include the prefrontal areas, sensory association
cortices, basal ganglia, and cerebellum (Seger, 1994a).
According to Seger (1994b), the functional nature of implicit learning may require that the
learner engage in many processing mechanisms. In other words, since implicit learning involves
learning abstract relationships the neural basis of implicit learning may involve perceptually-driven
streams of information from association, PF, or basal ganglia areas related to the nature of the
task to be learned. A careful scrutiny of the studies investigating implicit learning in brain
damaged populations supports this functionally-driven systems view of implicit learning. The
findings from brain-damaged populations during implicit learning suggest that 1) specific brain
areas are related to task specific implicit learning capability (Bondi et al.. 1993; Grober et al.,
1992; Heindel et al., 1989), and 2) long-term changes in motor memory and later stages of implicit
learning are associated with activity in striatal and cerebellar structures (Doyon et al., 1997; Doyon
et al, in press).
Dissociation between two implicit memory tasks was demonstrated in a study by Heindel
et al. (1989) comparing performance differences between normal controls, AD, HD, and PD
patients. All subjects were assessed on the Dementia Rating Scale (DRS). DRS scores were
used to identify 3 demented patient groups (AD, HD, demented-PD). The HD and demented-PD
subjects were impaired in motor learning ability compared to the other groups; measured as the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
difference in time on target during a pursuit-rotor task from the first to last practice blocks (6
blocks of 4 trials each). In contrast the AD and demented-PD subjects were impaired in lexical
priming compared to the other groups (i.e., significantly lower percentage of word stems
completed). The authors concluded that the double dissociation between the HD and AD subjects
on motor learning and lexical priming suggests that there are different forms of implicit memory
that depend upon distinct neuroanatomical systems. Demented-PD subjects were impaired on
both tests of implicit memory which suggests that neurologic damage had occured in the
neuroanatomic systems subserving both motor learning and verbal priming. Of particular interest
was the finding that non-demented PD subjects were not impaired in explicit (verbal recall) tasks
or the 2 implicit tasks used in this study. Therefore, the performance of non-demented compared
to demented-PD subjects suggests that implicit learning may be subserved by a neruoanatomic
system that involves the interaction of cortical and subcortical processes. In demented-PD, the
loss of cortical and subcortical neurons interfered with both of the implicit learning tasks examined
in this study. However, another consideration is that demented-PD subjects are also experiencing
progressive degeneration of the striatum.
Doyon et al. (1997) found differences between PD groups related to the severity of striatal
degeneration. In this study the SRTT paradigm was used to investigate implicit learning ability in
early and late acquisition phases for PD, cerebellar involvement, focal PF involvement, and
matched control groups. Subjects were given a 10-item sequence repeated 10 times to make a
100-trial block. Four blocks were given on 6 days dispersed across 6 weeks. This study revealed
several interesting findings. First, the PD group was stratified such that PD subjects with minimal
involvement (Hoehn & Yahr, Stage 1) were not impaired in implicit learning. However, subjects
with more severe striatal degeneration (Hoehn & Yahr, Stage 2-3) were more impaired in implicit
learning compared to controls. Neither group was demented. Therefore, this study suggests that
substantial, bilateral degeneration to the striatum can result in implicit learning deficits
independent of cortical damage. In addition, this study found that subjects with cerebellar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
damage presented with implicit learning deficits similar to that of the more involved PD group.
Subjects with focal PF damage did not demonstrate implicit learning deficits.
The findings of this study are significant because these investigators revealed a
neuroanatomic distinction between groups related to area (cerebellum, striate) and severity. In
addition, there was a relationship between the stage of skill acquisition that differentially
dissociated the neuroanatomic areas and explains the discrepancies found in other studies that
have only investigated changes in the early stages of skill acquisition. The key finding was that
subjects with bilateral striatal dysfunction and those with cerebellar involvement did not improve
performance compared to matched controls particularly in the later acquisition stages. In a follow-
up study, these same subjects were retested 10 to 18 months later in order to determine long
term retention of the skill (Doyon et al„ in press). Subjects were given four blocks of 100 trials.
Subjects with PD who had changed from Stage 1 to Stage 2 and cerebellar subjects
demonstrated deficits in long-term retention indicated by a greater deterioration in performance
compared to that of controls. The authors concluded that the cerebellum and striatum have a role
in the skillful acquisition of a visuomotor sequence as well as the long-term retention of this skill.
Furthermore, the loss of ability that accompanied progressive striatal degeneration suggests that
the striatum not only contributes to skill acquisition but may be a site for motor program memory
storage important for retention performance.
The study by Heindel et al. (1989) demonstrated a dissociation between groups for
implicit learning tasks that revealed cortical contributions to implicit learning. Differential cortical
contributions were further revealed in a study by Grober et al. (1992). This study demonstrated
implicit learning dissociation within a group of AD subjects. AD subjects were able to improve
performance on a repetition priming task of previously read words but did not acquire the skill of
mirror reading. These findings suggest that, within a neuroanatomic system, the processing may
be different for these two tasks (i.e., verbal and visual priming). It is interesting to note that
repetition priming does not appear to be a unitary process but is, in and of itself, task specific.
AD subjects demonstrated impairment in repetition priming for verbal, word-stern completion
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
tasks (Heindel et al.,1989), and yet in this study, AD subjects demonstrated performance
improvements in repetition priming for reading words. The authors suggest that the preserved
repetition priming effects reflect the function of the perceptual representation system which
appears to be intact for read words but impaired for other verbal tasks in individuals with AD.
To better understand the brain structures which mediate skill learning and forms of
priming, Bondi et al. (1993) compared the performance of AD, anterior communicating artery
aneurysm patients (ACoA), and herpes encephalitic patients with temporal lobe damage (HE) in
the following implicit learning tasks: fragmented pictures, word-stem completion, and pursuit-rotor.
The patient groups were impaired in all explicit memory tests (verbal recall). Both AD and HE
subjects were intact for fragmented picture and pursuit-rotor performance but were impaired in
word-stem completion compared to healthy controls. However, the ACoA subjects were not
impaired in any implicit learning task, consistent with the PF group findings of Doyon et al. (1997).
The findings from this study provide additional support for a systems view of implicit
learning. Both AD and ACoA subjects demonstrate evidence of prefrontal brain damage but the
capability for implicit learning is not impaired for focal lesions to PF areas compared to the more
widespread damage of temporal-parietal neocortical areas associated with AD. Furthermore, it
appears that the temporal lobes are involved in lexical priming but not in the repetition priming
associated with picture completion. Therefore, processing between PF and temporal areas
contribute to task-specific, in this instance lexical priming, implicit learning. The unimpaired
performance in the pursuit-rotor motor task and the lack of pathology in subcortical structures in
the patient groups from this study, provide further support for the role of the neostriatal system in
motor skill implicit learning.
Cortical and Subcortical Modules for Implicit Learning: Evidence from PET
There is compelling evidence from brain-damaged populations that 1) explicit learning is
dissociable from implicit learning, 2) the neural substrate for the explicit learning system is
neuroanatomically distinct from areas associated with the implicit learning system, and 3) there
are functional task-specific differences between implicit memory tasks which appear to be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
associated with specific cortical modules. It is not clear where or how learning-related changes
are processed within the brain. Several recent studies, that directly examined implicit learning,
provide some insight concerning these central processing issues.
Grafton, Hazeltine, and Ivry (1995) measured learning-related changes in rCBF during
sequence learning of the SRTT under dual-task and single-task conditions. During dual-task
performance, the subject is responding to the light sequence while monitoring a stream of tones.
The SRTT under dual-task conditions is believed to isolate the implicit learning system. As
expected, subjects' response times decreased during repeating sequences and returned to
baseline when a random sequence was introduced. As the subjects acquired the sequence
structure, changes in rCBF were noted as increasing activity in left Ml, SMA, left PF, left parietal
cortex, and bilateral putamen. These sites are the motor areas associated with execution of other
motor tasks (Deiber et al.,1991; Grafton, Mazziotta, Woods et al., 1992; Grafton, Mazziotta. Presty
et al., 1992). Interestingly, when subjects acquired the sequence explicitly these motor areas
returned to baseline levels. Learning-related changes attributed to explicit sequence learning
were associated with increased activation of right dorsolateral PF, right PM, and bilateral inferior
parietal cortex.
A recent study, using the SRTT paradigm with PET imaging during early and late skill
acquisition, has revealed that cortical activation patterns during visuomotor sequence learning are
associated with the stage of learning (Doyon et al., 1996). In the early learning stage, areas
known to be involved with the motor task had higher activation including the contralateral
sensorimotor cortex, supplementary motor areas, and ipsilateral cerebellar dentate nucleus. In
contrast later practice resulted in higher activation of the ventral striatum and cerebellar dentate
nucleus. This study supports the previous findings delineated in Chapter 2 that cortical activation
patterns change throughout the process of motor learning. These findings, together with the
studies of brain damage due to PD or cerebellar involvement (Doyon et al., 1997; Doyon et al., in
press), provide strong evidence that striatal and cerebellar areas are involved in implicit motor
learning particularly in the later stages of learning when movements become more automatic.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
In addition, these findings are consistent with the “ fast-/slow-leaming" processing modes
suggested by Kami et al. (1995). It appears that the “ fast-learning" processes associated with
establishing motor routines and characteristic of early acquisition are mediated in cortical motor
areas such as the sensorimotor cortex, supplementary motor areas, and ipsilateral cerebellum;
areas typically described for the control of motor actions. The “ slow-learning” processes
associated with long-term motor memory and implicit learning, and characteristic of later
acquisition, appear to be mediated by subcortical structures that include primarily the striatum and
cerebellum.
The functional anatomy of explicit and implicit memory retrieval for a verbal, word-stem
completion task was investigated in a PET study by Buckner et al. (1995). Imaging occurred
under three conditions: 1) baseline condition to identify areas associated with the processing of
the visual cue and producing the vocalization, 2) recall condition requiring explicit retrieval of
previously studied words, and 3) priming condition requiring implicit retrieval in a repetition
priming, word-stem completion task. By subtracting baseline activations from the activations
during recall and priming tasks, functional brain areas associated with explicit and implicit
memory, respectively, were identified. In both the recall and priming conditions, activations of the
extrastriatal cortex, Rolandic cortex, medial cerebellar, SMA, and left PF areas associated with
visual identification and speech production occurred. During recall, increased activation was
present in right anterior PF, left PF, and anterior cingulate cortices, most likely related to search
and retrieval strategies. In contrast, there was decreased activation in the sensorimotor areas
during the priming condition, particularly in the visual processing occipito-tempora! cortex. The
authors suggest that the neural correlate of priming may be a decrease in activation since
perceptual processing is more efficient after exposure to a stimulus.
Together the Grafton et al. (1995), Doyon et al. (1996), and Buckner et al. (1995) PET
studies provide corroborating evidence suggesting that implicit learning occurs within a functional
neuroanatomic network that includes specific components from cortical, subcortical, and
cerebellar structures. Grafton et al. (1995) found increases in cortical activation associated with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
implicit motor sequence learning. However, this study most likely revealed activation of structures
involved in the implicit motor action associated with early acquisition. In contrast Doyon et al.
(1996) replicated this early acquisition finding and went on to reveal a change in the
neuroanatomic substrate that mediates later stages of skill acquisition. Buckner et al. (1995)
found a decrease in cortical activation associated with implicit verbal learning; a finding that
suggests a neuroanatomic correlate for the implicit priming effect Together these studies provide
preliminary evidence that implicit learning is associated with processing changes that occur within
cortical and subcortical modules that are distinct neuroanatomically and functionally.
Implications of a Systems View of Learning for Skill Acquisition
In summary, there are at least two known dissociable learning systems within the brain.
The explicit learning system is supported by declarative memory and its neural substrate includes
the medial-temporal diencephalic system. The implicit learning system is supported by procedural
memory characterized by unconscious memory processes associated with such forms of learning
as skill acquisition or priming. The neural substrate associated with implicit learning is less clear.
However, careful scrutiny of studies in brain-damaged populations and the few PET studies which
have examined implicit learning suggests that implicit learning occurs: 1) outside the medial-
temporal diencephalic system, and 2) within the cortical modules associated with particular
perceptual processes, and subcortical and cerebellar structures associated with longer-term
acquisition and retention. For example, verbal priming occurs within the brain network known to
be involved with verbalization (Buckner et al., 1995), and implicit motor sequence learning occurs
within the motor effector areas for early learning (Grafton et al., 1995). However, longer-term
retention of motor skills involves subcortical structures in the striate and cerebellum (Doyon et al.,
1997; Doyon et al., in press).
Implicit motor learning is correlated with neural activity in the cortical-striatal-cerebellar
network (Doyon et al., 1997; Doyon et al., in press; Grafton et al., 1995). Brain-damaged
populations including Huntington’s Disease, Parkinson’s Disease, and cerebellar pathology
demonstrate some level of impairment in implicit motor learning related to the degree of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
pathology. Performance, ranging from intact skill learning to severely impaired skill learning,
varies within these groups as neurologic pathology progressively degenerates. Together the
findings suggest that 1) a critical mass of basal ganglia or cerebellar involvement results in
impaired implicit learning and 2) the rate of skill acquisition can be impaired for populations with
diffuse cortical damage particulary if there is concomitant involvement of striatal or cerebellar
structures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
CHAPTER 4
Behavioral Basis of Skill Acquisition and Motor Learning
Introduction
The neuroanatomic approach to motor skill acquisition provides important insights about
the brain areas involved in movement control. Since cortical activation patterns change
throughout the stages of skill acquisition, it is apparent that central processing is also changing.
The behavioral manifestation of these central processing changes can be investigated through
kinematic analysis of movement trajectories. It has been suggested that motor learning of
programmed movements progresses from closed-loop control to open-loop control as indicated
by movements which are discontinuous early in learning but become continuous with experience
(Brooks, 1979; Pew, 1966). While it has been well established that performance improves with
practice (Crossman, 1959; Mowbray & Rhodes, 1959; Schmidt, 1988; Snoddy, 1926; Woodworth,
1899), most skill acquisition studies focus on behavioral outcome measures such as movement
time or global error scores and do not investigate how the learner acquires the skill. A few studies
have used a kinematic approach to examine skill acquisition of rapid movements and have
demonstrated that in early acquisition movement trajectories are variable with discontinuous
velocity and acceleration profiles which become less variable and more continuous with practice
(Brooks & Watts, 1988; Darling & Cooke, 1987; Moore & Marteniuk, 1986).
Gross measures of motor performance and the absence of experimental designs that
systematically distinguish performance (i.e., acquisition) and learning (i.e., retention or transfer)
are two of the major limitations in most studies of motor skill acquisition from either a functional
neuroanatomic or behavioral perspective. Attention to methods that discriminate between central
processes associated with movement execution and motor learning would provide a means to
investigate behavior differences between the pragmatic and semantic systems. The purpose of
this Chapter is to review the following areas: 1) kinematic methods which describe the dynamic
changes that occur throughout the course of skill acquisition, and 2) behavioral experimental
designs that distinguish performance from learning effects.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
Kinematic Indicators of Skill Acquisition
There are two predominate theories of motor skill acquisition. The dynamic theory of skill
acquisition postulates that skill acquisition throughout practice is a search for the optimal
coordination and control of multiple variables (Fowler & Turvey, 1978). Therefore, learning is
acquiring the coordination between the perceptual environment action environment, and task
constraints (Newell, 1991). From a dynamical systems perspective, the early stages of learning
consist of acquiring the relative limb motions (i.e., topographical characteristics) with later practice
leading to more skillful scaling of these features (Newell, 1985). In other words, skill acquisition is
learning about the limb dynamics by experiencing the relationship between the action and the
outcome. It seems reasonable to assume that while the performer may be aware of the
displacement goal of a task at an explicit level, it is highly unlikely that they are aware of the limb
dynamic profile (i.e., accelerations) to skillfully reach that goal. Therefore, acquiring skill as
measured by changes in a higher-order variable such as acceleration may be a way to infer
implicit motor learning of a task.
A second predominate theory of motor skill changes during acquisition is Schmidt’s
Schema theory (Schmidt 1975). Schema theory postulates that motor learning involves acquiring
prescriptions for action that satisfy a task goal through the construction of recall and recognition
schema. The recall schemata is involved with the processes which select movement parameters
that specify force, duration, and amplitude. The recognition schemata involves processes that
evaluate the correctness of the movement. The strength of the recall and recognition schema
builds up as a result of practice and feedback. Together the recall and recognition schema
contribute to the organization of movement behavior into certain invariant features that Schmidt
(1985) termed the generalized motor program (GMP). According to this theoretical perspective,
skill acquisition is the process by which the performer acquires the GMP for a class of actions
which share the same relative timing, and leams to scale this motor program, spatially and
temporally.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
Changes that occur during motor skill acquisition can be analyzed at the behavioral level
through kinematic analysis. Based upon these two theoretical perspectives of skill acquisition,
kinematic analysis has been used as a method to investigate changes in motor performance with
practice. Kinematic analysis has been approached in one of two ways through investigation of 1)
changes in displacement velocity, and acceleration profiles which occur throughout practice, and
2) measures of the acquisition and parameterization of the GMP.
Changes in Displacement. Velocity, and Acceleration
One of the advantages of using kinematic analysis is that motor learning is emphasized
as a process rather than a product (Marteniuk & Romanow, 1983) such that there is a dynamic
transition in modes of control between early and late practice. The use of programmed
movements occurs as control progresses from closed-loop to open-loop processing (Brooks,
1979; Pew, 1966; Schmidt, 1987). Several studies have used kinematic analyses to demonstrate
that the control of rapid movement is characterized by discontinuous velocity and acceleration
profiles which become more continuous with practice (Brooks & Watts, 1988; Brooks, Hilperath,
Brooks, Ross, & Freund, 1995; Darling & Cooke, 1987; Marteniuk & Romanow, 1983; Moore &
Marteniuk, 1986).
Changes in movement speed, accuracy, and consistency that reflect more skilled motor
performance during later practice can be identified with kinematic analysis. In addition,
acceleration profiles with several zero crossings within a movement segment can be used to
identify on-line feedback-based adjustments (van Donkelaar & Franks, 1991). By identifying trials
with discontinuity within the acceleration profile, trials with primarily programmed movements can
be distinguished from trials where on-line feedback corrections are employed.
Marteniuk and Romanow (1983) used kinematic analysis to investigate how a skill was
acquired by using a horizontal lever task which involved producing a series of flexion-extension
movements to produce a 5 sec criterion movement pattern. The subject was presented with the
goal-movement pattern on a screen which went blank prior to the subject moving. After the
movement the subject saw the criterion and subject-produced trajectory. Two college students
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
practiced this task for 800 trials (80 trials/dy for 10 dys). Kinematic and Fourier analysis of the
trajectories was completed to characterize the process of skill acquisition. As expected, trajectory
variability decreased with practice; however, the variability decreased equally across the whole
trajectory suggesting that the whole movement was being learned not just parts of the movement.
From the Fourier analysis, it appeared that early in practice subjects relied on spatial or
displacement information to leam the movement and added higher-order control (i.e., velocity and
acceleration) information as practice proceeded. Separate cross-correlation analyses between
the subject-produced and criterion trajectory for displacement, velocity, and acceleration were
calculated to assess spatial and temporal ability across trials. This analysis led to the conclusion
that subjects work simultaneously on spatial and temporal control. In summary, young subjects
solved the skill acquisition problem by simultaneously learning the whole movement including the
spatial and temporal components. This type of kinematic analysis has not been done with brain
damaged populations. It is not known if damage to the sensorimotor cortical system would
interfere with this type of high-level strategy during skill acquisition; however, the evidence
provided in Chapter 2 would suggest that the sensorimotor areas are the likely neuroanatomic
substrate that would subserve this skillful behavior.
Darling, Cooke, and Brown (1989) used kinematic analysis to compare performance of a
visual step-tracking elbow flexion-extension movement between young (21-24 yrs) and elderly
subjects (68-95 yrs). Five of the elderly subjects practiced 30° flexion-extension movements for
180 trials each. Subjects were told to increase movement speed and accuracy during practice.
Elderly subjects had more discontinuous movements and increased trajectory variability
particularly in the acceleration phase compared to younger subjects. With practice, elderly
subjects reduced trajectory variability but did not increase speed. Movements became more
continuous but evidence of discontinuous segments were still present in the velocity profiles.
While elderly did demonstrate improvements in motor performance with practice, elderly subjects
continued to be more variable in the acceleration phase compared to young subjects. Kinematic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
analysis was effective in identifying performance improvements with practice and the nature of the
performance error present between groups.
Kinematic analysis has been effective in detecting performance deficits in motor control
due to hemispheric differences that could not be detected by more global performance measures
(Goodale. 1988; Winstein & Pohl, 1995). There has been only one study (Pohl & Winstein,
1998a) which has used this level of kinematic analysis in a brain-damaged population to
investigate the effects of practice. Using a reciprocal tapping task, individuals with unilateral
brain-damage were able to decrease MT with practice and used the same strategy as controls to
achieve this performance improvement Individuals with LCVA continued to demonstrate deficits
in the sequencing and timing of movement reversals that was not remediated with practice. It
appears that kinematic analysis may be a more sensitive method for detecting motor control and
motor learning differences between individuals with brain-damage and healthy controls.
Acquisition and Parameterization of the GMP
According to GMP theory (Schmidt, 1985), skill acquisition for a class of actions which
share the same relative timing involves acquiring the GMP and executing the GMP with spatial
and temporal accuracy (Wulf & Schmidt 1994). The GMP is a learned structure acquired with
practice in which sequencing, relative timing, and relative forces are specified (Schmidt, 1987).
The structure of the GMP is determined by identifying certain invariant movement features that
have fixed relative timing. For example, a movement would be considered to be under
programmed control if the timing of kinematic landmarks are in proportion to the total movement
duration. The process by which the performer acquires and retains the relative timing of a
movement would be a reflection of motor programming ability. Another aspect of skill acquisition,
from the programming perspective, is the effectiveness with which the GMP is parameterized.
Parameters are superficial features by which the GMP is scaled, such as movement duration and
amplitude. For a rapid movement with specified spatial and temporal goals, skilled performance
involves acquiring the movement structure (i.e., relative timing) and scaling the movement to meet
instantaneous spatial and temporal coordinates.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
Typically, motor learning studies use global performance measures such as root mean
square error (RMSE) or MT that captures both programming and scaling errors. To address this
lack of sensitivity, Wulf, Schmidt, & Deubel (1993) developed a method to discriminate between
GMP learning and parameterization learning based on previous work by Winstein (1988). This
residual-RMSE method, was developed to separate GMP error from parameterization error. The
method involves scaling the subject’s trajectory in time and amplitude so that the agreement
between the subjects's trace and the goal trace is maximized. The analysis results in a timing
ratio (>1, movement too slow; <1, movement too fast), spatial ratio (>1, movement too long; <1,
movement too short), and a residual-RMSE score (the remaining RMSE after temporal and spatial
scaling indicates the accuracy of the GMP).
The residual-RMSE method has been used to demonstrate that GMP and
parameterization learning are differentially affected by conditions of practice (Sekiya, Magill,
Sidaway, & Anderson, 1994; Wulf & Schmidt, 1994; Wulf, Lee, & Schmidt, 1994; Wulf, Schmidt, &
Deubel, 1993). These findings support the conclusion that programming and scaling capability
are separable and suggest that different central processing mechanisms may subserve and
influence the learning of these behaviorally dissociable movement attributes. In addition, Schmidt
(1991b) believes that acquiring the GMP may be a more difficult skill to learn; having a greater
impact on skill proficiency than parameterization ability. This may or may not be true in brain
damaged populations who have such severe motor performance deficiencies that poor scaling
ability (such as inability to walk quickly) has a substantial functional impact as costly as the
inability to effectively program the action. Nevertheless, the use of gross performance measures
has been a major limitation in skill acquisition studies with brain-damaged populations; limiting our
understanding of the nature of performance deficits post-brain injury and the process by which
brain-damaged individuals become more skilled with practice. Using more sensitive measures of
motor performance may prove more beneficial for investigating the relationship between brain
pathology and behavioral indicators of motor control and motor learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
Young and Schmidt (1990) examined changes in motor control with practice in young
subjects learning a coincident-timing task. The laboratory task was similar to a real-world activity
like batting a ball since the subject had to backswing a horizontal lever then forwardswing the end
of the lever to coincide with the last in a sequence of illuminated LEDs as it passed a coincident-
point Young and Schmidt (1990) proposed the use of a method termed the unit-analysis (Weiss.
1941) to determine the action units or motor programs that comprise a movement. Unit analysis
involves identifying movement landmarks (preferably from the acceleration profile which are easily
recognized and evenly distributed across the movement), and computing correlations between the
first landmark and successive landmarks and the various landmarks and the last landmark
(Schimdt & Young, 1991). If the pattern of correlations is consistent throughout the movement
then the movement is believed to be controlled by one unit (i.e., GMP). However, if there are
boundaries with low correlations within a movement this suggests that feedback processing is
occurring during that time. By using the unit-analysis method. Young and Schmidt (1990)
demonstrated that in early practice subjects used a one-unit action but the highest performance
scores were achieved with a two-unit movement In this particular task, the highest level of skill
occurred when subjects used visual extrinsic feedback during the movement to increase end
point accuracy.
The application of this method for a skill acquisition study is of interest because the unit-
analysis method was useful in determining the control processes which resulted in optimal
performance. The Young and Schmidt (1990) task required both programmed and on-line
feedback strategies to acquire the highest level of skilled performance. However, the method
could be applied for tasks which may be controlled by closed-loop processes early in practice but
evolve to more open-loop control later in practice. For example, if a rapid movement with spatial
and temporal constraints known to be optimally controlled by a one-unit action had intervals with
low correlations early in practice, then this would indicate trials where on-line, feedback or closed-
loop processing had occurred. However, as practice proceeds movements should evolve to a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
more continuous, one-unit action whereby kinematic landmarks are highly correlated throughout
the action.
Motor Performance and Learning Distinction
There is an important behavioral distinction between motor skill acquisition and motor
learning (Bjork & Bjork, 1992; Druckman & Bjork, 1991; Tolman, 1932). Motor skill acquisition is
considered a process that reflects the dynamic changes in control that occur with practice
(Marteniuk & Romanow, 1983). In contrast motor learning is the relatively permanent capability
to produce skilled action (Schmidt 1988); therefore, motor learning can be considered a product
of practice (Schmidt 1991c). Chapter 2 provides compelling evidence from neurophysiologic and
behavioral studies that the motor system is organized differently after practice that results in
improvements in motor performance; a robust finding in both non-human primates and humans
with and without brain-damage. However, there is considerable evidence from research in the
field of motor learning that conditions of practice that positively impact practice performance may
not engage the kinds of central processing activities that are required to improve performance
capabilities in retention or transfer situations, more credible indicators of motor learning (Bjork &
Bjork, 1992; Salmoni, Schmidt & Walter, 1984; Schmidt 1991a; Schmidt & Bjork, 1992; Winstein,
Pohl, & Lewthwaite, 1994; Winstein & Schmidt 1990).
The importance of the performance/learning distinction has led to experimental designs
for evaluating motor learning. In motor learning studies, the learner engages in task practice
during an acquisition phase. Practice performance is graphed as a function of practice trials;
improvements in performance as a result of practice are expected. In order to assess the relative
permanence of the practice improvements, performance is measured in retention or transfer tests
with long enough post-practice durations to ensure that the temporary effects of the practice
condition have dissipated.
Winstein and Schmidt (1990) conducted a motor learning study in which the frequency of
feedback (i.e., knowledge of results; KR) was manipulated during practice. Performance
improved with practice whether feedback was provided after every trial or after only 50% of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
trials. However, subjects receiving less frequent feedback performed better on delayed-retention
compared to those who received feedback after every trial. This is one example where a
condition of practice such as reduced feedback frequency had a more beneficial effect on
delayed-retention performance, an indicator of learning, than on acquisition performance.
Winstein and Schmidt (1990) suggest that feedback has a strong guidance effect that may
actually interfere with the active recall and retrieval processes required for motor learning. Other
motor learning studies have manipulated practice conditions by varying the schedule of tasks
during practice (Shea & Morgan, 1979; Lee & Magill, 1983), the method of scheduling KR
frequency (Lee & Carnahan, 1990), and the type of augmented feedback (Winstein et al., 1994).
Collectively, these studies demonstrate that conditions of practice that encourage active
information processing by the learner results in a greater capability to retain and generalize the
knowledge gained to other situations (Swinnen et al., 1994).
Conditions of practice that are heavily guided or have low contextual interference invoke
an information processing strategy that enhances practice performance over the processes
required for learning, and may have a particularly facilitatory effect on the pragmatic system where
optimal performance is the functional priority. The semantic system, involved with abstract rule
learning, may be more positively affected by information processing that invokes search and
retrieval strategies; processes where the learner is more actively engaged in using internal
representations of movement
Studies which manipulate practice conditions determine the learning effect between
groups by measuring performance differences in retention or transfer tests when the independent
variable has been equated (e.g., retention test where both groups perform with feedback
removed). Swanson and Lee (1992) manipulated both practice condition and subject group to
investigate the effects of aging on motor learning processes. Throughout acquisition and
retention phases, older adults were less accurate than younger subjects; however, the older
subjects were as effective as younger adults in using feedback during practice to leam a motor
skill. These findings were supported by significant group differences but no interactions of group
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
by trial block during practice or retention phases; an effect that distinguishes performance and
learning differences between groups.
Studies of Individuals with Brain-damage
Since the acquisition of complex motor skills is associated with diffuse and bilateral
cortical and subcortical activation, motor control deficits (i.e., decreased effectiveness of
programming or parameter scaling ability) of the limb ipsilateral to the damaged hemisphere
would be indicative of a specialized, hemispheric processing function. The advantage of studying
ipsilateral limb movements in cases with unilateral hemisphere damage is that the primary motor
impairments to the ipsilateral limb are not as limiting as those present on the contralateral side;
therefore, the capability to acquire motor skills will not be as directly affected as it would be in the
presence of severe hemiparesis or hemisensory loss (Haaland, Harrington, & Yeo, 1987).
Studies using this approach to investigate hemispheric differences in motor control
indicate that each hemisphere has specialty functions contributing to accurate movements. The
left-hemisphere has a greater role in the timing and sequencing of programmed movements; the
right-hemisphere has a greater role in visual-spatial integration (Fisk & Goodale, 1988; Goodaie,
1988; Haaland etal., 1987; Winstein & Pohl, 1995). Only one study (Pohl & Winstein, 1998) has
specifically examined the differential role of the cerebral hemispheres during extended practice.
Using Fitts' paradigm (Fitts, 1954), individuals in this study practiced aiming tasks under
conditions of task complexity ranging from predominately open-loop to closed-loop control. All
subjects improved performance with practice (i.e., decreased movement time); however,
individuals with unilateral brain-damage had significantly slower movement times compared to
controls in all conditions. Of particular interest was the finding that only the subjects with left-
hemisphere damage demonstrated motor control deficits in relative timing which were not
remediated by extended practice. The Fitts’ task does not have explicit temporal constraints;
subjects are instructed to be as fast and accurate as possible and are not given an explicitly
defined temporal goal. Threfore, it is not dear if the motor control deficits are relatively permanent
(i.e., not remediable with practice) or if the task allowed the subjects to chose a movement
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
strategy indicative of the neurologic damage (i.e., a movement strategy not requiring
programming). In either case, it appears that the cerebral hemispheres have a specialized role in
the movement execution function of the pragmatic processing system.
In order to investigate semantic processing system deficits, a study that examined the
effects of unilateral brain-damage on motor learning processes was conducted (Winstein et al.,
1998). Individuals with unilateral brain-damage from stroke benefited from practice with feedback
in a similar manner to matched controls and retained these practice improvements in retention.
As with the older subjects (Swanson & Lee, 1992), individuals with brain injury were less accurate
and consistent throughout practice and retention phases. However, a lack of a group by practice
block interaction suggests that the processes associated with motor learning did not differ
between groups. It appears that a motor learning paradigm (i.e., one that has acquisition and
retention phases within the experimental design) is effective for distinguishing differences
between groups due to execution errors from those processes associated with motor learning.
However, the nature of the performance improvements with practice and the persistent
performance errors (i.e., programming or parameter scaling) can best be determined through
kinematic analyses.
Summary
A behavioral approach to skill acquisition discriminates between processes that relate to
changes in control (i.e., motor performance) from those of learning (i.e., retention or transfer). In
the early stages of skill acquisition, movements are discontinuous which suggests that closed-
loop or feedback processes are involved in movement control (van Donkelaar & Franks, 1991).
However, with extended practice movements become more continuous, evident as velocity and
acceleration profiles that did not have discontinuities such as zero crossings (Brooks & Watts,
1988; Brooks et al., 1995; Darling & Cooke, 1987; Marteniuk & Romanow, 1983; Moore &
Marteniuk, 1986). These changes in motor control that occur with practice are indicators of
performance. It is only through experimental designs that incorporate retention or transfer tests
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
that the processes associated with motor learning can be investigated (Bjork & Bjork, 1992;
Saimoni et al., 1984; Schmidt 1991a; Schmidt & Bjork, 1992; Winstein & Schmidt 1990).
One theory of motor skill acquisition is the GMP theory (Schmidt 1985). The GMP theory
postulates that rapid movements are governed by a motor program which is scaled to meet task
temporal and spatial goals. Therefore, a global performance measure such as RMSE, which
includes programming and scaling error, may not be a sensitive indicator of motor control
changes that occur with practice. The residual-RMSE and unit-analysis methods were developed
to distinguish between these various motor control attributes. These methods have been used to
investigate how conditions of practice that manipulate feedback frequency affect action structure
(Young & Schmidt, 1990) or programming and parameterization capability (Sekiya et al., 1994;
Wulf & Schmidt, 1994; Wulf etal., 1994; Wulf etal., 1993). There have been no studies to date
that have used these methods to quantify changes in motor control processes that occur with
practice.
Combining neuroanatomic and behavioral approaches can be an effective means for
understanding brain-behavior relationships. The study of motor performance during the course of
skill acquisition in individuals who have an isolated deficit in an area of the movement control
neural system can reveal pertinent information about the function of the brain during movement
execution and motor learning. In addition, understanding the behavioral motor performance of
individuals with brain-damage can lead to predictions about the functional consequences of brain
damage and have practical application to rehabilitation strategies (Squire, 1987).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 5
55
DEFICITS IN MOTOR PROGRAM ACQUISITION AND EXECUTION BUT NOT MOTOR
LEARNING AFTER UNILATERAL SENSORIMOTOR CORTICAL SYSTEM DAMAGE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
Abstract
The sensorimotor cortical system has a well established role in the execution of fast and
accurate movements. It is less clear how the central processes that influence movement control
interact with those that influence motor learning. The purpose of this study was to investigate the
role of the sensorimotor cortical system in the control and learning of a rapid, upper limb
movement that can be controlled by a single motor program. Right-handed adults with unilateral
sensorimotor cortical system stroke and healthy, age-matched controls practiced a rapid upper
limb continuous movement with 3 flexion-extension reversals within 1000 ms for 200 trials on Day
1, and returned Day 2 for retention tests. Subjects with stroke used the limb ipsilateral to the
lesion. Kinematic analyses were used to quantify changes in motor control with practice.
Tnroughout the first 100 trials of practice, the stroke group was not as proficient in motor
programming compared to the controls; however, by the end of practice both groups had acquired
programmed movements and retained the movements to a comparable degree during retention.
The sensorimotor cortical system has a role in motor programming that is influenced by
hemisphere. Left-hemisphere damage resulted in greater motor programming error. In contrast,
subjects with right-hemisphere damage required more trials to acquire continuous movements.
The findings suggest that the sensorimotor cortical system has a major role in the acquisition and
control of a rapid limb movement that is influenced by hemisphere. However, the longer-term
learning of programmed actions was not affected by sensorimotor cortical system damage
regardless of hemisphere.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
Introduction
Motor skill acquisition of a rapid movement with spatially and temporally defined task
goals involves two distinct behavioral outcomes: one involved in the acquisition of movements
which are fast and accurate, and another associated with the ability to retain this motor skill and/or
generalize the motor skill to other novel actions. Jeannerod et al.(1995), using the cortical
mechanisms involved in grasp control as an example of a distributed neural network, have
proposed that diffuse cortical areas subserve the processing of information which relates to
different behavioral cognitive functions. The “ pragmatic" mode of processing functions to extract
the parameters relevant to an action. The “ semantic” mode of processing takes movement
attributes and binds them together to produce a unique precept; in other words, semantic
processing provides meaning and relevance to the action. In a similar manner, a distributed
neural network appears to be involved in the acquisition and retention of perceptual-motor skills.
Substantial neurophysiologic and behavioral evidence suggests that two functionally
distinct neuroanatomic systems contribute to the acquisition and retention of motor skills;
paralleling the concept of the pragmatic and semantic modes of processing proposed by
Jeannerod et al. (1995). One is a distributed cortical movement execution network involved in the
control of movements which are fast and accurate (Colebatch et al., 1991; Grafton, Mazziota,
Woods etal., 1992; Johnson, 1992; Kalaska & Crammond, 1992). This system, which has a
major role in the attributes of movement execution, has its own inherent learning properties such
as more efficient perceptual-motor processing with practice (Sasaki & Gemba, 1981, 1982;
Kubota & Komatsu, 1985; Kami et al., 1995). The neuroanatomic network associated with the
control of fast and accurate movements includes the primary and secondary sensorimotor
cortices and their corresponding afferent and efferent (i.e., intercortical, subcortical, striatal)
projections. This neuroanatomic network will be referred to as the sensorimotor cortical system.
The sensorimotor cortical system appears to share functional processing attributes synonymous
with the pragmatic system. The second is the procedural memory network which is involved with
a more general or abstract implicit learning capability indicated by the acquisition and retention of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
skills, procedures, and rules (Schacter, 1987; Seger, 1994a; Shanks & S t John, 1994; Squire,
1986; Squire, 1987). The neuroanatomic network associated with procedural memory and more
permanent indicators of motor learning includes prefrontal, striatal, and cerebellar structures
(Doyon et al., 1997; Doyon et al., in press; Grafton et al., 1995; Haaland & Harrington, 1990;
Ungerleider, 1995). This neuroanatomic network appears to share functional processing
attributes synonymous with the semantic system.
The pragmatic and semantic systems appear to be two functional neural networks
subserving motor skill acquisition. This functional system distinction provides a framework for
investigating the interaction of motor control and motor learning processes. Motor control is a
neurophysiologic process within the motor system governing the initiation and execution of
movement (Newell, 1991) and can be functionally associated with the pragmatic system.
Therefore, behavioral measures of movement initiation and execution and the changes which
occur over the course of skill acquisition are indicators of the integrity of the pragmatic system.
Motor learning is a set of central processes associated with long-term changes in the capability to
produce skilled action (Schmidt 1988) and can be functionally associated with the semantic
system. Therefore, behavioral measures of motor learning such as motor performance during
retention and transfer tests, amount of forgetting, and reacquisition capability are indicators of the
integrity of the semantic system.
By convention, one class of skilled movements are discrete movements which are
performed rapidly and accurately (Brooks, 1979; Fitts & Peterson, 1964; Woodworth, 1899).
Skilled performance, of a rapid movement with defined spatial and temporal goals, is evident in
the ability to acquire a control program with specified relative timing and to execute this control
program accurately and consistently in both spatial and temporal domains. Therefore, skill
acquisition involves at least two processes: one associated with the construction of the control
program and another associated with the accurate execution required to meet spatial and
temporal goals.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
The schema theory proposed by Schmidt (1975) is one theory that postulates about the
acquisition of movement behavior. From this theory, the construct known as the generalized
motor program (GMP) was defined (Schmidt, 1985). The GMP is an abstract representation of a
movement which is evident as the relative timing or temporal structure of a particular action;
therefore, the GMP governs a class of actions which share the same relative timing. While
certain features of a movement are invariant (i.e. the relative timing), other features are more
changeable such as those controlling spatial and temporal scaling. Movement response
specifications for a GMP such as temporal and spatial features are termed parameters.
According to the GMP theory, improvements in skill occur by enhanced accuracy of the GMP and
increased effectiveness with which the GMP is parameterized. Several recent studies have
demonstrated that programming and parameterization processes are empirically separable
(Sekiya etal., 1994; Wulf & Schmidt 1994; Wulf etal., 1994; Wulf etal., 1993). This dissociation
suggests that different central processing mechanisms are used to specify these two aspects of
movement control (Wulf & Schmidt, 1994).
Schmidt’s schema theory (1975) makes no predictions about how the movement program
is constructed. Instead, his theory describes the behavioral manifestations of programmed
movement However, Arbib (1981) proposes a control theory whereby neurophysiologic central
processes contribute to motor program construction, and encompasses the sensory and motor
interactions inherent in this process. According to Arbib, solutions to a movement problem evolve
as perceptual and motor schema combine to form a control program. Therefore, the process of
skill acquisition (i.e., improvements in speed and accuracy with practice) occurs through a
recursive process such that each practice trial contributes to the development of a localized
network of neurons that subserves fast and accurate movements (Jeannerod et al., 1995).
Single-cell recording work with primates (Sasaki & Gemba, 1981,1982; Kubota & Komatsu, 1985;
Mitz et al., 1991) and recent imaging studies with humans (Grafton, Mazziotta, Presty et al., 1992;
Haieretal., 1992; Jenkins et al., 1994; Rem yetal., 1994; Seitz etal., 1990) provide support for
the recursive nature of control program construction during motor skill acquisition. Activation
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
patterns within this sensorimotor cortical network change and are associated with motor program
acquisition and execution. In particular, the change in activation patterns during motor skill
acquisition, and the extensive connections between motor cortex and other cortical and
subcortical areas such as the basal ganglia, suggests a neural mechanism for motor
programming. Wickens, Hyland, and Anson (1994) propose that a possible mechanism for motor
programming is a cell assembly within the cortex stored as strengthened synaptic connections
between cortical pyramidal neurons.
Evidence does exist that supports the role of the primary sensorimotor cortex, not only in
the control of rapid, programmed movements, but of learning as well. Recent imaging studies in
humans have demonstrated that practice-related improvements in motor learning are associated
with activation pattern changes within sensorimotor cortical areas (Doyon et al., 1996; Grafton,
Mazziotta, Woods, et al., 1992; Grafton et al., 1995, Kami et al., 1995; Seitz et al., 1992). In
particular, the work of Kami et al. (1995) is noteworthy since they found experience-dependent
changes in motor cortex associated with early skill acquisition, which they termed “ fast-learning",
and later skill acquisition, which they termed “ slow-leaming". Fast-leaming processes are
associated with establishing task-relevant motor routines. Slow-learning is associated with
enhanced representations within the motor cortex related to long-term motor skill memory.
Clearly, the sensorimotor cortex has a role in motor skill acquisition but it is not clear how the
sensorimotor cortical system differentially contributes to practice-related changes in motor
performance (i.e., “ fast-leaming” ) from those associated with longer-term retention (i.e., “ slow-
leaming").
It is likely that the performance and learning of motor skills requires the processing
capabilities of both the cortical (i.e., pragmatic) and subcortical (i.e., semantic) motor systems
(Haaland & Harrington, 1990). The capability to construct and parameterize a control program
occurs within the pragmatic system with one aspect of motor memory and learning being an
inherent property of activated neuronal structures. The pragmatic system is primarily responsible
for the evolving perceptual and motor schema and is behaviorally evidenced as motor
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 1
performance improvements which occur with practice. The semantic system is a more highly
integrated processing system providing meaning and relevance to movement; functionally
contributing to the abstract rule learning capability of the implicit motor learning system.
Therefore, motor learning includes the ability to incorporate changes in motor performance with
practice into a representation that is retained and/or generalized to new experiences. The
neurophysiologic evidence would suggest that these capabilities reflect two different modes of
central processing (motor performance, motor learning); however, there have been no studies
which have been specifically designed to investigate the hypothesis that a lesion in the brain areas
primarily associated with the pragmatic system would have a greater effect on motor performance
than motor learning.
An understanding of the neural systems associated with motor control and motor learning
processes may be revealed by studying individuals who have damage to a particular component
of the neural substrate known to be involved with motor skill acquisition. Motor control deficits
after unilateral brain-damage can be revealed by studying the limb ipsilateral to the damaged
hemisphere. The advantage of studying ipsilateral limb movements is that the primary motor
impairments to the ipsilateral limb are not as limiting as those present on the contralateral side,
therefore, the capability to acquire motor skills will not be as directly affected as it would be in the
presence of severe hemiparesis or hemisensory loss. However, of greater import this approach
is advantageous for understanding neurophysiologic processes. Behavioral impairment
differences between participants with and without brain-damage can reveal the function of
lesioned neural areas and provide insight about central processes related to motor control and
motor learning. This approach has been effective in identifying ipsilateral limb deficits after
unilateral brain-damage (Haaland & Harrington, 1987; Jones, Donaldson, & Parkin, 1989; Pohl,
Winstein, & Onla-or, 1997; Smutok etal., 1989; Winstein & Pohl, 1995; Winstein et al„ 1998) and
hemispheric differences in motor control (Fisk & Goodale, 1988; Goodale, 1988; Haaland et al.,
1987; Winstein & Pohl, 1995). These latter studies have revealed that each hemisphere has
specialized functions that contribute to accurate movements. The left-hemisphere has a greater
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 2
role in the timing and sequencing of programmed movements; the right-hemisphere has a greater
role in visual-spatial integration.
Many studies have used brain-damaged populations to investigate the neuroanatomic
basis of motor control processes; however, a major limitation of brain injury research has been a
disregard of the processes involved in long-term retention or motor learning (Haaland &
Harrington, 1990). There have been a very limited number of studies which have specifically
examined the effects of unilateral brain-damage on motor learning. A few studies have
demonstrated that with practice, individuals with unilateral brain-damage can improve movement
speed or accuracy (Bondi et al.. 1993; Cushman & Caplan, 1987; Hanlon, 1996; Platz et al.,
1994), but the motor tasks were slow feedback-driven tasks not those demanding the recall and
recognition processes required in learning programmed movements (Schmidt 1975).
A study by Winstein et al. (1998) investigated the learning of a rapid upper limb
movement with explicit spatial and temporal constraints in individuals with unilateral brain-damage
performing with the ipsilateral limb. Subjects improved in motor performance throughout practice
and retained performance improvements during retention tests. Throughout all experimental
phases (acquisition, retention), the motor performance of individuals with unilateral brain-damage
was less accurate than that of healthy control subjects. There were no interactions between
group and performance blocks for either the acquisition or retention phases suggesting that the
processes associated with motor learning were similar in both groups. Due to methodological
limitations, no conclusions could be derived about motor control processing changes which
contributed to the improvements in performance accuracy in either group or the nature of the
persistent movement errors in the brain-damaged group. In addition, no exact lesion data were
available, thus, limiting any conclusions correlating lesion location/size and motor performance.
Despite these limitations, this study demonstrates the beneficial effects of practice for ipsilesional
limb motor skill acquisition in individuals with unilateral brain-damage. Of even greater interest
are the issues it raises about the type of movement errors which persist in the ipsilateral limb even
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
with extended practice and the effects of unilateral brain-damage on the processes underlying
motor skill acquisition.
The purpose of this study is to investigate the contribution of the sensorimotor cortical
system in the control and learning of a rapid, upper limb movement that is controlled by a single
motor program. Previous work (Winstein et al.. 1998) has suggested that unilateral stroke-related
brain-damage, presumed to involve the sensorimotor areas, affected the control but not the
learning of motor skills. However, specific lesion data and the lack of kinematic analysis
prevented definitive conclusions about the contribution of sensorimotor areas for motor control
attributes. This study will differ from previous work in two important ways: 1) techniques to
quantify lesion location and size will be incorporated to confirm the presence and extent of
sensorimotor cortical system damage, and 2) kinematic analysis will be used to quantify the
changes in control processes across practice and to determine the nature of control deficits
between groups with and without sensorimotor cortical system damage.
Since a major premise of this study is that the sensorimotor cortical system primarily
governs pragmatic system but not semantic system function, it is hypothesized that damage in the
sensorimotor cortical system will affect the control but not the learning of a rapid ipsilesiona! upper
limb movement If the sensorimotor cortical system is not involved in central processes
associated with motor learning, then there should be no difference in the pattern of change as
measured by accuracy for retention or reacquisition phases between groups with and without
sensorimotor cortical system damage. If the sensorimotor cortical system is involved in central
processes associated with motor control, then there should be differences in accuracy between
groups for all phases. Furthermore, differences in motor program acquisition and control may be
related to lesion site. If motor program construction is a recursive process resulting in organized
cell assemblies within sensorimotor cortical areas, then sensorimotor cortical damage should
result in differences in how a motor program is acquired and controlled compared to when the
sensorimotor cortical system is intact In addition, it is hypothesized that hemispheric differences
will be evident If the left-hemisphere is specialized for motor program acquisition and execution.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
then participants with left-hemisphere damage will demonstrate greater programming error
compared to all other groups. If the right-hemisphere has a role in visual-spatial integration, the
participants with right-hemisphere damage will demonstrate greater difficulty in acquiring
continuous movements particularly in the early stages of skill acquisition compared to all other
groups.
Method
Subjects
Forty right-hand dominant (determined by self-report and hand writing preference) adults
voluntarily consented to participate in this study. Participants included 20 individuals with
unilateral hemispheric brain-damage due to stroke (10 right lesions; 10 left lesions) and 20
neurologically healthy controls. Subjects were recruited from the greater metropolitan Los
Angeles area. Inclusion criteria for the participants with stroke were as follows; (1) participation
occurred at least 6 months from stroke onset (2) unilateral brain-damage within the sensorimotor
network due to middle cerebral artery stroke determined from CAT or MRI scan, medical record,
or clinical presentation, (3) had no evidence of any other neurologic condition, and (4) had
functional use of the ipsilateral upper limb. To control for the known effects of age and hand
asymmetries on motor performance (Roy, Kalbfleisch, & Elliott 1994; Pohl, Winstein, & Fisher.
1996), control subjects were matched to individuals with stroke by age and arm used.
Participants in the control group were neurologically healthy and had no evidence of any
musculoskeletal condition that would prevent them from performing the experimental task with the
designated upper limb. The control group consisted of 9 females and 1 1 males (range 43-80 yrs.
M=59.4±11.1 yrs). The stroke group consisted of 4 females and 16 males (range 45-81 yrs.
M=60.2±10.2).
To determine the homogeneity between groups (control, stroke) for visual, educational,
cognitive, motor, and physical function status, all subjects were assessed with the following
measures; (1) visual acuity (Rosenbaum Pocket Vision Screener), (2) peripheral vision of each
eye. (3) years of education, (4) Wechsler Memory Scale - Revised (WMS-R) digit span, (5) WMS-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
R visual span (Wechsler,1987)f (6) Mini-mental state exam (MMSE; Folstein, Folstein, & McHugh,
1975), (7) grip strength, (8) manual dexterity (Blocks and Box test Mathiowetz, Volland,
Kashman, & Weber, 1985), and physical function (physical function sub-scale of the SF-36; Ware
& Sherboume, 1992).
For the participants with stroke, post-stroke onset (i.e., time from stroke onset to study
participation) ranged from 6-147 mos (M=49.9±40.7 mo; median=38 mo). The majority of stroke
participants were 1-5 yrs post-stroke onset (n=12). Two subjects were 6 mos and 6 subjects were
greater than 5 yrs post-stroke onset To determine stroke severity and post-stroke functional
limitations, the upper extremity motor and sensory Fugl-Meyer Assessment (Fugl-Meyer, Jaasko,
Leyman, Olsson, & Steglind, 1975) and the mobility sub-scale of the Functional Independence
Measure (Hamilton, Granger, Sherwin, Zielezny, & Tashman, 1987) were completed on each
stroke participant respectively.
Unilateral stroke was confirmed from MRI or CAT scan reports. In addition, the site and
size of the brain lesion was determined from the actual radiologic image in 19 of the 20 stroke
participants (the actual scan for subject L4 was not available). A board-certified neurologist
outlined the outer brain and lesion areas from the MRI or CAT scan reports of each individual with
stroke. The JAVA video analysis software program was used for image analysis and processing
(Jandel Video Analysis, 1989). Lesion volume and location were determined from the digitized
images. Lesion volume was calculated as the percentage of the total brain volume. Total brain
volume included the cortical and subcortical areas excluding the cerebellum, brainstem, and
ventricles. In order to identify critical neuroanatomic landmarks for the various scan orientations,
the axial incidence (angle of scan orientation from horizontal) of each scan was ascertained.
Individual scans were cross-referenced by axial incidence to neuroanatomic atlases in order to
determine if lesion location involved cortical, capsular, or striatal structures (Damasio, 1995;
Truwit & Lempert, 1994). The greatest extent of each lesion using the functional atlas of Damasio
and Damasio (1989) is displayed in Fig 1. The full lesion extent for each individual stroke subject
is displayed in the Appendix. Lesion volume was not significantly different between right- and left-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
hemisphere lesions (right 11.5 ± 9.9%; left 7.4 ± 8.7%; p = .36). The clinical and neurologic
status of each stroke participant is summarized in Table 1. All subjects had brain damage within
the sensorimotor system (i.e., sensorimotor cortex, its subcortical white matter extensions, and/or
striatal areas) with one exception. Subject R7 had a medial pontine infarct but was included in the
subsequent analyses because of the corticospinal and medial lemniscal involvement associated
with medial pontine syndromes (Rowland, 1991).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
Figure 1. Lesions of the stroke participants. The greatest extent of each lesion is displayed using
the functional atlas of Damasio and Damasio (1989). The location of the central sulcus (filled
arrow) or Sylvian fissure (unfilled arrow) is indicated. Lesion data not available for L4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1. Clinical and neurologic characteristics of stroke participants.
Subject Age (yrs)/ Stroke onset Motor Sensory
number Sex time Fugl -Meyer Fugl-Meyer
(mos) Score t Score t
(max 66) (max 12)
R1 45 31 30 10
male
R2 55 41 15 2
male
R3 55 86 15 0
male
FIM mobility % lesion volume Lesion location
score t
(max 91)
79 2.0 Striatocapsular; frontal
subcortical white (corona
radiata) with extensions into
lentiform nucleus and posterior
limb internal capsule.
80 14.4 Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical
extensions into internal capsule
and lentiform nucleus, extends
to prefrontal, temporal-occipital
association areas.
52 22.9 Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical
extensions into internal capsule
and lentiform nucleus, extends
to prefrontal, temporal-occipital
association areas.
( J >
00
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1 (continued)
R4 62 35 8
female
R5 61 45 61
male
R6 60 6 50
female
R7 71 6 14
male
R8 68 14 53
female
R9 74 28 31
male
2 50 28.4 Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical
extensions into internal capsule
and lentiform nucleus, extends
to temporal-occipital
association areas.
10 90 7.7 Primary motor cortex without
internal capsule; some
extension into putamen,
temporal association, and
medial temporal areas.
8 83 0.4 Lacunar infarct; striate without
posterior limb internal capsule.
missing 79 not applicable Lacunar infarct; medial pontine,
data
12 80 2.8 Capsular infarct with thalamic
involvement; extends within
pyramidal tract from posterior
limb internal capsule to the
midbrain at the crus cerebri.
8 87 7.4 Striatocapsular + cortex; large
parietal infarct extending into
primary sensorimotor cortical
and subcortical areas including
portion of the lentiform nucleus,
some extension into temporal
and limbic areas. &
< o
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1 (continued)
R10 50 48 16
male
L 1 61 54 63
male
L2 59 67 8
female
L3 47 64 29
male
L4 68 12 23
male
1 87 17.2
12 90 0.1
2 76 3.9
7 90 2.7
2 68 scan not
available
Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical
extensions into internal capsule
and lentiform nucleus, extends
to prefrontal, temporal-occipital
association areas.
Primary sensorimotor cortex
only.
Striatocapsular; fronto-parietal
subcortical areas with
extension into lentiform nucleus
and posterior limb internal
capsule.
Striatocapsular; fronto-parietal
subcortical areas with
extension into lentiform nucleus
and posterior limb internal
capsule.
Per CT and MR reports:
Striatocapsular + cortex;
primary sensorimotor and
temporal cortical infarct with
subcortical extensions into
posterior limb internal capsule
and lentiform nucleus.
- '4
o
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1 (continued)
L5 64 147 19
male
L6 47 120 18
male
L7 49 27 64
male
L8 53 124 24
male
Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical
extensions into posterior limb
internal capsule and lentiform
nucleus.
Striatocapular + cortex primary
sensorimotor cortical infarct
extending thru majority of the
fronto-parietal region; includes
subcortical capsular, parietal-
occipital-temporal association,
and prefrontal areas, and
putamen.
Striate without posterior limb
internal capsule.
Striatocapsular + cortex;
primary sensorimotor cortical
infarct with subcortical
extensions into posterior limb
internal capsule and lentiform
nucleus, portions of temporal
association and medial
temporal areas.
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
L9 74 22 61 1 1
male
L10 81 21 8 12
female
R, right-hemisphere lesion; L, left-hemisphere lesion,
t Fugl-Meyer score of the arm contralateral to the hemisperic lesion,
i FIM, Functional Independence Measure (mobility sub-scale).
Striate + cortex; primary
sensorimotor cortical infarct
with subcortical extensions into
anterior portion of corona
radiata and lentiform nucleus,
no posterior limb involvement
per se.
Striate + cortex; primary
sensorimotor cortical infarct
with subcortical extensions into
posterior portion of corona
radiata and lentiform nucleus,
portions of temporal
association areas.
73
Instrumentation and task
A lightweight aluminum lever affixed to a frictionless vertical axle was anchored to a table
top and positioned parallel to the floor (Fig 2). A handle at the distal end of the lever was adjusted
to accommodate for the length of the subject’s forearm such that the subject's elbow was
centered over the axis of rotation. The subject held the lever at the “ home" position, located at 0°
The end of the lever, upon which the elbow rested, was attached to a vertical axle. A linear
potentiometer attached to the base of the vertical axle, transduced lever rotation. Therefore,
horizontal movement of the lever produced a change in voltage corresponding to the lever's
angular displacement This analog signal was converted to digital by an A-D board of a Dell 466v
personal computer and sampled at 200 Hz. The subject’s movement was displayed on a
A B
H a n d le
Subject
Feedback
0 200 400 600 800 1000
T im e (m s)
Figure 2. Experimental set-up with arm lever and feedback display. A) Lower part shows the
subject with lever and feedback monitor. Upper part shows an overhead view of the subject with
arm lever. B) An example of the augmented visual feedback with a graphic representation of the
subject’s movement trajectory for that trial superimposed over the target pattern and the RMSET
score.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
computer monitor. The Template software program (Hary, 1996) was used to signal pre- and
post-response experimental events and store trial data for later analyses.
The motor task was to produce a rapid horizontal plane arm movement with specific
spatial (three elbow flexion-extension reversals) and temporal (1000 ms) goals. Therefore, the
motor skill required a continuous movement of the arm. Participants in the stroke group used the
arm ipsilateral to the brain lesion. Control group participants used the arm as designated by
group assignment (e.g., a left control subject matched to a participant with left stroke, used the
left arm).
After the movement, subjects were presented with either post-response augmented visual
feedback (FB) or no-FB. For the FB trials, FB was displayed on the computer monitor positioned
in front of the subject The augmented feedback was provided as the: 1) trial number, 2) overall
numeric score (root mean square error in relation to target, RMSET \ the difference between the
target and the subject’s trajectory over 1000 ms), and 3) a graphic representation of the subject's
own response over the first 1000 ms of MT superimposed with the goal movement pattern (Fig
2B). For the no-FB trials, the screen remained biank except for the trial number. FB was
distributed across the 100 practice trials as follows: after every trial (100%) for trials 1 -33, after
67% of the trials between trials 34-66, and after 33% of the trials for trials 67-100. This faded FB
schedule provides high FB presentation in early practice which is gradually reduced (i.e. faded)
throughout a 100-trial practice session to result in an average relative frequency of 67% for the
entire session. The FB schedule was selected because FB presented with reduced relative
frequency is known to be more beneficial for motor learning than one with 100% FB frequency
(Winstein & Schmidt 1990).
'R M S E r^ If r -T g
-uM 200*
x, = subject’s position at time i
T = target position at time i
* the number of samples in the target array
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
The computer generated pre-movement time and response/post-response events for
each trial are shown in Fig 3. Figure 3A depicts the pre-movement time events. During the pre
movement interval, the target was displayed to the subject At the end of the pre-trial interval, the
target was erased. The subject was presented with a “Ready” tone and a yellow visual cue
displayed for 300 ms. After a 300 ms pause, the subject was presented with a “ Go” tone and a
green visual cue. In order to encourage programmed movement selection, the subject was forced
to move within 500 ms of the “Go” cue. If the subject moved prior to 150 ms of the “ Go” cue, the
subject received an error message that the movement started too early. If the subject moved
after 500 ms of the “ Go” cue, the subject received an error message that the movement started
too late.
Acceptable response
onset interval
150-500 ms from'Go*
Pre-trial interval
-2000 ms 0 300 600 750 1100 ms
Target displayed “ Too early" “ Good” “ Too late*
B
Acquisition collection time FB delay FB presentation Post-FB delay
(5000 ms) (2000 ms) (4500 ms) (2000 ms)
Movement time
Baseline detection
threshold crossed.
(-75°)
MT ends when
subject crosses 0°
Inter-trial interval
Figure 3. Schematic of the pre- and post-movement events. Top panel (A) depicts the pre
movement time events. The lower panel (B) depicts the response and post-response events for
the FB trials. For the no-FB trials, the post-response interval was displayed with no FB for 4500
ms. Note that time is not to scale.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
Figure 3B depicts the response and post-response events. The trial started when the
subject crossed the 0.75° baseline detection threshold and ended when the subject crossed 0°
Acquisition collection time was 5000 ms. Pcst-response events depended on whether the trial
was FB displayed or no-FB displayed. For FB displayed trials, a 2000 ms FB delay period was
followed by a 4500 ms post-response FB period. The screen was erased for a 2000 ms post-FB
duration; ending the trial. For no-FB displayed trials, the post-response duration was 4500 ms
during which the subject was only presented with the trial number therefore, the post-response
interval was 4000 ms shorter for no-FB compared to FB trials.
Procedure
Subjects read and signed an informed consent prior to starting the experimental protocol.
Subjects were seated in front of the computer monitor with their forearm resting on the lever in
approximately 90° of elbow flexion (Fig 2A). Subjects were instructed to move the lever in order
to replicate the goal movement pattern displayed on the monitor. A sample pattern was visually
displayed to orient the subject to the augmented error FB provided on the monitor however, the
subject was not allowed to move the lever prior to data collection. All subjects were informed that
1) FB presentation would be higher in early practice and gradually reduced as practice proceeded,
and 2) the second day would include two sessions similar to the practice phase on Day 1; subjects
were not explicitly informed that FB presentation was to be manipulated on Day 2.
Subjects practiced for 200 trials with a 30-minute break after the first set of 100 trials.
One day later, subjects returned for two 20-trial retention tests. For the no-FB retention test
neither RMSET or the subjects trajectory was presented during the post-response period; instead
the screen remained blank except for the trial number display. For the FB retention test the trial
number, RMSET and the subjects response was presented on the screen as experienced during
FB practice trials.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
Statistical analyses and dependent measures
Motor control and learning
Motor performance was measured as the root mean square error between the subject’s
total movement trajectory and the goal movement (RMSEs). The RMSEs 2 is a measure of
accuracy calculated as the average difference between the subject’s pattern and the goal
movement pattern over the subject’s total movement time (MT).
Differences in performance and learning were determined by grouping individual RMSEs
trial data into 10-trial blocks for the acquisition and delayed-retention phases. Performance
groupings included the acquisition phase (Blocks 1-20), no-FB retention phase (Blocks 21-22),
and FB retention phase (Blocks 23-24). For task acquisition, a 2 Group (control, stroke) x 20
Block (Blocks 1-20) analysis of variance (ANOVA) with repeated measures on the last factor was
used. For each retention test, a 2 Group (control, stroke) x 2 Block analysis of variance (ANOVA)
with repeated measures on the last factor was used. To further examine motor learning
differences between groups, specific comparisons were made to determine group differences in
forgetting (end of acquisition to first no-FB retention block, Blocks 20 and 21) and savings
(beginning of acquisition to first no-FB retention block, Blocks 1 and 21; beginning of acquisition to
first FB retention block, Blocks 1 and 23). A 2 Group (Control, Stroke) x 2 Block repeated
measures ANOVA was completed for each of these phases for RMSEs.
A Group main effect during acquisition or retention phases would suggest that differences
in motor control exist between those with and without sensorimotor system brain damage. A
Group x Block interaction in the aforementioned phases would suggest that differences in motor
learning exist between those with and without sensorimotor system brain damage. The
Greenhouse-Geisser degrees of freedom adjustment was used to compute the probability level
2 RMSEs = I K-.1t?
ns
x, = subject's position at time i
T = target position at time i; it is assumed that target values >1000 ms are 0°
ns = number of samples in the subject’s movement time
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
for the repeated measure factors. Post-hoc comparisons were performed to determine the locus
of any significant interaction effects using a Student’s independent samples f-test
Separate f-tests were conducted to assess group differences for the visual, cognitive,
motor, and functional tests. For all statistical tests, significance were set at p < .05.
Acquisition and execution of motor programs
Previous work has revealed that during motor skill acquisition individuals with stroke
perform with significantly greater error compared to matched controls (Winstein et al.,1998). To
determine if control differences between groups is due to differences in how a motor program is
acquired and executed, several analyses were completed. These analyses included changes
across acquisition in motor program accuracy, the number of discontinuous to continuous
movements, and the capability to acquire and execute a one-unit action.
Motor program accuracy
Since RMSEs is a global error measure (Schmidt, 1988), the residual-RMSE (rRMSE)
method described by Wulf et al. (1993) was used to separate the motor programming, temporal
scaling, and amplitude scaling errors of this total error. Based upon the rRMSE method of Wulf et
al. (1993), a computer software program (R. Sullivan, 1997) was developed to scale (stretch or
compress) the subject’s trajectory in time and amplitude in order to maximize the agreement
between the subject’s movement and the goal movement pattern. Specifically, the subject-
produced trajectory array and target array were synchronized when the subject’s movement
exceeded 0.75° displacement from baseline. The subject’s overall movement trace was
proportionately scaled from 0.2 to 2.0 by a time factor increment of 0.1 such that the subject’s
trace was interpolated to the number of samples in the target trace. This procedure normalized
the subject’s trajectory to the temporal goal essentially removing the temporal error component of
the subject’s movement. After scaling the subject’s trajectory in time, a similar procedure was
applied for amplitude which reduced the spatial error component of the subject’s movement. The
specific details of this scaling procedure and the resultant temporal and spatial parameter error
measures will be discussed in Chapter 6. After temporal and spatial scaling, the difference
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
between the subject’s scaled trajectory and the target array resulted in the rRMSE. The rRMSE is
considered to represent the accuracy of the motor program (Wulf et al., 1993). Figure 4 illustrates
the scaling process that results in RMSEs (subject’s total error prior to scaling) and rRMSE
(subject's programming error once temporal and spatial scaling is performed).
RMSEs 38.8
64
c
■
rRMSE 4.0
32 •a
•
E
E
o
e
•
E
o
u
( O
-32
a
o 400 800 1200 1600
Time (ms)
Figure 4. Example of movement trajectory scaling. The goal movement is displayed as the thick
black line. The subject’s actual movement is displayed as the thin black line (tic marks indicate
peak displacement for reversals). RMSEsis the difference, in degrees, between the goal pattern
and subject’s movement The rRMSE is the difference in degrees between the goal pattern and
the subject’s trajectory after the subject's trajectory is scaled in time (dashed line) and then
amplitude (dot-dashed line).
Kinematic analysis
In order to assess changes in the movement trajectory across the acquisition phase,
kinematic analyses were completed. Of the 200 total trials of practice, specific kinematic
measures were collected for blocks at the beginning (trials 1-10), middle (trials 91-100), and end
(trials 191-200) of practice Velocity and acceleration profiles were derived from the position data
using DATA-PAC III software from Run Technologies (Datapac, 1997). Position data were
conditioned using a low-pass filter with a 10-Hz cutoff frequency. Velocity and acceleration were
derived from a differentiation process such that the amplitude value for the middle data point in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
each set of 3 points was replaced with the slope of the regression line calculated for a minimum
interval. The minimum interval was equal to 2 times the sample period (5 ms); therefore, the
slope value was obtained over a 10 ms constant time interval.
Kinematic dependent variables were determined for the trials within the specific
acquisition blocks described above. Movement time (MT) was the absolute trial duration from the
initial zero velocity crossing to the last zero velocity crossing. The velocity profile was full-wave
rectified and then averaged to determine the absolute average velocity for each trial. Five
temporal variables were determined from the acceleration profile which included the absolute time
to the 1st 2nd, 3rd, 4th, and 5th acceleration peaks. Figure 5 displays the displacement velocity,
and acceleration profiles for the goal movement
Discontinuous movements
Differences in the process of skill acquisition between groups was also determined by
assessing the number of discontinuous movements across the acquisition phase. Several studies
have demonstrated that as rapid movements become more skilled velocity and acceleration
profiles evolve from a discontinuous to a more continuous pattern (Brooks & Watts, 1988; Brooks
et al., 1995; Darling & Cooke, 1987; Marteniuk & Romanow, 1983; Moore & Marteniuk, 1986).
The velocity and acceleration profiles for each trial were reviewed. If there was an observable
discontinuity within the velocity profile that was accompanied by an extra zero-crossing in the
acceleration trace then the trial was determined to be discontinuous (DC). In other words, a trial
was considered to be DC, if there were greater than 4 zero crossings in the acceleration-time
profile (i.e., greater than the number of zero crossings that occur in the goal movement see
acceleration-time profile in Fig 5) and a discontinuous velocity profile was visually evident The
number of continuous and DC movements were tallied for all subjects across the 200 acquisition
trials.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
70
c
(D
E
s
0
a
0 )
8
■ §
o
ca
c
o
E
0
9
300 -
o'
0 a
■ § >
0 w
| 0 - ? «
I S
o
-300 -
4 0 0 0 - b
-4000-
0 10 175 430 760 985 1000
Time (ms)
Figure 5. Displacement-, velocity-, and acceleration-time profiles for the goal movement The
temporal locations of landmarks a (movement onset), b (1st peak acceleration), c (2nd peak
acceleration), d (3rd peak acceleration), e (4th peak acceleration), f (5th peak acceleration), and g
(movement offset) are indicated. The times of these respective landmarks are displayed on the x
axis. The average velocity for the goal movement is 175 deg/sec.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
Unit structure correlation analysis
To determine if the movement was a single, programmed unit or if in fact the action
evolved to a one-unit action, the unit-analysis method described by Young and Schmidt (1990)
was applied. The acceleration-time profiles were used to determine seven kinematic landmarks.
These landmarks were present in every trial and were evenly distributed across the movement.
Figure 5 illustrates the landmarks labeled a through g from the earliest to the latest kinematic
event The time of occurrence for each landmark was determined and correlations were
computed between movement onset (a) and the time of each successive landmark, and between
the various landmark times and movement offset (g). Therefore, two sets of correlations were
calculated. The first set correlated absolute onset time with the absolute time of each successive
event The second set correlated the absolute offset time with the preceeding landmark events.
These onset and offset correlation sets were calculated for trials 1-10, 91-100, and 191-200 which
represented the beginning, middle and end of practice, respectively.
Using a task similar to the one used in this study Wulf et al. (1993), determined that an
action, a movement with a definable beginning and end, was considered to be a one-unit action
(i.e., governed by a single motor program) if the correlations within an action were relatively large
(near 1.0) or if the correlations gradually changed across the action (i.e., interval differences were
less than or equal to .20). This method suggests that two different processes relate to the
acquisition and execution of programmed actions. Abrupt changes across all correlation pairs
would suggest that the action was governed by more than one-unit; however, higher overall
correlations would suggest that the unit-action was executed with greater proficiency. For the
purposes of this analysis, each method (i.e., gradual change in correlations across an action and
overall correlation) were used to assess programming ability.
The average correlation across all correlations within the trial block was determined for
each group. In addition, the pattern of correlations was examined for abrupt changes between
correlation pairs. An abrupt change was defined as one where the change between correlation
pairs was greater than .20. For example, if the difference between correlation pairs (e.g., a-b, a-c,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
a-d) was less than or equal to .20 throughout the action then the movement would be considered
a one-unit action governed by a single motor program. If there was a difference greater than .20
between any pair of correlations this would indicate a pause in the action suggesting that the
movement was not executed as a single motor program.
In order to assess changes in motor program accuracy throughout practice, individual tiial
data for rRMSE was grouped into 10-trial blocks for the acquisition phase. A 2 Group (Control.
Stroke) x 20 Block (Blocks 1-20) ANOVA with repeated measures on the last factor was
completed for rRMSE. In a separate analysis, the total number of DC movements for the
acquisition phase was analyzed in a 2 Group x 4 Block repeated measures ANOVA. This analysis
included the following 4 blocks: Trials 1-50, 51-100, 101-150, and 151-200.
In order to assess changes in motor control at designated points during practice, a 2
Group (Control, Stroke) x 3 Block (beginning, Trials 1-10; middle, Trials 91-100; end, Trials 191-
200) repeated measures ANOVA was completed for each kinematic variable. The Greenhouse-
Geisser degrees of freedom adjustment was used to compute the probability level for the
repeated measure factors. A Student’s independent samples f-test was used for post-hoc
comparisons to determine the locus of any significant interaction effects.
Effects of lesion location
To assess the effects due to lesion location, two levels of analysis were used. First,
Group (Control, Stroke) x Arm-used (right, left) x Block repeated measures ANOVAs were
calculated with RMSEs for the acquisition and retention phases and with rRMSE for the acquisition
phase. Group (Control, Stroke) x Arm-used (right, left) x Block repeated measures ANOVAs were
also calculated for the kinematic variables for the beginning, middle, and late acquisition blocks.
The total number of DC movements during acquisition was analyzed in a Group x Arm-used x
Block repeated measures ANOVA for trial blocks: Trials 1-50, 51-100,101-150, and 151-200. Of
particular interest were the Group x Arm-used interactions since this interaction would suggest
that differences were due to the side of hemispheric damage and not simply group membership or
handedness alone.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
In addition, to determine if differences could be due to sensorimotor cortical system
damage regardless of hemisphere, a 4 Lesion-area (corticahtsubcortical, subcortical, pontine, no-
lesion) x 2 Arm-used factorial ANOVA was completed for rRMSE and the number of DC
movements. Participants were grouped into one of three neuroanatomic categories: (1)
Subcortical (SC) - lesion limited to sensorimotor subcortical structures only (i.e., capsular, striate,
and/or thalamic), (2) Cortical±subcortical (Cx±SC) - lesion included sensorimotor cortex with or
without subcortical structures, (3) Pontine - included the one subject with the medial pontine
infarct, and (4) Control - no lesion. These groupings were selected since cell-assemblies within
the cortex are hypothesized to be a neural mechanism for motor programming (Wickens et
al.,1994). If a Group effect was found in this analysis with the locus of the effect being between
the Cx±SC group and other areas, this effect would support the cortical cell-assembly hypothesis.
RESULTS
Impairment and disability
Group mean comparisons are summarized in Table 2. There were no differences
between groups for age or educational level. The participants with stroke were more impaired in
visual acuity (p=.02); however, this did not interfere with their ability to see feedback displayed on
the computer monitor. Peripheral vision did not differ between groups (left, p=.07; right p=.23).
The participants with stroke were more impaired than the control subjects in all memory tests
(p< 05) except forward visual span (p= 11), and were more impaired in cognitive function as
measured by the MMSE (p< 01). Functionally, the stroke group was more physically disabled as
measured by the physical function subscaie of the SF-36 (p<.001). In order to control for known
strength differences based upon age, gender, and hand dominance, individual strength values for
the right and left hand were compared to normative values and the difference from the normative
value was recorded (Mathiowetz, 1990). By comparing the differences between groups by hand,
a clearer depiction of ipsilateral and contralateral motor impairment was evident (Table 3). As
expected, stroke participants were significantly weaker on the contralateral side compared to
controls confirming the presence of hemiparesis. Ipsilateral grip strength was significantly less in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
the right stroke group compared to their matched controls but ipsilateral grip strength did not differ
between the left stroke and control groups. Upper limb dexterity was significantly impaired for
both the contralateral and ipsilateral arms in the right- and left-stroke groups compared to
controls.
Table 2. Group comparisons (excluding motor).
Assessment
Group
Control t Stroke t
P
Age (yrs) 59.4±11.1 60.2±10.2 .80
Male/female ratio 11/9 16/4
Visual acuity 20123 20/32 .02*
Peripheral vision (deg)
Left 77.5±5.3 71.5±13.1 .07
Right 76.8±4.7 73.3±11.7
.23
Years education 15.4±3.5 13.3±3.5 .07
Memory:
Digit span (n=15)
forward (12 max) 8.6±1.9 6.8±1.9 .01*
backward (12 max) 6.4±2.1 4.9±1.5 .04*
total (24 max) 14.9±3.8 11.7±3.1 .01*
Visual span (n=19)
forward (12 max) 8.2±1.9 7.1±2.3 .11
backward (12 max) 7.9±1.8 5.9±2.5 .01*
total (24 max) 16.1±3.0 13.1 ±4.2 .01*
Cognitive: (n=19)
MMSE (max 30) 28.9±1.9 25.2±4.2 < .01*
Physical function:
SF-36 (max 100) 92.8±18.0 39.0±30.4 < .0001*
values are mean ± SD.
* p < .05
t n=20.
t n=20 except where noted. Stroke subjects not tested due to expressive aphasia.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3. Group by hand upper limb motor comparisons.
Assessment Control
Group
Stroke
P
Riaht fn=9Vf Riaht (n=10)
Grip (lbs) **
Right 6.0±10.2 -13.4±24.8 04*
Left 11.7±11.54 -55.5±29.21 < .0001*
Box & Blocks (# in 1 min)
Right 67.0±6.7 55.0±11.1 <.01*
Left 62.1 ±8.3 8.7±15.3t < .0001*
Left (n=9)t Left fn=101
Grip (lbs) **
Right 7.5±16.8 -62.6±37.9 t < .001*
Left 14.3±14.8 2.0±18.6 .13
Box & Blocks (# in 1 min)
Right 67.4±6.4 14.3±23.1 t < .0001*
Left 66.0±7.4 57.5±8.5 .03*
values are mean ± SD.
* p < .05
** Grip strength values are reported as the difference from normative values controlled for age,
gender, and hand.
t Control subjects not tested due to upper limb limitations in the non-experimental limb such as
arthritis.
t Indicates the hemiparetic arm.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
Control and learning outcomes
In order to distinguish between motor performance and motor learning (Salmoni et al.
1984), the acquisition and retention phases were analyzed separately. Consistent with previous
findings (Winstein et al., 1998), there were no interactions of group with block for RMSEs in any of
the experimental phases but group differences did exist throughout as revealed by group main
effects evident in each phase.
The group block RMSEs means for the acquisition (Blocks 1-20), no-FB retention (Blocks
21-22), and FB retention (Blocks 23-24) phases are summarized in Fig 6. For both the stroke and
control groups, accuracy improved across practice (Block effect F(5,188)=16.62, p<001). Stroke
participants improved accuracy as indicated by RMSEs scores that improved from 35.1 deg in
early practice to 22.7 deg by the end of practice. Control subjects demonstrated similar practice-
related improvements in accuracy from 22.8 deg in early practice to 13.7 deg by the end of
practice.
To specifically examine differences in learning between groups with and without
sensorimotor system brain-damage, several separate analyses were completed. Performance
during the acquisition phase provides evidence of improved motor performance; however, it may
only reflect short-term changes biased by the temporary, though positive, influence of practice and
FB (Schmidt & Bjork, 1992). Therefore, differences between groups were assessed for delayed
retention (Blocks 21 and 22), reacquisition ability (Block 23 and 24), forgetting (Block 20 and 21),
and savings (Blocks 1 and 21; Blocks 1 and 23). Table 4 summarizes the change in RMSEs
between each of the aforementioned phases for each group. Statistically, there were no Group x
Block interactions for any of these phases (p>05). The pattern of change across practice and
persistence of that change during retention is similar between groups. This suggests that there is
no difference in motor learning due to sensorimotor cortical system damage.
Despite the preservation of motor learning capability, motor control deficits are evident in
the brain-damaged group. During all phases (acquisition and retention), the participants with
sensorimotor cortical system damage performed with more error (increased RMSEs) compared to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
the control group (Group main effect acquisition, F(1,38)=15.34, p< 001; no-FB retention,
F(1,38)=9.29, p<01; FB retention, F(1,38)=12.41, p<.01). Thus, differences in motor
performance were evident between groups; participants with sensorimotor cortical system
damage using the ipsilesional limb were less accurate than controls.
40
Control
Stroke
a> 30
a >
2 ,
t o
U J
C O
s
0C 20
10
• •
• • •
▲ ▲
▲ i
/A-
2 4 6 8 10 12 14 16 18 2021 22 23 24
NO-FB FB
Retention
(10-trial blocks)
Acquisition
Figure 6. Control and stroke group RMSEs block means for acquisition (Blocks 1-20), no-FB
retention (Blocks 21-22), and FB retention (Blocks 23-34) phases. A 30-min break occurred
between blocks 10 and 11. RMSEs represents overall accuracy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 4. Performance and learning effects indicated by change in block mean RMSEs by Group between acquisition and retention
phases.
Group
Practice:
Acquisition
Block 1 -
Acquisition
Block 20
Retention:
no-FB Retention
Block 21 -
no-FB Retention
Block 22
Reacquisition:
FB Retention
Block 23 - Block
24
Forgetting:
End acquisition
Block 20 -
no-FB retention
Block 21
Savings:
First acquisition
Block 1 -
no-FB retention
Block 21
Savings:
First acquisition
Block 1 -
FB retention
Block 23
Control
19.1
0.0
11.5 13.8 15.3 17.9
Stroke
112.4 10.4 10.6 13.5 18.8 110.9
RMSEs, root mean squared error of subject's trajectory. Units are in degrees. Blocks are 10-trial means.
Arrows indicate the direction of change from the earlier to the later block.
In all comparisons: Group main effect, p<01; Group x Block interaction, p>,05.
00
< o
90
Motor programming outcomes
Motor program accuracy
Residual RMSE group block means for the 200 trials of practice are displayed in Figure 7.
If the error difference between groups is due to parameter scaling only and not motor
programming, then there should be no group difference in rRMSE once the subject’s trajectory
has been normalized for temporal and spatial error. However, stroke subject's had higher rRMSE
than controls subjects (stroke: 10.9 deg; control: 8.8 deg). This resulted in a main effect for group
(F(1,36)=7.12, p = .01) suggesting that differences in motor programming accuracy exist between
those with and without sensorimotor cortical system damage. Both groups improved in motor
programming accuracy with practice (Block main effect F(5,188)=16.62, p<0001) and this pattern
o >
©
" O
U J
C O
(0
3
2
C O
©
c c
15
10
± Control
• Stroke
• •
• •
A A
A A
• • •
A A a A A
2 4 6 8 10 12 14 16 18 20
Acquisition
(10-trial blocks)
Figure 7. Control and stroke group residual RMSE block means for the acquisition (Blocks 1-20)
phase. A 30-min break occurred between blocks 10 and 11. After rescaling in time and amplitude
to reduce parameterization errors, the residual RMSE represents the accuracy of the motor
program. Group effect, p=.01; Block effect, p<.0001; Group by block interaction, p=.10).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
of change did not differ between groups across acquisition (Group x Block interaction:
F(6,213)=1.81, p = .10).
Individual trial data from a representative control subject and 2 stroke subjects during
early and late practice are displayed in Fig 8. Control subject JA and stroke subject R1
demonstrate improvements in overall accuracy and motor program accuracy with practice (i.e.,
decrease in RMSEs and rRMSE). It is evident in subject R1 during late practice that a large
proportion of the total error is due to temporal and spatial scaling error once the subject’s
trajectory is scaled the magnitude of rRMSE is low (5.7 deg). However, subject R6 demonstrates
both programming and scaling error during late practice. While a large proportion of the total
error appears to be due to scaling ability, the rRMSE is relatively high (i.e., greater than the stroke
group rRMSE mean) suggesting a greater degree of motor programming error.
In summary, throughout the acquisition phase the stroke group demonstrated higher
rRMSE (less motor program accuracy). Post-hoc analysis comparing motor program accuracy
between Blocks 1-10 and Blocks 11-20 revealed a difference in the pattern of acquisition between
groups in the early compared to the later practice blocks. During the first 100 trials, the stroke
group decreased motor program error by 29% (rRMSE: Block 1, 14.8 deg; Block 10,10.5 deg).
Motor program error decreased by 14% in the control group (rRMSE: Block 1, 10.6 deg; Block 10,
9.1 deg). This resulted in a group by block interaction for Blocks 1-10 (F(4,136)=2.47, p = .05). In
contrast the pattern of change was not different between groups in the last 100 trials of practice
(Group x Block interaction: F(4,136)=1.41, p = .22). However, consistent with the previous
analysis the stroke group performed with greater residual programming error than the control
group throughout early and later practice (Group main effect Blocks 1-10, F(1,36)=4.31, p = .05;
Blocks 11-20, F(1,36)=9.42, p < .01). These findings suggest that after sensorimotor cortical
system damage the capability to acquire motor programs appears to be retained to some degree;
however, the proficiency and development of motor programs is impaired.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Early
Late
92
RMSES17.8
R M S E s 9 .2
Control
JA
rRMSE 9.2
1000 isoo 500 3 isoo
rRMSE 4.1
32
0
300 •00 900 1200
RMSEstO.S
Stroke
R1
RMSE*34.3
8 32
32
10 00 ISOO 500 0 9 00 1 0 0 0 I S O O 9
Stroke
R6
RMSES37.5
> rR M S E nt.a
ISOO 2000 1000 500 0
RMSEs39.9
Tims (ms)
32
0
-32
500 000 1 9 0 0 0 II
Urns (ms)
Figure 8. Individual subject trial data. Rows show data from a control (top) and two stroke
subjects (bottom) during early (left) and late (right) practice. The goal movement is the thick black
line. The subject’s actual movement (thin black line), time-scaled movement (dash line), and
amplitude-scaled movement (dot-dash line) are displayed. The trial error scores for the unsealed
(RMSEs ) and scaled (rRMSE) trajectories are in the table.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
Kinematics
The process of skill acquisition is best revealed by examining changes in the kinematics
that occur over the course of practice. Representative examples of the displacement-, velocity-,
and acceleration-time profiles for a control (Fig 9) and stroke (Fig 10) subject provide a visual
summary of the movement trajectory changes that occurred with practice. With practice,
movements became less variable and more continuous. Movement times progressively
decreased for the stroke subject across practice. For the control subject, MT was within 200 ms
of the goal movement by the middle to end of practice. By the end of practice, both the control
and stroke subject presented with acceleration profiles that approximated the pattern of the goal
movement (see Fig 5). Table 5 summarizes the group kinematic variable means and standard
error of the means for beginning (Block 1, Trials 1-10), middle (Block 2, Trial 91-100), and end
(Block 3, Trials 191-200) of practice.
Movement time. Throughout practice, the stroke group had longer MTs compared to
controls (control: 1253±305 ms; stroke: 1719±684 ms) resulting in a Group main effect
(F(1,36)=31.7, p< .0001). For both groups, MT decreased with practice (Block main effect
F(1,41)=51.4, p< .0001). The stroke group decreased MT by over 1000 ms between Blocks 1 and
2 contributing to a significant group by block interaction (F(1,41)=17.0, p< .001).
Average velocity. Control subjects achieved faster speeds than the stroke group. The
average velocity of the control group was 171±29 deg/sec while the stroke group averaged
143±41 deg/sec (Group main effect: F(1,36)=15.4, p< .001). The main effect for Block and the
Group by Block interaction were significant (Block main effect F(2,60)=16.9, p< .0001; Group x
Block interaction: F(2,60)=3.45, p = .05) but this undoubtably was due to the stroke subjects who
progressively increased movement speed from the beginning to the end of practice (see Table 5).
In contrast, control subjects actually increased movement speed in Block 2 prior to decreasing
movement speed to come within 1 deg/sec of the average goal movement velocity in Block 3.
Temporal variables. The absolute time to each of the 5 peak accelerations were
determined because these variables represent key kinematic landmarks within the goal
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
Q .
v
1440 ms
0
Q .
C D
b
o
o
<
0 1120 ms
CL
X.
1295 ms
Time (ms)
Figure 9. Displacement-, velocity-, and acceleration-time profiles for a control subject (JA). Each
panel displays 10-trials from the A) beginning (Trials 1-10), B) middle (Trials 91-100), and C) end
(Trials 191-200) of practice. Movement onset is synchronized at 0 ms. The vertical marker on
the right indicates the end of movement for each trial.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A
95
Q .
ja
a
> »
o
o
£
o
£
4615 ms
0
CL
jn
Q
. V .
V
o
0 2150 ms
C L
*n
Q
>*
o
o
$
1630 ms
Time (ms)
Figure 10. Displacement-, velocity-, and acceleration-time profiles for a stroke subject (R4). Each
panel displays 10-trials from the A) beginning (Trials 1-10), B) middle (Trials 91-100), and C) end
(Trials 191-200) of practice. Movement onset is synchronized at 0 ms. The vertical marker on
the right indicates the end of movement for each trial.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 5. Mean (+SE) for Kinematic variables by group across practice blocks.
Block 1 (Trials 1-10) Block 2 (Trials 91-100) Block 3 (Trials 191-200)
Variable Control Stroke Control Stroke Control Stroke
MT (ms)a b c 1436(103) 2399 (162) 1171 (27) 1402 (69) 1152 (21) 1355 (71)
Average velocitya b d
(deg/sec)
160 (8) 114(7) 180 (7) 155 (9) 174 (4) 159 (9)
No.of discontinuous
movementsa b c 2.25 (.73) 7.15 (.71) .20 (.16) 1.25 (.48) .30 (.30) 1.05 (.48)
Absolute time (ms) to:
1st peak acca a 87(6) 160 (34) 76(5) 91 (5) 77 (3) 90(7)
2nd peak acca b 0 331 (21) 545 (47) 281(9) 326 (16) 273 (7) 305(17)
3rd peak acca b c 658 (61) 1191(120) 493(15) 601 (35) 481 (10) 569 (34)
4th peak acca b 0 1016 (76) 1685(126) 782 (23) 936 (53) 747 (17) 882 (46)
5th peak acca b 0 1330(107) 2245(153) 1044(25) 1255 (62) 1025(17) 1183 (55)
MT = movement time, acc = acceleration.
a Group main effect, p < .001;b Block main effect, p < .0001,c Group x Block interaction, p < .001;
d Group x Block interaction, p = .05; * Group x Block interaction, p = .01.
$
97
movement and will be used for the forthcoming unit-analysis (see Fig 5 for acceleration temporal
landmarks). The probability for the Group and Block effects and the Group by Block interactions
are provided in Table 5. To summarize, this Table reveals that all subjects decreased absolute
times across Blocks. However, the control subjects achieved each temporal landmark earlier
than the stroke subjects.
Discontinuous movements.
The mean number of DC movements for the beginning, middle, and end of practice for
each group and the probability for each effect is presented in Table 5. Paralleling previous
findings, a repeated measures ANOVA revealed significant main effects for Group (stroke group
made more DC movements than the controls) and Block (all subjects decrease DC movements
across practice), and a significant Group by Block interaction (due to the stroke subjects
decreasing DC movements substantially between Blocks 1 and 2). However, the magnitude of
the group differences was best revealed by comparing the total number of DC movements across
all 200 practice trials. Figure 11 displays the total number of DC movements for Trials 1-50, 51-
100,101-150, and 151-200. For the stroke group, 22.2% of all trials had DC movements (888 out
of 4000 trials). In contrast, 4.9% of the movements made by control subjects contained DC
movements (196 out of 4000 trials). This resulted in a significant main effect for Group
(F(1,36)=13.7, p < .001). In addition, a Group by Block interaction was present (F(2,55)=7.59, p <
.01). Post-hoc analysis revealed that the stroke group made significantly more DC movements
compared to controls during trials 1-50 (f=-4.49, p <.001) and trials 51-100 (t=-2.93, p <01).
In summary, the kinematic analyses reveal that with practice both groups were able to
increase movement speed, approach temporal landmarks consistent with the goal movement
pattern, and decrease the number of DC movements. Despite the substantial improvement in
motor performance (i.e., faster, increased accuracy, and more continuous movements) with
practice, the stroke group consistently performed with less proficiency compared to controls.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
23 C ontrol
50-Trial Blocks
Figure 11. Frequency histogram of the total number of discontinuous movements for each group
during 1) Trials 1-50, 2) Trials 51-100, 3) Trials 101-150, and 4) Trials 151-200. Group effect
p< 001; Block effect p<.000T, Group x Block interaction, p< 01. The asterisk (*) indicates the
locus of the interaction effect
Unit-analvsis
The purpose of the unit-analysis was to provide a description of programmed actions in
order to examine changes in motor control with practice. Using the seven temporal landmarks
delineated in Figure 5, within-subject correlations were averaged for each pair of correlations (i.e.,
between movement onset (a) and movement offset (g) and each of the peak accelerations).
Figure 12 illustrates the pattern of correlations for the Control and Stroke groups for Block 1
(Trials 1-10), Block 2 (Trials 91-100), and Block 3 (Trials 191-200). If points a through fand b
through g were landmarks within a one-unit action or motor program, then the interval difference
between these pairs of correlations should be less than or equal to .20. However, interval
differences greater than .20 would indicate a pause in the action suggesting that the movement
was not executed as a single motor program but involved multiple units.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
T rials 1-10
1.00
S troke
Control
0.80
5 0 60 5 0-60
0.40 0.40
0.20 0.20
0.00 0.00
» - d a - f g - b g - c g - d g -c g - f
Trials 91-100 r - .7 7
1.00
0 .8 0 - [:
t -
M 0.80 ^
« p
•
o 0.40 T '& \
O - t r
0.20
0.00
t - b a - c a - d * . « a - r g -b g - c g -d g -a g -f
r - .6 6
5 0.60
a ~ b a -c * - d a - e a - f g - b g - c g - d g -c g - f
T rials 191-200
7 8
if:
Si
• - ’ S ?
i
n
i i
a - b a « c « -d a -# a - f g -b g - c g - d g -c g -f
r - .7 5
£ 0.60
0 .0 0
a -b a -c « - d a -a a - f g -b g -c g - d g -c g -f
Figure 12. Averaged within-subject correlations (+SE) between the start of movement (a) and
kinematic landmarks (b through f) and the end of movement (g) and kinematic landmarks (b
through f) for the Control (left) and Stroke (right) groups during the beginning (Trials 1-10), middle
(Trials 91-100), and end (Trials 191-200) of practice. The asterisk (*) indicates intervals between
correlation pairs greater than .20. The average correlation for each group of comparisons is
indicated.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
In Figure 12, an asterisk indicates an interval change greater than .20. During Block 1.
both the control and stroke groups demonstrated a break in the pattern of correlations, thus,
indicating a movement comprised of more than a one-unit structure. The control group executed
the action in two units of behavior; the first corresponds to the 1st peak acceleration, the second
corresponds to the acceleration events c f , e, and f that comprise the 3rd displacement reversal
and return to the home position. The stroke group also produced movements that had more than
a one-unit structure; however, the first unit corresponds to the acceleration events b and c that
comprise the first displacement reversal (from movement onset to the 1st peak displacement).
The second unit included the same kinematic landmarks as was observed in the control group.
According to Young and Schmidt (1990), this type of movement represents two units that are
strung together with a period of information processing that separates the initial from the final
action unit Therefore, it appears that the actions within this beginning practice block were not
governed by a single motor program.
In contrast the pattern of correlations in Blocks 2 and 3 are consistent with a one-unit
structure for both the control and stroke groups; the correlations progressively and gradually
decrease from movement onset and increase from movement offset with all correlation intervals
less than or equal to .20.
In order to examine the overall motor programming proficiency, the average correlation
for each group of comparisons for both the stroke and control groups was calculated. The
average correlation for each block by group is indicated in Figure 12. Correlations did not differ
for Block 1 (control, r=.67; stroke, r=.58; p= 11) and Block 3 (control, r=.78; stroke, r=75; p=53);
however, group differences were present in Block 2 (control, r=.77; stroke, r=.66; p=.05). By the
end of practice, both groups were executing movements as a one-unit action with a comparable
level of programming proficiency.
In summary, the unit analysis reveals a systematic reorganization of motor control. In the
beginning of practice, movements are discontinuous and there is no indication that the action is
governed by a single motor program. However, with practice, the number of DC movements
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
101
decrease and the temporal landmarks are systematically correlated suggesting that the action is
governed by a single motor program. Differences between groups exist in the process of motor
program acquisition in that stroke subjects were not as proficient in motor programming within the
first 100-trials of practice. However, according to the pattern and magnitude of correlations
revealed by the unit-analysis, after 200 practice trials both groups were proficient in the execution
of a one-unit action.
Effects of lesion location
There were several interesting findings concerning the question of performance
differences due to the side of hemispheric damage. First, there were no two-way (Group by Arm-
used) interactions in overall accuracy (RMSEs) when collapsed across blocks for acquisition
(F(1,36)=1.36, p=.20), no-FB retention (F(1,36)2.41, p=. 13), or FB retention (F(1,36)=1.36, p= 25).
In addition, there were no 3-way interactions (Group x Arm-used x Block) for any of the
aforementioned phases. These findings suggest that motor learning differences were not affected
differentially by right-or left-hemisphere brain-damage. Despite the preserved capability for motor
learning, the persistent group effects reported previously suggest differences in motor control.
Therefore, to better understand the nature of motor control differences between those with and
without sensorimotor cortical system damage all of the subsequent analyses are focused on the
nature of group differences throughout the acquisition phase.
There were no Group by Arm-used interactions for MT (F(1,36)=3.09, p=.09) or average
velocity (F(1,36)=.31, p=.58). However, Group by Arm-used interactions were statistically
significant for motor program accuracy (rRMSE), the percentage of DC movements, and the
motor program acquisition process.
Motor program accuracy
Figure 13 illustrates the mean rRMSE for each group during the beginning and end of
acquisition. Motor program accuracy was affected by the side of hemispheric damage evident by
a Group (control, stroke) by Arm-used (right left) interaction for rRMSE during the beginning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
102
Right control
Right stroke
M B Left control
Left stroke
1 2
Acquisition
100 Trial blocks
Figure 13. Bar graph of the mean residual RMSE (+SE) by group for acquisition blocks 1) Trials
1-100 and 2) Trials 101-200. For both blocks, the group by hand interaction was significant
(p< 05). Post-hoc comparisons indicated that the locus of the effect was a statistically significant
difference between the left control and stroke groups (*, p<01) but not between the right control
and stroke groups (ns).
Trials 1-100 (F(1,36)=4.31, p=.05), and end, Trials 101-200 (F(1,36)=9.42, p<01) of practice.
Post-hoc analysis revealed that the locus of the effect was between the performance of the left-
stroke subjects compared to the left controls. In both blocks, the left stroke group had greater
rRMSE than left controls (Beginning, f(1,18)—3.02, p<.01; End, f(1,18)=-3.30, p<01). There was
not a statistically significant difference between right- stroke and control groups (Beginning,
f(1,18)= -.49, p=.63; End, f(1,18)=-1.42, p=.17). These findings suggest a role for the left-
hemisphere in motor program accuracy.
Figure 14 illustrates the relationship between motor program accuracy and lesion location
(SC, Cx±SC, pontine, control). Motor program accuracy was affected by lesion location (Lesion
main effect F(3,33)=4.77, p<01). Subject’s with lesions in the SC or Cx±SC areas had higher
rRMSE than controls or the subject with a pontine infarct (rRMSE: SC, 11.6 deg; Cx±SC, 11.0
deg; pontine, 5.8 deg; control, 8.9 deg). Post-hoc analysis revealed the locus of this effect to be
between Cx±SC and SC groups compared to controls (rRMSE:Cx±SC /controls, f(1,31)=2.82.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subcort i cal 1 0 3
M W CorticaH-/-SC
Control
n-6 n-13 n-1 n-20
Figure 14. Bar graph of the mean residual RMSE (+SE) during acquisition by lesion location
compared to that of controls. Group effect, p< 01. The subcortical (SC) and cortical +/- SC
groups were not statistically different from each other (p=.64) but both were significantly greater
than controls (subcortical, p=.02; cortical+/-SC, p< 01).
p<01;SC /controls, f(1,24)=2.54, p=. 02). The difference between Cx±SC and SC groups was not
statistically different (f(1,17)=0.48, p=.66). Together these findings suggest that
differences in motor program accuracy are related to damage within the sensorimotor cortical
system that involves both cortical and subcorticai cell assemblies. This observation is further
supported (though with caution) by the performance of the one subject with a pontine infarct
Involvement of motor pathways at the level of the pons did not interfere with the central processes
associated with motor program accuracy.
Finally, an analysis that included lesion location and arm-used was completed to
determine if the side of hemispheric damage interacted with lesion location. Table 6 summarizes
the mean rRMSE for the neuroanatomic lesion groupings by arm-used. The Lesion location by
arm-used interaction was significant (F(2,33)=3.21, p=.05); however, post-hoc analysis revealed
no significant locus for this effect Therefore, it appears that the sensorimotor cortical system, in
general, and the left-hemisphere, in particular, have functionally significant roles in motor program
accuracy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 6. Group by arm-used mean (+S£) residual RMSE (deg).
104
Group Right
Arm-used
Left
Subcortical (SC) 9.5 (.38) 13.7 (.34)
(n=3) (n=3)
Cortical+/-SC 11.6 (.36) 10.4 (.23)
(n=6) (n=7)
Pontine 5.8
(n=1)
Control 9.2 (.20) 8.4 (.15)
(n=10) (n=10)
Group main effect, p< 01; Hand main effect p=84; Group by Hand interaction, p=.05.
Post-hoc comparisons reveal no locus for the Group by Hand interaction.
Discontinuous movements
The total number of DC movements was affected by the side of hemispheric damage
evident by a Group (control, stroke) by Arm-used (right,left) interaction that approached
significance (F(1,36)=3.87, p=.06). Illustrated in Figure 15, subjects with right stroke made
substantially more DC movements throughout acquisition than all other groups (right stroke, 593;
right control, 63; left stroke, 295; left control, 133). Post-hoc analysis revealed that the locus of
the effect was between the performance of the right-stroke subjects compared to the right
controls. The right stroke group had a greater number of DC movements than right controls
(f(1,18)=-4.11, p < 01). There was not a statistically significant difference between left- stroke and
control groups (f(1,18)=-1.20, p=.25). These findings suggest that right-hemisphere damage
interferes with the ability to perform continuous movements; possibly related to this hemispheres
role in visual-spatial processing (Winstein & Pohl, 1995; Haaland & Harrington, 1989).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
2 600
£ 350
100
RigM control
Right stroke
Left control
Left stroke
Figure 15. Frequency histogram of the total number of discontinuous movements for each group
by arm-used. Group effect, p<.001; Group x Hand interaction, p=.0568. Post-hoc comparisons
indicate that the locus of the effect was a statistically significant difference between the right
control and stroke groups (*, p<001) but not between the left control and stroke groups (ns).
Pontine
Figure 16. Bar graph of the mean number (+SE) of discontinuous movements during acquisition
by lesion location compared to that of controls. Group effect, p=.01. The subcortical (SC) and
cortical+/-SC groups were not statistically different from each other (p=.81) but both were
significantly greater than controls (p<.01). The difference between the subject with pontine infarct
and controls was not statistically significant (p=.67).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7. Group by Arm-used mean (+SE) number of discontinuous movements.
106
Group Right
Arm-used
Left
P
Subcortical (SC) 39 (15) 46 (23) .81
(n=3) (n=3)
Cortical+/-SC (Cx±SC) 76 (17) 22(8) .01*
(n=6) (n=7)
Pontine 20
(n=1)
Control 6(2) 13(10)
(n=10) (n=10)
Group main effect p< 01; Hand main effect p= 21; Group by hand interaction, p=.01.
Post-hoc comparisons of the subcortical groups was not significant (ns) but was significant
for the Cx±SC (p=. 01).
The mean number of DC movements across acquisition was affected by lesion location
(SC, Cx±SC, pontine, control). This resulted in a main effect for lesion group (F(3,33)=5.40,
p< 01). Figure 16 displays the mean number of DC movements during acquisition for the different
lesion location groupings. Post-hoc analysis revealed the locus of this effect to be between
Cx±SC and SC groups compared to controls (DC movements: Cx±SC/ controls, f(1,31 )=2.96,
p< 01; SC/controls, f(1,17)=-0.24, p<01). The difference between Cx±SC and SC groups was
not statistically different (f(1,17)=0.48, p=.81). Interestingly, the interpretation of these findings is
influenced by a significant lesion group by arm-used interaction effect (F(2,33)=5.05, p=.01).
Table 7 summarizes the mean number of DC movements during acquisition by lesion group and
arm-used. Post-hoc analyses revealed that right Cx±SC lesions resulted in a significantly greater
number of DC movements than all other groups. It appears that the number of DC movements is
particularly related to damage in the right-hemisphere that includes the cortex since there was a
statistically significant difference between right- and left-Cx±SC groups (f(1,11 )=2.98, p=.01) but
not between right- and left-SC groups (f(1,4)=-0.25, p<81). In summary, it appears that
sensorimotor cortical system damage results in a higher tendency to make DC movements.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107
However, it appears that lesions within the cell assemblies that include the right cortex may
contribute differentially to this central processing impairment
Motor program acquisition process
Differences in motor program acquisition are evident in the unit analysis comparisons for
the right- stroke and control groups (Fig 17) and the left- stroke and control groups (Fig 18).
During the beginning of practice (Trials 1-10), both left- stroke and controls groups and the right
control group execute movements comprised of more than a one-unit structure (interval breaks
are indicated by the asterisk). However, by the middle (Trials 91-100) and end (Trials 191-200) of
practice, movements appear to be executed as a one-unit action for both left-stroke and control
groups and the right control group, in contrast, the right-stroke group demonstrates greater than
one-unit actions during both the beginning and middle of practice. One-unit actions are not
evident in the right-stroke group until the end of practice.
A Group by Arm-used analysis of motor program proficiency (i.e., the average correlation
for each group of comparisons indicated in Fig 17 and Fig 18) was calculated separately for the
beginning, middle, and end of practice. In the first 10 trials, the right-stroke group was less
proficient in motor programming compared to all other groups (correlation coefficient right stroke,
.47; right control, .67; left stroke, .69; left control. .67). This difference resulted in a Group by Arm-
used interaction that approached significance (F(1,36)=3.64, p=.06). However, by the middle of
practice this apparent hemispheric effect had dissipated (Group x Arm-used interaction:
(F{ 1,36)=1.42, p=. 24) but group differences were evident (Group main effect F(1,36)=4.08,
p=.05). By the end of practice, both groups executed one-unit actions, presumed to be governed
by a single motor program, with comparable proficiency (Group main effect F(1,36)=.40, p=.53;
Group x Arm-used interaction: F(1,36)=.63, p=.43).
In summary, the unit-analysis reveals that the process of motor program acquisition is
affected by sensorimotor cortical system damage that is also influenced by the side of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
108
Trials 1-10
Right control
0.80
S 0.00
b « ~ c « -d a - f g - b g -c g - d g - « g -f
1.00
0.80
0.80
0.40
0.20
0.00
Right stroke
r - .47
a - b a - c a - d « •• a - f g - b g -c g - d g * « g -f
T rials 91*100
r - .81
0.20
0.00
r - .64
a - b a - c a - d a - « a -f g - b g -c g - d g - a g -f
; |
i
-
1 I
. n
j §
p i
a - b a - c a - d a - a a -f g - b g -c g - d g - a g -f
T rials 1 9 1 -2 0 0
r - .85
a - b a - c a -d « - • a -f g - b g -c g - d g - a g - r
r - .79
2 0.80
a - b a - c a - d a - a a - f g - b g -e g - d g - a g -f
Figure 17. Averaged within-subject correlations (+SE) between the start of movement (a) and
kinematic landmarks (b through /) and the end of movement (g) and kinematic landmarks (b
through f) for the Right control (left) and Right stroke (right) groups during the beginning (Trials 1-
10). middle (Trials 91-100), and end (Trials 191-200) of practice. The asterisk (*) indicates
intervals between correlation pairs greater than .20. The average correlation for each group of
comparisons is indicated.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
Left control Left stroke
Trials 1-10
Trials 91-100
r - .72
K-b a-c *-d a-« a-f g-b g -c g-d g-e g-f
1.00
0.80 &
r - .67
2 0.60
o 0.40
0.20
0.00
a-b a-c a-d a-a a-f g-b g-c g-d g-a g-f
Trials 191-200
r - .70
1 o o
0.80
£ 0.60
e 0.40
0.20
0.00
&
a-b a-c a-d a-a a-f g-b g-c g-d g-a g -f
.71
0.80
£ 0.60
o 0.40
0.20
0.00
m
a-b a-c a-d a-a a-f g-b g-c g-d g-« g-f
Figure 18. Averaged within-subject correlations (+SE) between the start of movement (a) and
kinematic landmarks (b through /) and the end of movement (g) and kinematic landmarks (b
through f) for the Left control (left) and Left stroke (right) groups during the beginning (Trials 1-10),
middle (Trials 91-100), and end (Trials 191-200) of practice. The asterisk (*) indicates intervals
between correlation pairs greater than .20. The average correlation for each group of
comparisons is indicated.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
hemispheric damage. Sensorimotor cortical system damage results in more discontinuities in
movement execution during early practice and less motor programming proficiency throughout the
first 100 trials of acquisition. Differences in motor program acquisition were evident with right-
hemisphere damage particularly in the beginning to middle practice phases. However, with
extended practice, the capability to acquire and execute actions governed by a single motor
program was achieved for those with sensorimotor cortical system damage regardless of
hemisphere.
Discussion
The results of this study suggest that the sensorimotor cortical system has a major role in
the acquisition and control of motor skills but not the long-term learning of those skills. These
findings support that a distributed neural network subserves the control and learning of
perceptual-motor skills. The sensorimotor cortical system has a primary role in pragmatic system
function. The pragmatic system functions to extract parameters relevant to action. In this study,
sensorimotor cortical system damage resulted in movements that were less accurate throughout
practice and retention phases. These deficits were in the ipsilateral limb and provide evidence
that bilateral sensorimotor cortical system processing is required for the control of rapid,
programmed upper limb movements with highly specified temporal and spatial constraints.
In contrast, sensorimotor cortical system damage did not result in long-term motor
learning deficits associated with semantic system function. Throughout acquisition and retention
phases, the pattern of change was remarkably similar between those with and without
sensorimotor cortical system damage. This finding provides evidence that the neural substrate
associated with long-term motor skill memory or “ slow-learning” (Kami et al., 1995) does not
include the sensorimotor cortical system. However, the sensorimotor cortical system does have a
role during the acquisition phase or “ fast-learning” (Kami et al., 1995) associated with the
establishment of motor routines. Through the use of kinematic analysis, differences in the
process of motor skill acquisition related to motor program construction were evident
Sensorimotor cortical system damage resulted in more discontinuous movements and less
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 1 1
programming proficiency during the first 100 trials of practice. However, like the control group, the
group with sensorimotor cortical system damage was able to acquire and execute movements
governed by a single motor program with extended practice. This finding supports the recursive
nature of motor program construction (Jeannerod et al., 1995) within the sensorimotor cortical
system and suggests that additional iterations (i.e.. practice trials) are required to develop a
proficient cell assembly within the cortical system capable of generating programmed actions.
Fundamental control deficits
Despite the performance improvements with practice that were sustained during
retention, sensorimotor cortical system damage resulted in persistent motor control deficits.
Sensorimotor cortical system damage resulted in movements that were less accurate due to
inefficiencies in both motor programming and temporal and spatial scaling ability. The deficits in
temporal and spatial scaling will be addressed in Chapter 6. If sensorimotor cortical system
processing primarily controls parameter scaling and not motor programming, then movements
that have been normalized to reduce temporal and amplitude error should result in comparable
motor program accuracy between groups with and without sensorimotor cortical system damage.
However, in this study motor program accuracy was impaired after sensorimotor cortical system
damage due to processing deficiencies attributed to the left-hemisphere. Throughout acquisition
and retention phases, the group with sensorimotor cortical system damage executed movements
with greater programming error. However, the locus of this effect was due to left-hemisphere
sensorimotor cortical system damage. Previous work has suggested that the left-hemisphere has
a specialized role in the timing of goal-directed movements particularly in the open-loop.
feedforward phase of the action (Fisk & Goodale, 1988; Goodale, 1988; Haaland, Harrington &
Yeo, 1987; Winstein & Pohl, 1995). This study, which specifically investigated the acquisition of
programmed actions, supports the differential role of the sensorimotor cortical system, particularly
the left-hemisphere, in motor programming. The lack of a hemispheric effect across acquisition
and retention phases suggests that hemispheric differences do not impact motor learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
112
This study also provides support for the validity of the residual-RMSE method as a means
for extracting movement control error that discriminates between programming and scaling ability.
Previous work that used RMSE, a global measure of both programming and scaling error, did not
reveal hemispheric asymmetries in motor control in a group with unilateral brain-damage due to
stroke compared to controls (Winstein et al., 1998). However, the influence of hemispheric
specialization on motor control and learning is evident in this study supporting the need for
methods that can discriminate between functionally distinct motor control attributes that may be
differentially affected by neural substrates throughout a neural network.
Deficient skill acquisition processes
The cortex contributes to skill acquisition as supported by neurophysiologic studies with
primates and positron emission tomography (PET) studies with humans. The initiation and
execution of a sensory cued movement includes posterior parietal, prefrontal, supplementary
motor, premotor, and primary motor areas which become active at different times relative to
movement onset (Johnson, 1992; Kalaska & Crammond, 1992). The interaction of these different
cortical areas contributes to the construction and execution of a control program (Grafton et al.,
1992). The cortical activation patterns associated with sensory cued programmed movements
actually changes over the course of skill acquisition. During earty training, there is diffuse cortical
activation which is hierarchically and serially ordered (Kuboto & Komatsu, 1985; Sasaki & Gemba.
1981,1982). As skilled performance develops in later training, cortical activation becomes more
localized (Aizawa et al., 1991; Haier et al., 1992; Jenkins et al., 1994; Remy et al., 1994: Kuboto &
Komatsu, 1985; Mushiake et al., 1990; Sasaki & Gemba, 1981, 1982; Seitz etal.. 1990). These
findings suggest that the neural system involved in motor skill acquisition is a distributed system
that involves the cortex (KaJaska & Crammond, 1992) such that changes in information
processing occur throughout the course of motor learning (i.e., acquiring programmed
movements). In addition, these observations support the rule of synaptic modification proposed
by Hebb (1949), whereby, connections between neurons are strengthened as a result of
experience.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
113
This study supports the importance of subcortical structures within the neural system that
subserves the skillful execution of programmed movements. Lesions that involved damage to
both cortical and subcortical structures resulted in less accurate programmed movements.
However, motor control deficits were similar when lesions included subcortical structures only.
Together these findings suggest that the accuracy of programmed movements is subserved by a
neural network that includes both cortical and subcortical structures.
The neuroanatomic approach to skill acquisition provides important insights about the
brain areas involved in movement control. Since cortical activation patterns change throughout
the course of skill acquisition, it is apparent that central processing is changing. It has been well
established that performance improves with practice (Woodworth, 1899; Snoddy, 1926;
Crossman, 1959; Schmidt 1988). The behavioral manifestation of practice-related improvements
in performance can be quantified and qualified through kinematic analysis of movement
trajectories. Motor learning of programmed movements progresses from closed-loop control to
open-loop control as indicated by movements which are discontinuous earty in learning but
become continuous with experience (Brooks, 1979; Pew, 1966). Kinematic approaches that
investigate skill acquisition of rapid movements have demonstrated that in early acquisition
movement trajectories are variable with discontinuous velocity and acceleration profiles which
become less variable and more continuous with practice (Brooks & Watts, 1988; Darting & Cooke,
1987; Moore & Marteniuk, 1986).
The present study provides evidence at both a neuroanatomic and behavioral level that
motor programs evolve over the course of practice. Groups with and without sensorimotor cortical
system damage demonstrated improvements in performance over the course of skill acquisition.
In early practice, movements were discontinuous and were not governed by a single motor
program as evident by discontinuous velocity and acceleration profiles which demonstrated that
movements were controlled by multiple-action units. However, by the end of practice both groups
executed movements as a continuous, one-unit action. Differences in the skill acquisition process
existed between groups. Subjects with sensorimotor cortical system damage required additional
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
114
trials to achieve a one-unit action with motor program proficiency comparable to controls. In
addition, differences in motor program acquisition were affected by right-hemisphere damage.
Damage to the right-hemisphere of the sensorimotor cortical system resulted in a greater number
of DC movements particularly from the beginning to middle of practice. Winstein and Pohl (1995)
found that right stroke results in more frequent corrective submovements which may reflect
deficits in the ability to control on-line feedback based adjustments during aiming movements. In
the present study, right sensorimotor cortical system damage resulted in a substantially greater
number of corrective submovements throughout the acquisition phase that parallels the findings of
Winstein and Pohl (1995).
However, despite differences in motor program acquisition that were attributed to
sensorimotor cortical system damage and influenced by hemispheric asymmetries, all subjects
were able to generate programmed actions with extended practice. This finding supports the
hypothesis that motor programs are constructed within the sensorimotor cortical system through a
recursive process. The capability of acquiring and executing programmed actions is retained after
sensorimotor cortical system damage; however, additional iterations (i.e., practice) are required to
achieve behavioral proficiency (i.e., continuous movements with accurate relative time between
kinematic temporal landmarks). The behavioral differences between groups can be attributed to
the efficiency of cell-assemblies within the sensorimotor cortical system.
Differences between groups in the skill acquisition process provides support for the
observations of Kami and colleagues (1995) that two different central processing modes support
“ fast-learning" versus “ slow-learning". It appears that the sensorimotor cortical system has a
major role in the central processes associated with “ fast-leaming”. In this study, there were no
differences between groups in motor learning which may be correlated with the “ slow-learning"
processes attributed to the establishment of motor skill memories described by Kami et al.(1995).
In contrast differences existed in the process by which motor programs were acquired and is
consistent with the “ fast-leaming" processes involved in the establishment of motor routines.
Limitations
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
115
For the purpose of this study, lesions were considered within the sensorimotor cortical
system if brain damage included the sensorimotor cortex and/or its afferent and efferent
subcortical projections. The subcortical areas could include the immediate subcortical white
matter such as the corona radiata, the posterior limb of the internal capsule, or the striatum. This
somewhat broad classification was necessary due to the heterogeneity endemic within stroke
populations. Subjects were admitted into the study if the stroke was in the distribution of the
middle cerebral artery. No attempt was made to distinguish between infarcts within the main
(anterior, posterior) or lenticulostriate branches. As a result, 16 of the 20 stroke subjects had
infarcts that included some portion of the striatum, unilaterally. The striatum has a significant role
in both the control and learning of motor skills (Cote & Crutcher, 1991; Doyon et al., 1997; Doyon
et al., in press). However, research that investigates the effects of Parkinson's Disease reveals
that motor control and learning deficits are functionally significant and behaviorally discemable
when extensive and bilateral degeneration is present (Doyon et al., 1997; Doyon eta!., in press;
Hoehn & Yahr, 1967). Further research is needed that investigates the effects of unilateral
infarcts localized to the cortex and striatum in order to delineate the specific contributions of these
areas in the control and learning of programmed movements.
Summary
In summary, this study provides evidence that the acquisition and retention of perceptual-
motor skills involves a distributed neural network. The sensorimotor cortical system has a major
role in establishing motor routines through an iterative process. Practice-related improvements in
motor performance that include the acquisition of fast and accurate programmed movements are
subserved by this neuroanatomic network. The retention of these motor skills does not include
the sensorimotor cortical system directly but involves subcortical areas such as the basal ganglia
and cerebellum with which the sensorimotor cortical system is richly connected. Hence, the
sensorimotor cortical system has a major role in the acquisition and control of rapid limb
movements associated with short-term learning but may have less of a role in the longer-term
learning of these motor skills.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 6
116
DEFICITS IN MOVEMENT PARAMETER CONTROL: SPATIAL AND TEMPORAL COUPLING
AFTER UNILATERAL SENSORIMOTOR CORTICAL SYSTEM DAMAGE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
117
Abstract
The sensorimotor cortical system has a direct role in the planning and execution of fast
and accurate movements. Neuroimaging studies have revealed that learning-related changes
occur in this network over the course of motor skill acquisition. These dynamic changes suggest
that control parameters become better represented within the sensorimotor network with practice.
The purpose of this study was to investigate the contribution of the sensorimotor cortical system in
the parameter control and learning of a rapid, upper limb movement with specific temporal and
spatial goals. A series of 3 experiments was used to examine changes in temporal and amplitude
parameters across acquisition and retention phases in groups with and without sensorimotor
cortical system damage due to stroke. The subjects with stroke performed with the ipsilesional
upper limb. Kinematic analyses were used to quantify the proportion of temporal and amplitude
error. In addition, a component analysis was completed that compared the differences between
groups in temporal and amplitude scaling strategies used across practice.
All participants benefitted from practice and sustained practice-related improvements
during retention. Despite these improvements, subjects with unilateral sensorimotor cortical
system damage performed with greater error in all phases such that movements were slower, and
hypermetric compared to controls. In addition, damage within this network resulted in a
synchrony between temporal and spatial parameters such that slower movements were
associated with larger amplitude movements, faster movements were associated with smaller
amplitude movements. This coupling was not observed in healthy subjects of different ages, and
was not remediated in subjects with stroke who participated in extended practice. These findings
suggest that the sensorimotor cortical system has a major role in the parameter control of rapid
single-joint movements. The control of unimanual rapid movements with highly specified temporal
and spatial goals is modulated by processing within the sensorimotor cortical system of both
hemispheres. Sensorimotor cortical system damage results in the coupling of temporal and
spatial control parameters consistent with a pulse-width control strategy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118
Introduction
The functional anatomy for movement control involves a distributed neural system that
includes several cortical areas (Colebatch etal., 1991; Donoghue & Sanes, 1994; Grafton,
Mazziota, Woods etal., 1992; Johnson, 1992; Kalaska & Crammond, 1992). The primary
sensorimotor cortex, supplementary motor area, premotor area, and posterior parietal cortex have
been shown in both single-cell recording studies with primates (Aizawa etal., 1991; Brinkman &
Porter, 1979; Halsband & Passingham, 1985; M itzetal., 1991; Kobata & Komatsu, 1985;
Mushiake etal., 1990; Passingham, 1988; Sasaki & Gemba 1981,1982; Shibasakietal., 1993;
Tanji et al., 1988; Weinrich & Wise, 1982; Wise, 1985; Wise & Mauritz, 1985) and imaging studies
with humans (Doyon et al., 1996; Grafton et al., 1996; Grafton, Mazziotta, Presty, et al., 1992:
Haier et al., 1992; Jenkins et al., 1994; Kawashima et al., 1994a; Remy et al., 1994) to have a
direct role in the planning and execution of rapid limb movements. The motor cortical areas of
this sensorimotor cortical system have been implicated in the control of motor behaviors such as
the somatotopically organized execution of individual muscles and different motor plans (Grafton,
Mazziota, Woods et al., 1992; Sanes & Donoghue, 1992; Tanji et al., 1988), the execution of
specific movement parameters such as force, direction, and distance (Fu et al., 1993;
Georgopolous etal., 1993; Jeannerod, 1988; Johnson, 1992), and the preparation and execution
of complex movements through the bilateral activation of Ml and primary sensory cortices
(Grafton, Mazziota, Presty, et al., 1992; Rao et al., 1993; Remy et al., 1994; Sadato et al., 1996;
Shibasaki et al., 1993; Tanji et al., 1988).
Fu et al. (1993,1995) have demonstrated that the primary motor and premotor cortices
have a role in the modulation of neuronal pools that control movement parameters such as
direction and distance. Single-cell recording techniques in 2 rhesus monkeys during an
unconstrained arm reaching task were used. The task involved moving a 2-joint manipulandum to
48 targets in 8 different directions and 6 distances. Their results demonstrated that direction and
distance were encoded independently (Fu et al., 1993) and were serially ordered such that the
direction-related discharge occurred first followed by activation related to target position, then
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
119
movement distance (Fu et al., 1995). According to these authors, the encoding of multiple
parameters occurs within primary and secondary motor cortical areas. These studies only
examined contralateral motor cortex fields during movement parameter control. However,
evidence for bilateral motor cortical contributions in the execution of complex movements has
been demonstrated (Grafton, Mazziota, Presty, et al., 1992; Rao et al., 1993; Remy et al., 1994;
Sadato et af„ 1996; Shibasaki et al., 1993; Tanji et al., 1988 ). The tasks in these studies were
either finger sequencing tasks or pursuit-tracking tasks. While it is evident that parameter
specification occurs in contralateral motor cortical areas (Fu et al., 1993, 1995), it is not clear if
parameter specification requires bilateral cortical processing especially for motor tasks that may
be considered complex because of highly specified temporal and spatial demands.
One behavioral hypothesis is that with practice the learner acquires a generalized motor
program (GMP) and the capability to scale it in temporal and spatial domains to meet task
constraints or environmental demands (Schmidt, 1975, 1985). The GMP is an abstract
representation of the movement that defines the relative time or temporal structure of an action.
Parameters are the temporal and spatial features by which the GMP is executed (i.e., scaled).
Therefore, skilled performance of a rapid movement with specific accuracy requirements involves
acquiring the specified motor program and executing it accurately to meet temporal and spatial
constraints. It has been demonstrated in several behavioral studies that programming and
parameterization are separable constructs (Sekiya et al., 1994; Wulf & Schmidt, 1994; Wulf et al.,
1994; Wulf et al., 1993) and suggests that different central processes could mediate the control of
programming separately from that of parameter scaling.
Wulf etal. (1993) based on previous work by Winstein (1988), developed the residual-
RMSE method to differentiate between movement errors as a result of programming from those
attributed to temporal or amplitude scaling. The analysis involves scaling the subject’s movement
trajectory in time and amplitude to maximize the agreement between the subject’s actual
movement and the task goal. This analysis results in a timing factor (i.e., proportion of temporal
error) and amplitude factor (i.e., proportion of spatial error). These variables can be used to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
quantify the practice-refated improvements in parameter scaling across acquisition and learning-
related changes that occur during retention. The residual-RMSE method may be an effective
means to quantify changes in central processes related to temporal and spatial parameter control.
Brain-behavior relationships involved in motor skill acquisition can be investigated by
studying motor learning in brain-damaged populations known to have a lesion within the
sensorimotor cortical system. Previous work (Winstein et al., 1998) suggests that unilateral brain-
damage resulted in impairments in the control but not the learning of a rapid movement performed
by the limb ipsilateral to the brain damage. This finding suggests that bilateral hemispheric
processing contributes to the accurate control of rapid movements. However, this study did not
include lesion data in order to confirm damage within the sensorimotor cortical system or
kinematic analyses to examine changes in motor control during the skill acquisition process.
Therefore, no conclusions could be derived about the neural correlates that contribute to motor
control and motor learning processes. Another issue not addressed in this study was whether the
persistent motor control deficits were resistant to further practice.
It is now understood that the sensorimotor system is highly adaptable as evidenced by
experience-dependent changes in cortical representations. This capability for reorganization has
been demonstrated through motor recovery after motor cortical lesions (Aizawa et al., 1991,
Chollet etal., 1991; Nudo, Wise, SiFuentes, & Milliken, 1996; Pavlides et al., 1993) and in
practice-related changes in motor cortex observed over the course of skill acquisition (Kami et al.,
1995; Nudo, Milliken etal., 1996). Recent imaging studies using PET and fMRI have revealed
that during motor skill acquisition learning-related changes occur within a cortico-striatal-cerebellar
network (Doyon et al., 1996; Grafton et al., 1995) and within the motor cortex itself (Grafton et al.,
1995; Kami et al., 1995). Dynamic changes in the sensorimotor network associated with practice
would suggest that control parameters become better represented in the network with repetition.
However, triai-by-tria! changes in control parameters cannot be assessed using imaging
technology.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
121
The purpose of this study was twofold: 1) to investigate the contribution of the
sensorimotor cortical system in parameter control and learning of a rapid, upper limb movement
with specific temporal and spatial goals, and 2) to determine the effects of extended practice on
parameter control processes in individual's with unilateral sensorimotor cortical system damage.
The sensorimotor cortical system appears to have a major role in the execution and parameter
specification of motor skills but not the long-term learning of those motor skills. Therefore, it is
hypothesized that the sensorimotor cortical system is involved in central processes associated
with parameter control but not the retention of these scaling attributes.
If the sensorimotor cortical system is not involved in central processes associated with the
retention of parameter specification, then there should be no difference in the pattern of change
for temporal and spatial scaling factors between groups with and without sensorimotor cortical
system damage. If the sensorimotor cortical system is involved in central processes associated
with parameter specification control, then there should be differences in temporal and spatial
accuracy between groups for both acquisition and retention phases. Furthermore, differences in
parameter control may be related to lesion site. If movement parameter specification is a result of
neuronal modulation within cortical cell-assemblies, then sensorimotor cortex damage should
result in greater parameter control deficits than that due to subcortical damage.
In addition, it is hypothesized that differences in parameter control may be influenced by
hemispheric specialization (Fisk & Goodale, 1988; Goodale, 1988; Haaland et al., 1987; Winstein
& Pohl, 1995). The left-hemisphere is believed to have a specialized role in the timing of
programmed movements; therefore, it is hypothesized that participants with left-hemisphere
damage will have greater temporal error compared to all other groups. The right-hemisphere is
believed to have a specialized role in visual-spatial integration; therefore, it is hypothesized that
participants with right-hemisphere damage will have greater spatial error compared to all other
groups. Finally, it is hypothesized that participants with sensorimotor cortical damage will benefit
from extended practice indicated by improvements in temporal and spatial scaling accuracy;
however, motor control deficits will be evident that reflect persistent neurologic damage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
122
The Chapter is organized into a series of 3 experiments. Experiment 1 is a continuation
of the experiment presented in Chapter 5 (i.e., same subjects and methods). However, the
analyses in this Chapter are focused on different motor control variables that are related to
temporal and amplitude parameterization. This Experiment examines changes in temporal and
amplitude parameters across acquisition and retention phases in groups with and without
sensorimotor cortical system damage. Experiment 2 examines changes in temporal and
amplitude parameters across acquisition and retention phases in neurologically healthy young and
older groups. The purpose of this experiment is to determine if parameter control movement
strategies are related to sensorimotor cortical damage by comparing the performance of two
healthy groups of different ages. If parameter control movement strategies (i.e., correlation
between temporal and amplitude scaling) are related to damage in the sensorimotor cortical
system, then there should be no difference in movement parameter strategies even though
differences in parameter error (i.e., absolute temporal and spatial error) may be evident between
healthy young and older groups. Experiment 3 examines changes in temporal and amplitude
parameters with extended practice in selected participants with sensorimotor cortical system
damage. If parameter control is a reflection of persistent deficits in temporal and spatial scaling,
then there should be no change in parameter control with extended practice.
Experiment 1
Method
Subjects
Twenty individuals with unilateral hemispheric brain-damage due to stroke (10 right
lesions; 10 left lesions) and 20 age-matched neurologically healthy controls participated in this
study. All subjects were right-hand dominant Brain-damage was confirmed by MRI or CAT scan
reports. The clinical and neurologic status for the stroke group and the demographic and
functional status of both groups has been reported previously in Chapter 5.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
123
Instrumentation, task and procedure
The instrumentation, task, and procedures are the same as has been described in detail
in Chapter 5. Briefly, a lightweight aluminum lever affixed to a frictionless vertical axle was
anchored to a table top. Horizontal movement of the lever was sampled at 200 Hz and displayed
on a computer monitor. The motor task was to produce a rapid horizontal plane arm movement
with specific spatial (three elbow flexion-extension reversals) and temporal (1000 ms) goals.
Individuals with stroke performed with the ipsilesional limb. Control subjects performed with the
limb that corresponded to matched-group assignment
Subjects practiced for twenty 10-trial blocks (Acquisition phase, 200 trials). A 30-minute
break was provided mid-way through practice between blocks 10 and 11. Post-response FB was
provided on the computer monitor positioned in front of the subject FB was provided as RMSET
and a graphic representation of the subject’s own response superimposed with the goal
movement pattern. One day later, the subject returned for 2 retention tests; a no-FB retention test
(20 trials) and a FB retention test in which FB was reintroduced (20 trials).
Statistical analyses and dependent measures
Movement accuracy A global performance score, root mean square error in relation to
the subject's trajectory (RMSEs), was used as the dependent measure of overall accuracy.
RMSEs is a measure of accuracy calculated as the average difference between the subject’s
movement pattern and the goal movement pattern over the subject’s total movement time (MT)3 .
Since RMSEs is a global error measure (Schmidt 1988), it reflects both errors in motor
programming and parameter scaling. Therefore, the residual-RMSE method described by Wulf et
al. (1993) was used to separate the motor programming, temporal scaling, and amplitude scaling
3 RMSEg = £ fa t *[).-
-*-'1 ns
X j = subject’s position at time i
T = target position at time i; it is assumed that target values >1000 ms are 0°
ns = number of samples across the subjects movement time
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
124
of this total error. The analyses in this chapter will be limited to the temporal and amplitude
scaling measures that result from this method.
Temporal and amplitude scaling Based upon the residual-RMSE (rRMSE) method of
Wulf et al. (1993), a computer software program (R. Sullivan, 1997) was developed to
proportionately scale (stretch or compress) the subject’s trajectory in time and amplitude in order
to maximize the agreement between the subject's movement and the goal movement pattern.
Specifically, the subject-produced trajectory and target were synchronized when the subject's
movement exceeded 0.75° displacement from baseline. The subject’s overall movement trace
was proportionately scaled from 0.2 to 2.0 by a time factor increment of 0.1 such that the subject’s
trace was interpolated to the number of samples in the target trace. The time scaling factor that
produced the highest correlation between the rescaled subject trace and the target trace was
saved as a time factor (TF). A TF of 1.0 indicates a movement with 100% agreement with the
target goal (i.e., 1000 ms). Therefore, the TF indicated if the subject’s movement was too slow
(values greater than 1.0) or too fast (values less than 1.0). It should be noted that this scaling
factor represents the proportional scaling of the subject's total movement duration; therefore, any
pauses or phase shifts within the subject's movement were preserved.
After scaling the subject’s trajectory in time, a similar procedure was applied for
amplitude. Using the median target and subject amplitude trace values, the subject’s trace was
proportionately expanded or compressed until the highest correlation between the subject’s time-
scaled trajectory and the target pattern was reached. For example, the median value of the
subjects time-scaled trajectory array would be determined. If the proportionate scaling factor for
an iteration was x tnen each point in the subjects median time-scaled trajectory would be
multiplied by x. The median target value was an effective means to preserve the relationship
between movement reversals (i.e., extension for the first peak, flexion for the second peak,
extension for the third peak) since the scaled values from the median were multiplied by a positive
value if they were above the median or a negative value if they were below the median. If the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
125
RMSEs 38.8
a
a
T J
C
■
5
E
c 0
v
E
o
o
0
0 *
5
400 0 800 1200 1600
Time (ms)
Figure 1. Example of movement trajectory scaling. The goal movement is displayed as the thick
black line. The subject's actual movement is displayed as the thin black line (tic marks indicate
peak displacement for reversals). RMSEs is the difference, in degrees, between the goal pattern
and subject’s movement TF and AF are calculated after the subject’s trajectory is scaled in time
(dashed line) and then amplitude (dot-dashed line).
correlation between this iteration and the target was the highest of all attempted iterations, then
the subject’s amplitude factor (AF) would be x. The AF indicates whether the subject’s overall
trajectory was hypermetric (values greater than 1.0) or hypometric (values less than 1.0). Figure
1 illustrates the scaling process that results in RMSEg, TF, and AF for an individual trial.
Data analysis Individual trial data were grouped into 10-trial blocks for the acquisition and
delayed-retention phases. Performance groupings included the acquisition phase (Blocks 1-20),
no-FB retention phase (Blocks 21-22), and FB retention phase (Blocks 23-24). A Group (Control,
Stroke) x Hand (right, left) x Block repeated measures analysis of variance (ANOVA) was
completed for each phase for each of the dependent measures. A Students independent
samples (-test was used for the post-hoc comparisons to determine the locus of any significant
interaction effects. For all statistical tests, significance were set at p < .05, and the Greenhouse-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126
Geisser degrees of freedom adjustment was used to compute the probability level for the
repeated measure factors.
In addition, to assess group differences attributable to the side of hemispheric damage,
special attention was given to any Group x Hand interactions for TF or AF. If a Group x Hand
interaction was present this would suggest that temporal or amplitude scaling ability was related to
hemispheric damage and not group membership or handedness alone.
To determine the nature of practice-related improvements, a post-hoc component
analysis that compared the difference in temporal and amplitude scaling strategies used between
control and stroke subjects was completed. Visual inspection of group and individual trial data
revealed a pattern of change between TF and AF that appeared to be influenced by group
membership; therefore, the component analysis was completed to quantify this observed
relationship. The component analysis determined the degree of coupling between temporal and
amplitude scaling factors by calculating the correlation across the 200-trials of practice for each
subject To normalize the correlation coefficient the correlation for each individual was
transformed to a z' score and group differences were determined using an independent samples
Student f-test
To determine if lesion location influenced parameter control strategies, subjects were
grouped into one of the following neuroanatomic categories: (1) Subcortical (SC) - lesion limited to
sensorimotor subcortical structures only (i.e., capsular, striate, and/or thalamic), (2) CorticafctSC
(Cx±SC) - lesion included sensorimotor cortex with or without subcortical structures, (3) Pontine -
included the one subject with the medial pontine infarct, and (4) No lesion - included the 20 control
subjects. These groupings were used since the cortex is known to have populations of cells
whose neuronal firing controls the direction and amplitude of movements (Georgopolis, 1991).
These neuronal populations may affect the control strategies used for temporal and amplitude
scaling that could be revealed by comparing the performance of those with and without damage
that includes the cortex. A factorial ANOVA with lesion location as the independent factor was
completed to determine differences in temporal and spatial coupling as a function of lesion
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
127
location. Post-hoc comparisons for the effect of lesion location were determined using an
independent samples Student f-test
RESULTS
Motor control deficits
Reported in Chapter 5, the brain-damaged group demonstrated preservation of motor
learning capability but motor control deficits were evident In order to distinguish motor
performance deficits due to parameter scaling, TF and AF were analyzed separately for the
acquisition and retention phases.
Parameter error
The proportion of temporal and amplitude error, averaged into 10-trial block means, for
the acquisition, no-FB retention, and FB retention phases are summarized in Figure 2. Collapsed
across practice, stroke participant's movement response relative to the goal-movement was 41%
slower and 32% hypermetric. Control subject’s movements were 9% slower and 13% hypermetric
compared to the goal movement Throughout all phases and summarized in Table 1, the stroke
group performed with slower movements (Group main effect for TF: acquisition, F(1,36)=13.78,
p<001; no-FB retention, F(1,36)=8.25, p<.01; FB retention, F(1,36)=6.96, p =01) that were
hypermetric (Group main effect for AF: acquisition, F(1,36)=3.60, p=.06; no-FB retention,
F(1,36)=5.05, p=.03; FB retention, F(1,36)=3.95, p=.05) compared to that of the control group. It
should be noted that the effect for AF during acquisition approached statistical significance.
All subjects increased movement speed and improved spatial accuracy across practice
as indicated by a significant Block effect for both TF (F(5,162)=28.33, p<.0001) and AF
(F(6,221)=9.04, p<.0001). In fact, the degree of practice-related improvement was higher in the
stroke group than the control group. Participants with stroke improved movement speed by 48%
(Block 1, M=2360 ms; Block 20, M=1240 ms) and spatial accuracy by 32% (AF: Block 1, M=1.84;
Block 20, M=1.25). In contrast control subjects improved movement speed by 26% (Block 1,
M=1400 ms; Block 20, M=1040 ms) and spatial accuracy by 14% (AF: Block 1, M=1.23; Block 20,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 2 8
2.50 .50 | #
_ .o o /
1.30 / •
* Control
• Stroke
1.80
S. 1.40
®
E
l - 1.20
1.00
0 .8 0
2.00 ,
1.75 /
1.60 X
• •
* . •
• • * - *
• • * •
- / / -
2 4 6 8 10 12 14 16 18 20 21 22 23 24
B
o
O 1.40
a
®
■ o
J 1.20
a
E
<
1.00
0.80
• •
• • • •
• # * • • •
• • • •
* * A A A ^ a
A A A
* * • A A A „
-y / -
2 4 6 8 tO 12 14 16 18 2021 22 23 24
Acquisition N0-f B FB
(io-triai blocks) Retention
Figure 2. Block means for acquisition (Blocks 1-20), no-FB retention (Blocks 21-22), and FB
retention (Blocks 23-24) for A) TF comparison between control and stroke groups, and B) AF
comparison between control and stroke groups. Horizontal line at 1.00 indicates movement goal.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
129
Table 1. Mean (+SE) temporal and amplitude scaling factors by Group for the acquisition
and retention phases.
Group Acquisition no-FB Retention FB Retention
Control
TF 1.09 (.01) 1.06 (.03) 1.04 (.02)
AF 1.13 (.01) 1.13 (.02) 1.09 (.01)
Stroke
TF 1.41 (.03) 1.38 (.08) 1.28 (.06)
AF 1.32 (.02) 1.42 (.09) 1.29 (.07)
TF, temporal scaling factor; AF, amplitude scaling factor.
Value of 1.0 would be in 100% agreement with target temporal and spatial goals.
In all comparisons, Group main effect p s .05 except AF for acquisition phase, p = .06,
moderate effect size, ES=. 51
M=1.06). This difference between groups across practice resulted in significant Group x Block.
interactions for both TF (F(5,162)=8.90, p<0001) and AF (F(6,221)=3.37, p<01). Individual trial
data from a representative control subject and stroke subject during early and late practice are
displayed in Figure 3. Both subjects decreased temporal and spatial error with practice.
Interestingly, for the stroke subject, TF and AF decreased with practice, yet RMSEs remained
high. This finding substantiates that RMSEs is a global error measure that is not sensitive to
practice-related improvements in parameter control.
Post-hoc analysis comparing performance during Blocks 1-10 and Blocks 11-20 revealed
differences in temporal and spatial parameterization between groups in the early compared to the
later practice blocks. During the first 100 trials, the stroke group decreased temporal error by
46% (TF: Block 1, 2.36; Block 10, 1.27) and amplitude error by 33% (AF: Block 1, 1.84; Block 10,
1.24). For the control group, temporal error decreased by 24% (TF: Block 1,1.40; Block 10,1.07)
and amplitude error decreased by 6% (AF: Block 1,1.23; Block 10,1.16). This resulted in a
group by block interaction for Blocks 1-10 for TF (F(3,108)=9.80, p < .0001) and AF
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Early Late
130
R W S & I1 7 .8 I
T F 1 .1 2 !
T F . 9 4 1
Control
(subject JA)
-32 32
500 0 1000 T 5 0 0 0 3 0 0 6 0 0 • 0 0
Stroke
(subject R6)
6 4
32
32
0 5 0 0 1000 1S00 5 0 0 1000 2000 1S00 0
Time (ms) Time (ms)
Figure 3. Individual subject trial data. Rows show data from a control (top) and stroke subject
(bottom) during early (left) and late (right) practice. The goal movement is the thick black line.
RMSEg f°r the subject’s actual movement (thin black line), TF from the time scaled movement
(dash line), and AF from the amplitude scaled movement (dot-dash line) is displayed in the table
for each trial. Note different time scales between trials.
(F(4,140)=5.07, p <.001). In contrast, the pattern of change was not different between groups in
the last 100 trials of practice (Group x Block interaction: p>.05). This analysis suggests two
salient points. First the stroke subjects demonstrate a remarkable capability to improve
movement speed and accuracy with practice. This resulted in a substantial amount of change in
parameter scaling ability in the first 100 trials. In contrast, control subjects did not demonstrate
such a large degree of change; however, temporal and spatial scaling were already approaching
the goal movement parameters even within the first 100 practice trials.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
131
Parameter control strategies
Figure 4 displays the within group changes in temporal and spatial scaling during the
acquisition phase for the control and stroke groups. To examine the difference in movement
A
TF
r « .27
z' = .3 1
1.50
I .
O
o
0 1.17
Q
C
0
o
C O
1.08
1.00
10 12 14 16 18 20 6 8 2 4
B
r= .59
— AF
1.70
1.60
O
—
a
a 1.40
a
c
a
o
(0
1.20
1.00
10 12 14 16 18 20 6 8 2 4
Acquisition
(10-Ufai blocks )
Figure 4. Change in block mean TF and AF scaling for the acquisition phase (Blocks 1-20).
Correlation coefficient (r) and normalized correlation coefficient ( t score) for the control group (A)
and stroke group (B) are indicated.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
132
Loft control (HW)
r - .07
TF .75
AF
1 .5 0
Left stroke (Li)
r - 85
TF
AF
1 -2 5 I
a
o
to ►. •
¥ ----------
2 4 6 a to 1 2 1 4 1 6 IB 2 0
2 4 « 8 1 0 1 2 1 4 1 $ IS 2 0
R ight control (CP)
r - -.25
Right stroko (R4)
2 1 -2 5
• 1.00
o
* 4
t
r •
2 4 S 8 1 0 1 2 1 4 is 1 8 2 0
Acquisition
(10-tru i W odta)
2 4 S • 1 0 1 2 1 4 1 5 ’8 2 0
A cquisition
(1 0 -trtt M eets)
Figure 5. Individual trial TF and AF data across the acquisition phase (Blocks 1-20) for two
representative control (left) and stroke subjects (right). Horizontal line at 1.00 indicates movement
goal.
control strategies between groups, a correlation analysis that analyzed the change in temporal
and spatial scaling for the acquisition phase was calculated. For each subject the correlation
between TF and AF was calculated over the 200 practice trials. Each subject correlation was t
transformed in order to statistically determine group differences. Changes in spatial and temporal
error were tightly coupled for the stroke group (r=.59±.22) and far less coupled for the control
group (r=.27±.29). This difference was statistically significant (stroke, z’=.75±.38; control,
z’=.31±.37; t=-3.73, p<.001). Individual subject data that demonstrate the higher correlation
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
133
between temporal and spatial scaling across practice for stroke compared to control subjects are
displayed in Figure 5.
Lesion effects on parameter control
Of particular importance to the effect of hemispheric differences, there were no Group by
Hand interactions for TF or AF in any experimental phase (Acquisition: TF, F=.98, p=.33, AF,
F=. 19, p=.67; No-FB retention: TF, F= 18, p=.67, AF, F=.06, p=.80; FB retention: TF, F=1.84,
p=. 18, AF, F= 03, p=.87). These analyses provide strong evidence that temporal and amplitude
parameterization are not influenced by the side of hemispheric damage.
The effect of lesion iocation on temporal and spatial coupling was determined by a
factorial ANOVA comparing subjects with SC lesions, Cx±SC involvement the one subject with
the medial pontine infarct, and the neurologically healthy controls. Evident in Figure 6, subjects
with sensorimotor cortical system involvement had the highest correlation between temporal and
spatial coupling (SC, z'=.65; Cx±SC, z -8 4 ; Pontine, z,=.11) compared to controls (z’= 31) that
resulted in a significant effect of lesion location (F=6.49,p<.01). The locus of this effect is such
that the z' scores for the SC and Cx±SC groups were not different from each other (f=-1.48,
p=. 16); however, both groups had significantly greater correlations than the control group (SC,
f=2.13, p=.04; Cx±SC 1=3.94, p<001). In addition, the individual with pontine infarct did not differ
significantly from the control group (f=-.53, p=.60). After applying the Bonferroni correction for the
number of post-hoc comparisons, the probability level for the SC and control group comparison
would not be considered significant However, the magnitude of group differences was
considered large for both the SC and Cx±SC groups compared to controls based upon effect size4
(SC /control: ES=.82; Cx±SC /control: ES=1.40) According to Thomas, Salazar, and Landers
(1991), an ES greater than .80 indicates large group differences. These findings suggest that the
coupling of temporal and spatial scaling is related to damage within the sensorimotor cortical
4 ES = M1 - M2
pooled SD
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
134
o
C O
sc
1.00
r - .69
m Cx+/-SC
0.80
1
tn
0 0
Pontine
— H H
Control
® 0.60
o
n 0.40
0.20
r - .27
n-6 n-13 n-1 n-20
Figure 6. Bar graph of mean t scores (normalized correlation coefficient) and SE for the
temporal and spatial scaling correlations by lesion location compared to control group. The
correlation coefficient for each group is indicated. Group effect p<01. The subcortical (SC) and
cortical ±SC ( Cx±SC) group were not statistically different from each other (p=. 16) but both were
significantly greater than controls (SC, p=.04; Cx±SC, p<001). The difference between the
subject with pontine infarct and controls was not statistically significant (p=.60).
system that is affected by damage within the cortex itself and/or its subcortical projections.
Further, the locus of this effect is supported, albeit cautiously, by the performance of the one
subject with pontine infarct. Damage to motor pathways at the level of the pons did not result in
substantial temporal and spatial coupling suggesting that the central processes associated with
parameter control must occur within the sensorimotor cortical network prior to movement
execution.
Discussion
Established in Chapter 5, the sensorimotor cortical system is involved in the control but
not the retention of a skilled action. During motor skill acquisition, significant group differences
persist throughout practice and delayed-retention phases. This experiment reveals that one
aspect of the motor control deficit is a deficiency in the absolute scaling of temporal and spatial
movement parameters that is related to sensorimotor cortical system damage independent of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
135
hemisphere. Sensorimotor cortical system damage resulted in movements that were slower and
hypermetric. Persistent movement deficits in parameter scaling present in the ipsilesional limb
suggests that accurate temporal and spatial scaling for a rapid upper limb movement with highly
specified temporal and spatial goals requires bilateral activation of sensorimotor areas.
In addition, changes in temporal and spatial scaling were highly coupled after unilateral
sensorimotor cortical system damage and significantly less coupled in healthy controls. This
suggests that the central processes controlling parameterization are altered. Damage within this
network results in synchrony between temporal and spatial parameters such that slower
movements are associated with larger amplitude movements, faster movements are associated
with smaller amplitude movements. This coupling between temporal and spatial control
parameters suggests that different control strategies are used between those with and without
sensorimotor cortical system damage.
An explanation for these findings may be derived from proposed control models that
hypothesize the nervous system control in rapid, single-joint movements. These models relate
the pattern of motor neuron excitation which results in muscle contraction to emergent kinematic
and EMG patterns. Gottlieb, Corcos, and Agarwal (1989a) proposed a “ dual-strategy* hypothesis
that suggested that an organizing principle underlying the control of rapid, single-joint movements
is that the nervous system uses two possible mechanisms to generate task-appropriate forces.
One mechanism is to develop large amounts of force to increase movement speed or amplitude
by increasing the number of motor units recruited and to develop more force within these units.
This mechanism has been referred to as a puise-height (Gordon & Ghez. 1987) or speed-
sensitive (Gottlieb, Corcos, & Agarwal, 1989a; Corcos, Gottlieb, Agarwal, 1989b) control strategy.
Another mechanism is to increase force by increasing the duration of motor unit excitation. This
has been referred to as a pulse-width (Hoffman & Strick, 1993) or speed-insensitive (Gottlieb,
Corcos, & Agarwal, 1989a; Corcos, Gottlieb, Agarwal. 1989c) control strategy.
Gottieib et al. (1989a) used the speed-sensitive (SS) and speed-insensitive (SI) strategies
to relate the excitation pulse (i.e., neural input to the alpha motoneuron pool) and kinematic task
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
136
constraints (i.e., the control of distance and accuracy). In their terminology, a strategy is the set of
rules used to perform a task. According to these authors, an SS strategy is used when the
performer is expected to control the speed of movement such that motoneuron pool excitation is
proportionate to movement speed. The excitation pulse is modulated by increasing intensity;
therefore, change in the rate of rise would be expected with pulse duration held relatively constant
(consistent with pulse-height modulation). Typically, the SS or pulse-height control strategy has
been implicated in the control of movements where distance is varied (i.e., rapid movements to
different target locations). However, another category of tasks where the SS strategy is
appropriate is for movements with fixed distance and load executed in an explicitly defined
duration. Therefore, the goal of the task can only be achieved by acquiring the “ correct" speed or
average velocity. In contrast the SI strategy is used when control over movement speed is not
explicitly defined. In this case, the excitation pulse intensity is held constant and the duration is
modulated (consistent with puise-width modulation). Kinematically, this results in a constant initial
rate of acceleration. With this strategy (assuming constant inertial loads), distance and time are
directly related.
The task in the present study is to acquire a rapid limb movement with explicitly defined
temporal and spatial constraints. This task differs from the tasks cited in the preceding studies in
that the task includes 3 amplitude reversals within a designated MT goal; the other studies involve
various amplitude movements but to only one target However, the acceleration demands for the
present study's task are independent of the amplitude goals (i.e., speed not correlated with
amplitude). Therefore, it appears that the SS (pulse-height) strategy would be the appropriate
strategy to meet task goals. The control subjects in this study performed the task in such a way
as to support a pulse-height control strategy. First, as presented in Chapter 5, the control group
learned the average velocity by the end of acquisition (task goal: 175 deg/sec; control group: 174
deg/sec. Chap 5, Table 5) suggesting that they had acquired the correct speed for the task goal.
Second, temporal and spatial scaling were not highly correlated; spatial accuracy and MT did not
vary proportionately.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
137
In contrast temporal and spatial scaling were highly coupled in the stroke group. The
direct correlation between temporal and spatial scaling suggests that MT was proportionate to
distance. It appears that the stroke group was using a pulse-width control strategy. If the initial
rate of rise in acceleration cannot be changed, there will be a direct relationship between distance
and MT. This was observed in the stroke group kinematics. If a particular trial included a
movement that was too slow, then the correct response on the next trial would be to increase
movement speed. However, the decrease in MT was usually coupled with a movement that was
spatially hypometric. Conversely, the opposite was true; increases in movement amplitude were
correlated with longer MTs. It appears that the stroke group was limited in their ability to vary the
rate of acceleration rise, consistent with a pulse-width control strategy.
Recent work has suggested that there is not a dichotomous distinction between these
control strategies. Rather the nervous system uses pulse-height or pulse-width modulation
depending on the nature of the forces required for task demands (Gottlieb, Chen, Corcos, 1995;
Gottlieb, Chen, Corcos, 1996; Pfann, Hoffman, Gottlieb, Strick, & Corcos, 1998). In a study that
varied distance and inertial loads at the wrist, elbow, and ankle, Pfann et al. (1998) were able to
demonstrate that there are common control rules that result in modulation patterns influenced by
several biomechanical factors such as muscle properties, muscle strength, visco-elastic
properties, and external loads. Hoffman and Strick (1993), by varying movement amplitude and
external load, demonstrated that as force demands increase there is a shift from pulse-height to
pulse-width modulation. Therefore, force output is modulated as a result of task demands not a
shift or choice between movement control strategies.
In the present study, task demands were held constant but differences in movement
strategy were evident between those with and without sensorimotor cortical system damage.
Differences between subjects during constant task conditions was also found by Hoffman and
Strick (1993). In their study, the weakest subject performed the same tasks using a pulse-width
control strategy. These authors propose that in conditions where force requirements are high
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
138
o
o
C O
N
1.S0
1.00
o .s o
0.00
o
r = -.38, p= 02
0
o
0 0
0 0 ft
o o o
° 0 °
▲
▲
o
A Control
o Stroke
O
o
A A O
A
A O
A A
-40 0 40
Grip strength difference from normative values
(lbs)
Figure 7. Scatter plot of grip strength by temporal and spatial coupling correlation (normalized
correlation coefficient) for control (triangle) and stroke (circle) subjects.
(e.g., large externally applied loads) or when central drive to the motoneuron pool is maximized
then force output cannot be increased by further pulse-height modulation but must be augmented
by pulse-width modulation.
Evidence from the present study suggests that unilateral sensorimotor cortical system
damage may result in diminished central drive to motoneuronal pools that restricted stroke
subjects to a pulse-width control strategy. First, the stroke subjects in the present study were
significantly weaker than control subjects as measured by grip strength differences controlled for
age, gender, and hand-used. The relationship between muscle strength and agonist modulation
proposed by Hoffman and Strick (1993) is supported in this study by comparing the correlation of
temporal and spatial scaling and grip strength measures within the stroke and control groups. If
evidence exists for a pulse-width control strategy that is related to muscle strength, then subjects
with muscle weakness (i.e., grip strength below normative values) should have high temporal and
spatial coupling correlations (i.e., suggestive of pulse-width control). Figure 7 plots grip strength
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
139
by the normalized correlation coefficients for temporal and spatial scaling for the stroke and
control participants. Evident in Figure 7, there is a indirect correlation between grip strength and
temporal and spatial coupling (r=-.38, p= 02), higher temporal and spatial correlations are
associated with decreased grip strength, that appears to be influenced by sensorimotor cortical
system damage. Participants with sensorimotor cortical system damage tended to cluster in the
range associated with decreased strength and high coupling in temporal and spatial parameter
control.
Differences in control strategies between groups with and without sensorimotor cortical
system damage performing the same task suggests that control strategies may be influenced by
the amount of excitation to motoneuron pools. Most likely lesions within the sensorimotor cortical
system would limit the number of recruitable motor units such that the amplitude of a burst would
be diminished; therefore, the only mechanism to grade force is to increase pulse duration. If this
is the case, then it appears that a unilateral lesion within the sensorimotor cortical system results
in a fundamental control impairment that is influenced by the loss of neuronal tissue within the
cortex and its subcortical projections.
To further clarify if this deficit is related to sensorimotor cortical system damage and to
determine if the motor control impairment can be influenced by additional practice, two additional
experiments were completed. In Experiment 2 the performance of healthy young and older
subjects was compared. The aging literature consistently reveals examples of group differences
in speed and accuracy between young and older subjects (Pohl & Winstein, 1998b; Swanson &
Lee, 1992; Welford, 1985). Therefore, it was hypothesized that older subjects will perform with
more error in all phases compared to younger subjects (i.e., Group main effect for age); however,
parameter control strategies will be similar between groups. If the coupling of movement
parameter scaling is not different between neurofogically healthy young and older subjects, then it
would suggest that the pattern of force recruitment is similar between these two groups. This
finding would support that the coupling of temporal and amplitude scaling observed in the subjects
with stroke resulted from sensorimotor cortical system damage and not that of aging alone.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
140
Experiment 3 was designed to determine if deficits in motor performance and parameter
control strategies were a persistent reflection of sensorimotor cortical system damage or if motor
performance and/or control strategies could be changed with extended practice. If temporal and
spatial coupling are a reflection of persistent deficits related to sensorimotor cortical system
damage, then there should be no change in the correlation between the TF and AF throughout an
extended practice phase. However, if motor control deficits are remediable than there should be
an improvement in motor performance and/or a decrease in correlation between the TF and AF
across extended practice.
Experiment 2
Methods
Subjects
Ten neurologically healthy subjects voluntarily consented to participate in this study.
Subjects were assigned to group based on age. The young group was comprised of 4 females
and 1 male (age range 23-26 yrs, M=24±1.2). The older group was comprised of 3 females and 2
male (age range 58-80 yrs, M=68±8.3 yrs). The participants in this older group were from the
Experiment 1 control group.
Instrumentation. Procedures. Task, and Analyses
The instrumentation, procedures, task, and analyses were the same as that described for
Experiment 1.
Results
Movement accuracy and parameter control
Figure 8 summarizes the group RMSEs means for the acquisition (Blocks 1-20), no-FB
retention (Blocks 21-22), and FB retention (Blocks 23-24) phases. Both groups improved
movement accuracy across practice (Block effect, F(4,30)=6.27, p< 01), and sustained practice-
related improvements in a similar manner during retention (All phases: Group x Block interactions,
p>.05). Older subjects tended to have higher RMSEs scores than younger subjects throughout all
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
141
25
a Old
• Young
^ 20
□ >
■ o a 1 *
t 1 5
L U
C O
10
A A
A A
A A A
A A #
✓ / --------------------
2 4 6 8 10 12 14 16 18 2021 22 23 24
NO-FB F B
Acquisition
(10-trial blacks)
Retention
Figure 8. Young and older group RMSEs block means for acquisition (Blocks 1-20), no-FB
retention (Blocks 21-22), and FB retention (Blocks 23-34) phases.
phases but these differences did not reach statistical significance (Group effect Acquisition,
F(1,8)=3.37, p=.10; no-FB retention, F(1,8)=2.68, p=14; FB retention, F(1,8)=3.54, p= 10).
However, the non-significant group effects were due to the small sample size. ES greater than
.80 indicates large group differences (Thomas et al.,1991). Large group differences in accuracy
were evident in each phase as revealed by ES (Acquisition, ES=.83; no-FB retention, ES=1.04;
FB retention, ES=1.04).
The temporal and amplitude scaling factors for the young and older groups across
acquisition is displayed in Figure 9. There were no group differences for TF (Group effect p=. 57)
or AF (Group effect p=.55). The correlation between TF and AF across the 200 acquisition trials
was similar for both groups (young, r=.31; old, r=.30). Differences between the group normalized
correlation coefficients was not significant (young, z'=.36; old, Z-.34; f(1,8)=-.09, p=.93). These
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
142
< s
o
CO
1.40 1
1.25 /
r - .31
TF
AF
1.20 /
1.10
C 9
1.00
0.90
8 6 10 12 16 18 20 2 4 14
B
1.40 |
1.25 \ k
r = .30
TF
AF
1.20
o
a i.io
a >
c
1.00
0.90
2 4 6 8 10 12 14 16 18 20
Acquisition
(10-crial blocks)
Figure 9. Change in block mean TF and AF scaling for the acquisition phase (Blocks 1-20).
Correlation for the young group (A) and older group (B) are indicated. Horizontal line at 1.00
indicates movement goal.
findings suggest that temporal and amplitude scaling and movement parameter control strategies
did not differ in healthy subjects of different ages. Both young and older subjects were able to
control temporal and spatial scaling relatively independently.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
143
Discussion
The results of Experiment 2 provide further evidence that the difference in control
strategies identified in Experiment 1 are related to sensorimotor cortical system damage. Older
subjects had a tendency to be less accurate than younger subjects in both the acquisition and
retention phases. This tendency did not reach statistical significance which is most likely due to
the small sample size; however, the effect size was large. Despite the difference in movement
accuracy, both groups made practice-related improvements in motor performance and used
similar motor control strategies for temporal and spatial parameter control. Both groups were able
to control temporal and spatial scaling relatively independently suggesting that there was no
difference in control strategies between groups. It appears that older and young subjects used a
pulse-height control strategy for the rapid limb movement used in this experiment
Experiment 3
Methods
Subjects
Three participants with stroke from Experiment 1 who had a Block 20 RMSEs score less
than the stroke group mean (Block 20 RMSEs: stroke group, 22.8; R4, 25.6 deg; L2, 30.0 deg; L7,
25.1 deg) voluntarily agreed to participate in an extended practice session.
Instrumentation. Procedures. Task, and Analyses
The instrumentation, procedures, task, and analyses were the same as that described for
Experiment 1 with the following exception. Participants in the extended practice study returned
within 2-4 weeks of the initial practice day for 200 additional trials of practice. Similar to
Experiment 1, participants practiced for 200 trials on Day 3 and returned Day 4 for a no-FB and
FB retention test (20 trials each).
Results
Individual Block mean RMSEs scores for the acquisition and retention phases for both
the initial and extended practice sessions (400 total practice trials) are displayed in Fig 10.
Regression lines have been added to illustrate the performance changes across all phases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject L7
144
«< * R U S E
I
C I " , • • •
s 3 0 • • * . •* .
W • # • • •
2 j •
I U i . \
2 ! •
2
( E i
a t
4 8 1 2 I S 2 0 2 4 2 B 3 2 3 1 4 0 4 4 4 8
Subject L 2
2 5 t
!
i
! / /-----------------------
4 I 1 2 IS 2 0 2 4 2 8 3 2 3 6 4 0 4 4 4 8
Subject R4
4 5 ;
/ / ---------------------------
4 8 1 2 I S 2 0 2 4 2 8 3 2 3 6 4 0 4 4 4 8
Acquisition 1 n o - f b j f b Acquisition 2 x o - f & t b
RatantiM mhow
Figure 10. Individual stroke subjects block RMSEs scores for acquistion 1 (Blocks 1-20), no-FB
retention (Blocks 21-22), FB retention (Blocks 23-24), acquisition 2 (Blocks 25-44), no-FB
retention (Blocks 45-46), and FB retention (Blocks 47-48). Regression lines have been added to
illustrate the direction of performance changes across phases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
145
Subject L7
3.00 i
>.» r
2.00 Y
i
< 0
— 1.50 i
o
C O
TF
AF
r - .69 f - 84
I - -
• •
< 0
o
C O
1.00
5 10 15 20 25 30 35 40
Subject L2
- .50 r - 76
2.10
2.00
> .
I.
©
o
.E ! .50
1.00
• •
r
5 10 15 20 25 30 35 40
Subject R4
r - 66 • r - 78
2.80
2 .7 5 f
c o
o
0
o
C O
• •
♦ •
» •
5 10 15 20 25 30 35 40
Acquisition 1 Acquisition 2
( l 0 > t r i a t M o c k s )
Figure 11. Individual stroke subject TF and AF data for acquisition 1 (Blocks 1-20) and acquisition
2 (Blocks 21-40). The correlation coefficient of time and amplitude factors for each acquistion
phase is indicated.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
146
RMSEs scores decreased across Blocks for subjects L7 and R4, and increased for subject L 2.
This gives the impression that performance was deteriorating with extended practice for subject
L2. However, evident in Figure 11, all three subjects improved in temporal and amplitude
parameter scaling; movements became faster (decrease in TF) and more spatially accurate
(decrease in AF) with extended practice.
Individual trial data from the extended practice phase for subject L2 demonstrates that
while overall accuracy decreased (i.e., global RMSEs performance scores increased), the fit
between the subject's trajectory and target pattern when scaled for timing and amplitude error was
improving (i.e., little difference between the temporally and spatially scaled subject trajectory and
the target Fig 12). In fact Block TF and AF means (Fig 11) for subject L2 demonstrate that
scaling capability improved with extended practice. For both TF and AF, accuracy improved and
variability decreased between the initial (Blocks 1-20) and extended (Blocks 21-40) practice
phases (initial: TF, M=1.65±.30; AF, M=1.54±.24; extended: TF, M=1.51±.15; AF, M=1.43±.18).
it is interesting to note that subject R4 also demonstrated increased RMSEs scores during the
beginning of the extended practice phase but made a substantial gain in performance during trials
270-400. Individual trial data for R4 reveals improvement in movement accuracy and parameter
scaling throughout the extended practice trials (Fig 13). Both cases demonstrate the limitations of
the global RMSEs score for measuring changes in skill acquisition across practice. Each of these
three subjects demonstrated improvements in motor performance with more discriminating
measures of parameter control. Improvements in parameter control are evident by all three
subjects (Fig 11); yet an apparent decrement in overall accuracy seems evident in subject L2 as
RMSEg increases throughout the extended practice phase (Fig 10).
To determine if parameter control strategies changed with extended practice,
comparisons of TF and AF correlations for the initial and extended practice phases were
performed. The TF and AF Block means and correlations for Acquisition 1 (Trials 1-200) and
Acquisition 2 (Trials 201-400) are displayed for each subject in Figure 11. All three participants
increased speed and decreased spatial error by the end of the extended practice session (L7:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
147
Subject L2
T rial 211
RMS E s 126.7
Trial 281
T F 1 .1 8
o
•32
300 0 • 0 0 0 0 0 1200 1 S 0 0
RM S Es * 34.6:
TF 1 35
AF 1 22
0
300 0 0 0 1200 1500 0 •00
Trial 311
RMS E s 142.9' T ria l 381
• 4
TF 1.72!
AF 1.59
32
0
3 2
300 0 0 0 1200 0 • 00 1 0 0 0
Time (ms)
RM S Es 38.6
AF 1 43
0
1200 tooo 0 400 • 0 0
Tima (ms)
Figure 12. Individual trial data for subject L2 during extended practice. The goal movement is the
thick black line. RMSEs for the subject’s actual movement (thin black line), TF from the time
scaled movement (dash line), and AF from the amplitude scaled movement (dot-dash line) is
displayed in the table for each trial.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
148
Subject R4
T ria l 2 28
RUSEs 43.0
9 4
0 1
TF 1.68
AF 1.52
a n
o
3 0 0 1 2 0 0 0 9 0 0 t OOQ I I
T rial 264
A
:•!
.7
RUSEs 42.8
TF 158
AF 1 38
1200 1500
T ria l 337
RHSEt 35.3
TF 1.33
AF 1.45 2 2
0
5 0 0 1000 0 1500
Tims (ms)
Trial 382
RUSEs 18.0
TF
AF 1.04
9
1000 5 0 0 9
Time (ms)
Figure 13. Individual trial data for subject R4 during extended practice. The goal movement is
the thick black line. RMSE for the subject’s actual movement (thin black line). TF from the time
scaled movement (dash line), and AF from the amplitude scaled movement (dot-dash line) is
displayed in the table for each trial.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
149
Block 1. TF=2.60, AF=1.68; Block 40, TF=1.10, AF=1.12; L2: Block 1, TF=2.31, AF=1.86; Block
40, TF=1.46, AF=1.53; R4: Block 1, TF=2.95, AF=2.56; Block 40, TF=1.20, AF=1.06). However,
the correlation between TF and AF for the 2 acquisition phases did not decrease for any of the 3
subjects (correlations for Acquisition 1 and Acquisition 2: L7, r,=.69, r2 =.84; L2, r,=.50, r2 =.76; R4,
r,= 66, r2 =.78). In fact, in all 3 cases, the correlations significantly increased between the initial
acquisition and the extended practice period (f=-4.14; p=.05). These findings suggest that
throughout both phases temporal and spatial scaling remained tightly coupled even though
improvements in motor performance were evident Interestingly, the couping of temporal and
amplitude scaling increased with extended practice suggesting that the subjects with sensorimotor
cortical system damage were reinforcing this movement parameter strategy across practice.
Discussion
A feature of the SI or pulse-width control strategy is that the initial rate of rise in
acceleration is constant therefore, MT and amplitude are proportionately and directly related to
one another. Sensorimotor cortical system damage results in the coupling of temporal and spatial
control parameters consistent with a pulse-width control strategy. This motor control deficit,
present after unilateral sensorimotor cortical system damage, appears to be a fundamental motor
control impairment since even with extended practice a pulse-width control strategy is evident
General Discussion
The results from this series of experiments suggest that the sensorimotor cortical system
has a major role in the parameter control of rapid single-joint movements. Unilateral damage to
the sensorimotor cortical system resulted in movements that were slower and hypermetric
compared to that for controls. These deficits were in the ipsilesional limb and suggest that
accurate temporal and spatial scaling of programmed movements requires bilateral activation of
sensorimotor areas. In addition, sensorimotor cortical system damage resulted in impairments in
control strategies that reflect a deficit in motor control that did not change with the extended
practice provided in this study. After unilateral sensorimotor cortical system damage, kinematic
analysis reveals that temporal and spatial parameters are tightly coupled; therefore, stroke
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
150
subjects appear to be restricted to a pulse-width control strategy. This observation appears to be
directly related to damage within this neuroanatomic network since neurologically healthy subjects
of different age groups were able to use a pulse-height control strategy, a strategy more
appropriate for the task constraints imposed by the movement task goals used in this study
(Gottlieb etal., 1989a).
The persistent deficit in parameter control may reflect the role of the sensorimotor cortical
system in the modulation of motoneuronal pools. After sensorimotor cortical system damage, the
ability to modulate intensity is impaired: therefore, the only way to meet increased force demands
is to modulate pulse duration. The decrease in central drive to motoneuron pools is most likely
attributable to the loss of cortical neurons and their associated subcortical projections. The
performance during extended practice of the three stroke subjects suggests that this motor
control deficit is not remediable with practice and may reflect a relatively permanent deficit in
parameter control after unilateral sensorimotor cortical system damage.
Gottlieb et al. (1989b) have suggested that one of the purposes of theory is not only to
predict behavior but to provide a norm with which deviations can be identified. Most motor control
studies that have examined the control of rapid, single-joint movements have varied task
constraints in order to evoke a nervous system control response. In this study, the task goal was
to acquire a rapid limb movement with specific temporal and spatial constraints under two varying
nervous system conditions. Previous work (Winstein et al., 1998) has revealed that control
differences exist between those with and without unilateral brain-damage learning a rapid upper
limb movement but because of limitations in methods and the lack of lesion data on brain
damaged subjects there could be no correlation between motor control deficits and the neural
contribution to that control. This study has addressed the limitations of this previous work by
confirming the presence of unilateral sensorimotor cortical system damage and including
kinematic analyses. One of the limitations of the current study is there were no a priori
hypotheses about control strategies for the subjects in this study. Therefore, future studies that
include EMG would allow for comparisons between kinematic and EMG findings that would more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
151
definitively provide evidence for pulse-height or pulse-width control strategies in the acquisition
and execution of rapid limb movements after brain damage.
Effects of lesion location
Temporal and amplitude parameter control was influenced by damage within the
sensorimotor cortical system but deficits in parameter control were not related to hemisphere.
Group differences in parameter control but a lack of group by arm-used interaction in acquisition
and retention phases suggests that sensorimotor cortical system damage affects temporal and
amplitude parameter scaling that is independent of hemisphere. This finding was in contrast to
hypotheses that predicted an effect of left-hemisphere damage on temporal scaling and right-
hemisphere damage on spatial scaling. One explanation for this lack of a hemispheric effect may
have to do with the dissociation in central processes related to programming and
parameterization. Previous studies (Sekiya et al., 1994; Wulf & Schmidt 1994; Wulf et al., 1994;
Wulf et al., 1993) have demonstrated a behavioral dissociation between these two processes that
has been replicated in Chapters 5 and 6. Together the results from Chapters 5 and 6 extend
previous work and provide evidence for a neuroanatomic basis for programming and
parameterization processes. Chapter 5 established that sensorimotor cortical system damage
affected motor programming that was influenced by hemispheric differences. Motor program
accuracy and acquisition was affected by the left- and right-hemispheres, respectively. The
results from this study reveals that deficits in parameter control are influenced by sensorimotor
cortical system damage but not by hemispheric differences. This finding suggests that neuronal
pools within the sensorimotor cortical system, independent of hemisphere, modulate the control of
temporal and spatial parameterization. The findings from Chapter 5 and this study provide
support for separate functional neuroanatomic networks that subserve the control of programming
from that of parameterization.
Further support for a sensorimotor cortical system network modulating parameter control
processes comes from the lesion location analysis that compared temporal and spatial coupling in
groups with 1) cortical and subcortical damage, 2) subcortical damage only, 3) pontine damage.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
152
and 4) healthy controls. In this analysis, group differences in temporal and spatial coupling were
large in the cortical and subcortical group and subcortical only group compared to controls. In
contrast the one subject with pontine damage had extremely low temporal and spatial coupling
(r=.11). Together these findings suggest that parameter control strategies are influenced by
sensorimotor cortical system damage at either the level of the cortex or within subcortical
structures. Although cautious interpretation is warranted, damage at the level of the pons did not
result in temporal and spatial coupling which may suggest that the central processing related to
parameter control occurred in neural networks rostral to the pons. In addition, this finding
suggests that parameter control is not relegated to cortical cell-assemblies exclusively. In fact,
the presence of these parameter control deficits in the ipsiiateral limb and with lesions localized to
subcortical structures provides strong evidence that the control of rapid movements with highly
specified temporal and spatial goals involves neuronal modulation between the sensorimotor
cortical systems of both hemispheres. Further support for this observation is provided by a study
examining the spatial and temporal coupling of bimanual movements in healthy controls and
individuals with callostomy (Franz, Eliassen, Ivry, & Gazzaniga, 1996). Using a bimanual upper
limb coordination task, these authors demonstrated that spatial coupling between the upper limbs
required bilateral hemispheric activation mediated through the corpus callosum. Temporal
coupling was maintained regardless of corpus callosum integrity. This study provides evidence
that the integration of parameter processes does occur between hemispheres. Future research
that examines the bilateral sensorimotor cortical system contributions to parameter control are
needed.
Summary
In summary, this study provides evidence that the sensorimotor cortical system has a
major role in temporal and spatial parameter control that is not influenced by the side of
hemispheric damage. Practice-related improvements in movement speed and accuracy support
the capability of the sensorimotor cortical system to adapt, leading to experience-dependent
improvements in motor performance. Despite this capability to improve, persistent deficits in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
153
absolute temporal and spatial error (i.e., slower, hypermetric movements) and parameter control
strategies (i.e., coupling of temporal and spatial parameters) may reflect fundamental motor
control impairments suggestive of persistent neurologic damage. Therefore, it appears that the
control of rapid unimanual movements with highly specified temporal and spatial goals is
modulated by processing within the sensorimotor cortical system of both hemispheres.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 7
154
PREDICTING UPPER EXTREMITY MOTOR IMPAIRMENT AND PHYSICAL DISABILITY
AFTER MIDDLE CEREBRAL ARTERY STROKE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
155
Abstract
The disablement model proposed by Nagi (1991) was used as a conceptual framework to
investigate the predictive validity of 1) post-stroke pathology (lesion location and extent) on the
severity of upper extremity (UE) motor impairment and 2) a clinical disablement model that
includes indices of pathology, impairment (contralateral and ipsilateral Box & Block and grip
strength, contralateral UE Fugl-Meyer motor score), and functional limitation (Functional
Independence Measure; FIM) on post-stroke physical disability outcome (SF-36). Nineteen
subjects with unilateral middle cerebral artery stroke and one subject with basilar stroke who
presented with hemiparesis and hemisensory loss participated in this study.
Univariate correlation revealed that UE motor severity is directly related to lesion extent;
larger lesions resulted in more severe UE motor impairment (r = -.59; p < .01). However,
multivariate regression analysis revealed that lesion location accounts for 87% of the variance in
motor severity when both lesion location and extent are known. Lesions that include the
striatocapsular region are associated with greater motor severity. Multivariate stepwise
regression with all disablement model components revealed that the FIM was the single
independent correlate of physical disability post-stroke such that 51% of the variance in SF-36
was predicted by the FIM. A separate analysis was completed to determine the impairments that
contribute to physical disability. Contralateral UE FMMS, contralateral grip strength, and
ipsilateral grip strength were independently correlated with physical disability and accounted for a
comparable proportion of the variance explained in physical disability as the FIM score alone (R2 =
.63). This study suggests that a physical disablement model can be an effective framework to
investigate the multivariate contributions to motor severity and physical disability post-stroke.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
156
Introduction
This final chapter makes a deliberate departure from the previous chapters in order to
investigate the impact that sensorimotor cortical system stroke has on physical disability. In
particular, this chapter focuses on the relationship between the primary pathology when stroke
occurs in the sensorimotor cortices, the adjacent association areas, and/or the subcortical
striatocapsular region and the resultant sensorimotor neurologic impairments. Therefore, this
chapter is not concerned with the cognitive processing that involves sensorimotor areas during
complex motor skill acquisition but rather the direct effect that sensorimotor dysfunction has on
motor execution and functional ability.
Stroke is a leading cause of disability in the United States (Gresham, Duncan, Stason et
al.. 1995) and across the world (Bonita, Solomon, & Broad, 1997; Wyller, Bautz-Holter, & Holman,
1994). Determining the prevalence of stroke within a population is difficult since it is influenced by
incidence, life-span duration post-stroke, and the effects of age and gender within the population
(Bonita et al., 1997). However, studies that compare different methods suggest that stroke
prevalence is about 1% of the population 50 years and younger, and 10% of the population over
80 years. Approximately, 33% of the younger group and 75% of the older group will not recover
completely. These stroke survivors will live with persistent impairments and disability from the
combined effects of stroke and other comorbid factors (Wyller, 1998).
Historically, the profession of physical therapy has been concerned with the disabling
consequences of disease and trauma (Rothstein, 1994). Despite a fundamental commitment by
our profession to remediate disability, a preponderance of outcome research and education is
devoted to the quantification and remediation of impairment Considering the prevalence of
stroke-related disability and the role of physical therapy in physical disability remediation, it is
necessary to begin to develop and test conceptual models within physical therapy that focus on
stroke-related disability. By understanding the relationships between impairment functional
limitation, and disability in a population with stroke, a theoretical basis for physical therapy stroke
intervention can be developed (Duncan, 1994).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
157
Jette (1994) has proposed that disablement models may provide a ‘basic conceptual
architecture for disablement research in physical therapy” (Jette, 1994, p. 380). According to
Jette (1994), disablement is a result of chronic or acute conditions that reflect the consequences
that the disease or injury has on human functioning. Disablement is a complex process that
involves the dynamic interaction of physical, cognitive, social, and environmental factors (World
Health Organization, 1980; Pope & Tartov, 1991). Several disablement models have been
developed to reflect these complexities (See review, Jette, 1994). However, for the purposes of
this study, the conceptual disablement scheme developed by Nagi will be used (Nagi, 1991). The
disablement terms used in Nagi’s model include pathology, impairment, functional limitation, and
disability. For each of these constructs, the definitions, indicators, and outcome measures that
will be used in this study are summarized in Table 1.
Stroke is characterized by the rapid, focal onset of neurologic signs that last longer than
24 hrs and is caused by a vascular pathology (Aminoff, Greenberg, & Simon, 1996). The majority
of ischemic strokes, approximately 64%, are within the distribution of the middle cerebral artery
(MCA; Garcia & Anderson, 1997). The MCA distribution (Gabella, 1995) includes central
branches (medial striate arteries to the (entiform nucleus; lateral striate arteries to the internal
capsule and caudate nucleus) and cortical branches (frontal branches to precentral, middle, and
frontal gyri; parietal branches to the postcentral gyrus, inferior portion of the superior parietal
lobule, and entire inferior parietal lobule; temporal branches to the lateral surface of the temporal
lobe). Strokes within this distribution are commonly associated with signs and symptoms such as
hemiparesis and hemisensory loss, visual field defects, and in left-hemisphere lesions, aphasia
(Aminoff et al., 1996). Recent studies have suggested that the severity of motor impairment and
the degree of motor recovery post-stroke are related more to lesion location than lesion volume
(Kunesch, Binkofski, Steinmetz, & Freund, 1996; Binkofski, Seitz, Arnold, Classen, Benecke, &
Freund, 1996). The greatest motor severity and the poorest functional recovery were associated
with lesions that included striatocapsular areas. While studies such as these relate pathology with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission.
Table 1. Disablement Model: Nagi Scheme
Disabling Process:
Pathology Impairment
Functional
Limitation
Disability
Definition:
Disease process; Interruption
of normal processes.
Loss or abnormality at tissue,
organ, or body system level.
Limitation in performance at
the level of the whole person.
Limitation in performance of
socially defined roles and
tasks within a socioeconomic
or physical environment.
Indicators:
Diagnoses of disease or injury.
Measures:
Lesion location
Lesion volume
Signs, symptoms; dysfunction
in specific body systems (i.e.,
neuromuscular, musculo
skeletal).
Contralateral UE FMMS
Contralateral UE FMSS
Contralateral grip strength
Ipsilateral grip strength
Contralateral Box & Blocks
Ipsilateral Box & Blocks
Restrictions in basic physical
and mental actions such as
ambulate, reach, produce
speech.
FIM (Motor subscale)
Difficulty in activities of daily
life such as job, household
management, self care.
SF-36 (Physical function
subscale)
Upper extremity (UE), Fugl-Meyer motor score (FMMS), Fugl-Meyer sensory score (FMSS), Functional independence measure (FIM).
158
159
impairment severity, no studies were found that examined if pathology predicts physical disability
post-stroke.
Several studies have demonstrated that initial motor impairment seventy is associated
with poor recovery of motor function (Bonita & Beaglehole, 1988; Duncan, Goldstein, Matchar,
Divine, & Feussner, 1992; Gowland, 1984; Heller, Wade, Wood, Sunderland, Langton-Hewer, &
Ward, 1987; Olsen, 1990; Wade, Langton-Hewer, Wood, Skilbeck, & Ismail, 1983) and there is a
direct correlation between stroke-related impairment and functional outcome (Chae, Johnston,
Kim, &Zorowitz, 1995; Roth, Heinemann, Lovell, Harvey. McGuire, & Diaz, 1998; Sonoda,
Chino,Domen, & Saitoh, 1997). A meta-analysis of 78 research studies was completed to identify
variables that predicted functional outcome after stroke (Kwakkel, Wagenaar, Kollen, &
Lankhorst, 1996). The results indicated that functional recovery after stroke was predicted by:
age, previous stroke, urinary continence, consciousness at onset disorientation to time and place,
severity of paralysis, sitting balance, admission ADL score, level of social support and metabolic
rate of glucose outside the infarct area in hypertensive patients. This comprehensive list reveals
the multiple factors that contribute to stroke disablement However, no studies were found that
explicitly investigated the Nagi disablement model and the relationships between pathology,
impairment, and functional limitation on physical disability post-stroke. Additionally, no physical
therapy studies were found that have used a disablement model as a conceptual framework for
guiding physical therapy practice in stroke rehabilitation.
The purpose of this study was to examine the correlation between pathology, impairment
functional limitation, and physical disability in a stroke population with neurologic damage within
the distribution of the MCA. This Chapter will focus on the following relationships. The first aim
was to determine if lesion location and extent is predictive of the severity of contralateral upper
extremity (UE) motor recovery as measured by the Fugl-Meyer Motor Score (FMMS). It is
hypothesized that lesions which include the striatocapsular areas will result in greater upper
extremity motor impairment compared to other lesion locations regardless of lesion extent The
second aim of this Chapter is to investigate the predictive validity of a clinical disablement model
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
that includes indices of pathology, UE sensorimotor impairment and functional limitation on
physical disability outcome.
Method
Subjects
The twenty individuals with stroke described in Chapter 5 were included in this study. All
subjects voluntarily signed an informed consent that included 1) release of CAT or MRI scans for
lesion location and extent analysis and 2) approval for clinical assessment of sensorimotor
performance, functional status, and level of physical disability.
Testing Procedures
All lesion analyses and clinical assessments were completed by the author, a California
licensed physical therapist Lesion locations were confirmed by a Board Certified Neurologist
The primary outcome variables (See Table 1, Measures) were selected to represent the
constructs of pathology, impairment functional limitation, and physical disability as described by
the Nagi disablement process model (Nagi, 1991). Impairment assessments were limited to UE
function.
Pathology
Pathology was indexed by the extent and location of the neurologic lesion. Lesion extent
was calculated as the percentage of the total cortical and subcortical brain volume. Therefore, the
ventricles, cerebellum, and brainstem volumes were excluded. The procedure for the volume
analysis is described in detail in Chapter 5. The lesion location for each individual stroke subject
is described in Chapter 5, Table 1. All subjects had stroke within the distribution of the MCA with
one exception. Subject R7 had a basilar stroke that resulted in medial pontine damage. This
subject was not included in the analysis of lesion location and extent but was included in all other
analyses because the sensorimotor clincal presentation of medial pontine stroke is similar to MCA
stroke. For the analyses in this Chapter, subjects were categorized into lesion groupings based
upon pyramidal tract or capsular involvement Previous work has demonstrated that motor
severity post-stroke is greatest with striatocapsular (SC) lesions that include pyramidal tract (i.e.,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
161
posterior limb of the internal capsule, PT) damage (Binkofski et al., 1996; Kunesch et al., 1995).
Therefore, scans were reviewed to determine the specific neuroanatomic location of each lesion.
This review resulted in the following lesion location groupings; 1) Cortex + Striatocapsular
(Cx±SC), 2) SC only (SC), 3) Thalamic + PT (Thal+PT), 4) Striate no PT (STR), 5) Cx no PT (Cx),
and 6) Pontine (P).
Impairment The following variables were measured to determine the level of motor and
sensory impairment
Selective motor control The UE motor component of the Fugl-Meyer assessment was
used to determine the severity of motor selectivity for the contralateral upper limb. The Fugl-
Meyer sensorimotor assessment is a valid and reliable measure of motor control and
somatosensory function post-stroke (Berglund & Fugl-Meyer, 1986; Duncan, Prodst, & Nelson,
1983; Fugl-Meyer et al., 1975). The total possible score for the UE motor component is 66 points;
lower scores indicate more severe motor impairment
Somatosensory function The UE sensory component of the Fugl-Meyer assessment was
used to determine the degree of sensory impairment for the contralateral upper limb. The
maximum value for the Fugl-Meyer sensory score (FMSS) is 12 points; lower scores indicate
more severe sensory impairment
Strength Ipsiiateral and contralateral grip strength was determined with a standard Jamar
hand-held force dynamometer. The same dynamometer was used for all assessments. Subjects
were seated with the elbow in approximately 90 degrees of flexion. Subjects were encouraged to
grip the dynamometer handle and to squeeze as hard as possible. During contralateral grip
measurements, if needed, the experimenter would support the weight of the dynamometer without
interfering with the subject’s grip performance. Three measurements were taken. The average of
the three measurements was calculated and is reported in pounds as the difference between the
subject's performance and normative values controlled for gender, age, and hand-used
(Mathiowetz, 1990). Therefore, the weakest participants were the ones with negative grip strength
values indicating strength below normative means.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
162
Dexterity Ipsilateral and contralateral upper limb dexterity was determined with the Box
and Blocks test (Mathiowetz et al., 1985). The object of the test is to pick up blocks (one at a
time) from one side of a 2-compartment wooden box and to drop the block on the other side of a 6
inch high barrier. Performance was measured as the total number of blocks moved in 1 minute.
Subjects always started with the ipsilateral limb test, followed by the contralateral limb test
Functional Limitation
The motor sub-scale of the Functional Independence Measure (FIM) was used to
determine the level of functional limitation (Hamilton et al., 1987). The FIM is a commonly used
clinical measure of physical function that reflects the physical burden of care required to perform
13 basic activities of daily living in self-care (eating, grooming, bathing, upper body dressing, lower
body dressing, toileting), sphincter control (bladder-, bowel-management), transfers (bed-, chair-,
wheelchair-transfer, toilet tub or shower), and locomotion (walk or wheelchair, stairs). The
assessment was administered as an interview in which the subject reported the level of
assistance required to complete each of the mobility tasks (score of 7 = complete independence,
score of 1 = complete dependence). The maximum score for the mobility component of the FIM
is 91. Thus, lower scores indicate greater level of functional limitation.
Disability
The physical functioning subscale of the SF-36 was used as the disability outcome
measure for this study (Ware & Sherboume, 1992). The assessment was administered as an
interview in which the subject reported to what degree their current health limited performance on
10 designated physical tasks (tasks range from vigorous activities such as running and lifting
heavy objects to walking one block and bathing/dressing self). The SF-36 is a measure of health
status in which raw scores are transformed to a 0-100 point scale; lower scores indicate greater
level of physical disability.
Data Analysis
Descriptive statistics (means and standard deviations) were calculated for each of the
outcome measures excluding lesion location. Univariate correlations were calculated between
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
163
lesion extent and FMMS, and each of the model independent variables and SF-36. A one-way
factorial ANOVA was used to determine the effect of the categorical grouping factor, lesion
location, on FMMS.
Two separate multivariate regression analyses were completed. For the first regression
analysis, stepwise linear regression was used to determine the independently significant
correlates of lesion location and lesion extent on FMMS. The second regression analysis used
stepwise linear regression to determine independently significant correlates of physical disability.
In this regression, all variables from the proposed disablement model were included to determine
physical disability outcome.
Student t-tests were used for all significance testing with the significance level set at p =
.05.
Results
Table 2 summarizes the data from the 20 subjects with stroke used in the subsequent
univariate correlations and multivariate regression analyses.
Effects of lesion location and extent
Table 3 summarizes the univariate correlation between FMMS and lesion extent The
subject with the pontine infarct was not included in this analysis since volume calculations were
determined by the lesion percent of the cortical and subcortical areas with the brainstem
excluded. Therefore, a brainstem lesion volume is not comparable to lesions within the cortical
and subcortical areas. The severity of upper extremity motor impairment was indirectly correlated
with lesion volume (r = -.59, p < .01). In other words, larger lesion volumes were correlated with
lower FMMS. Figure 1 illustrates the indirect relationship between contralateral UE FMMS and
lesion volume.
Summarized in Table 4, contralateral UE FMMS was significantly affected by lesion
location (main effect p < .0001). Evident in Figure 1, subjects with lesions that included the
striatocapsular area (Cx±SC, SC) had lower FMMS. Post-hoc comparisons revealed that there
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2. Descriptive statistics for outcome variables.
164
Variable n Mean (SD) Range
Pathology:
Extent (% of total volume) 18 9.4 (9.3) .1-28.4
Location
SC+Cx
SC
Thal+PT
STR
Cx-no PT
Pontine
10
3
1
2
3
1
Impairment:
UE Fugl-Meyer motor score (0-66) 20 30.5 (20.3) 8 -6 4
UE Fugl-Meyer sensory score (0-12) 19 6.5 (4.2) 0-12
Ipsilateral grip strength (lbs) f 20 -5.7 (22.8) -58.5 - 33.2
Contralateral grip strength (lbs) f 20 -59.0 (33.1) -109.9-9.1
Ipsilateral Box & Blocks (# in 1 min) 20 56.3 (9.7) 31 -71
Contralateral Box & Blocks (# in 1 min) 20 11.5(12.3) 0 -5 3
Functional Limitation:
Functional Independence Measure
(13-91)
20 80.0 (11.9) 50-91
Disability:
SF-36 (0-100)
20 39.0 (30.1) 0 -9 5
t Grip strength values reported as the difference from normative values controlled for gender,
age, and hand-used, was not a statistically significant difference in FMMS between lesions that
included Cx±SC or SC
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3. Univariate correlation of motor impairment (upper extremity
Fugl-Meyer motor score) with lesion extent
165
Variable r
P
Pathology:
Extent (% of total volume) -.59 < .01
Table 4. Mean upper extremity Fugl-Meyer motor score (+SD) by
lesion location.
Variable Mean (SD)
P
Pathology:
Location
Cx+SC
Thal+PT
STR
Cx-no PT
17.7 (7.1)
22.3 (12.4)
53.0
57.0 (9.9)
61.7(1.2)
p < .0001
Location main effect p<0001. Post-hoc comparison of the Cx+SC and
SC groups was not significant p=.42.
Table 5. Stepwise multivariate analysis of motor impairment (upper extremity
Fugl-Meyer motor score) with pathology.
Construct/Variables
P
R2
Pathology: .87
Extent (%lesion volume) .20
Location (striatocapsular or no-
striatocapsular involvement)
<.0001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
166
only (p = .42). Therefore, for all subsequent analyses subjects were assigned into one of the
following dichotomous groupings: striatocapsular or no-striatocapsuiar involvement
Multivariate stepwise regression with lesion extent and location in the analysis
demonstrated that lesion location was independently correlated with the severity of contralateral
UE motor impairment as measured by the FMMS (Table 5; R2 = .87, p < .001). In other words,
87% of the variance in contralateral UE FMMS was accounted for by lesion location independent
of lesion volume.
o
o
co
o
o
C D
>»
C D
40
— 20
o>
1 1 1
Z5
▲ O
o
▲
+
qb
m
a SC
♦ Cx+SC
o STR
+ Thal+PT
* Cx-no PT
♦
♦
10 15 20
% Lesion Volume
25 30
Figure 1. Upper extremity (UE) Fugl-Meyer scores of the contralateral limb by percent lesion
volume (r = -.59, p <.01). Symbols designate individual lesion location. The horizontal line
indicates a score of 33 out of a maximum score of 66.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
167
Table 6. Univariate correlations of physical disability (SF-36) with model variables.
Variable r
P
Pathology:
Extent (% of total volume) -.20 .44
Location (striatocapsular or no-
striatocapsular involvement)
.31 .19
Impairment:
UE Fugl-Meyer motor score .56 or
UE Fugl-Meyer sensory score .35 .15
Ipsilateral grip strength .20 .40
Contralateral grip strength .10 .69
Ipsilateral Box & Blocks .43 .60
Contralateral Box & Blocks .49 .03*
Functional Limitation:
Functional Independence Measure .72 < .001*
‘statistically significant correlations
Physical disablement model
Table 6 summarizes the univariate correlations between physical disability and the model
independent variables. Physical disability was significantly associated with FMMS (r = .56, p =
.01), contralateral Box & Blocks (r = .49, p = .03), and FIM (r = .72, p < .001). Greater physical
disability (i.e., lower SF-36 score) was associated with lower FMMS (Fig 2), lower contralateral
Box & Blocks score (Fig 3), and lower FIM scores (Fig 4).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 6 8
o
o
c o
w
O
o
©
> »
©
6 0
40
"5. 20
3
U -
Ui
3
▲
▲
A A
A A *
A A
20 40 60
SF-36 Score
80 100
Figure 2. Upper extremity (UE) Fugl-Meyer scores of the contralateral limb by SF-36 score
(r=56, p=01).
o
O
C O
o
o
ca
X
O
C Q
©
©
a s
a s
o
o
5 0
4 0
30
20
10
A
A
A A A A A
2 0 4 0 6 0
SF-36 Score
80 100
Figure 3. Contralateral Box & Blocks score by SF-36 score (r = .49, p = .03).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
169
m
< 0
©
E
©
o
c
©
T 3
C
©
Q _
©
T J
C
Z 60
90
80
70
▲ A
A
A
‘ i A *
u
c
50
20 40 60
SF-36 Score
80 100
Figure 4. Functional independence measure (FIM) by SF-36 score (r = 72, p < .001).
Table 7 summarizes the results of separate regression analyses that determined the
predictive validity of the full disablement model and each of the individual constructs (i.e.,
pathology, impairment functional limitation) on physical disability outcome. For the full model, the
FIM score was the single independent correlate of physical disability such that 51 % of the
variance in SF-36 is predicted by the FIM (R2 = .51, p < .001). Higher FIM scores were associated
with higher SF-36 scores; therefore, as expected physical disability is less in individuals who have
greater levels of physical independence in ADL’s. There were no significant correlations between
physical disability and either of the pathology measures (i.e., lesion location, lesion extent). In
contrast the impairment measures of FMMS, contralateral grip strength, and ipsilateral grip
strength were independently correlated with physical disability accounting for a comparable
proportion of the variance explained in physical disability as the FIM score alone (R2 = .63, p <
.01).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
170
Table 7. Stepwise multivariate analysis of physical disability (SF-36) with
model constructs.
Construct/Variables
P
Ra
Full model: .51
Functional Independence Measure < .001
Pathology only:
Not predictive
Impairment only: .63
UE Fugl-Meyer motor score <05
Contralateral grip strength <05
Ipsilateral grip strength <01
Functional Limitation only: .51
Functional Independence Measure < .001
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
171
Discussion
Clinical disablement model
The relationship between pathology, impairment and functional limitation on stroke
physical disability was examined in this study. Increased disability was associated with increased
physical dependence in ADL’s. In an analysis that determined the UE impairments associated
with physical disability, contralateral FMMS, contralateral grip strength, and ipsilateral grip strength
were independently correlated with physical disability.
This study suggests that a physical disablement model can serve as a conceptual
framework to investigate the multivariate relationships that contribute to physical disability.
Understanding the impact of impairment and functional limitations on physical disability can
enlighten and guide clinical practice. As expected, but tested empirically, poor functional status
(i.e., higher levels of physical assistance in ADL's) is associated with increased physical disability.
This finding supports the importance of task specific training as a clinical intervention since it
suggests that improvements in the level of functional independence most likely will be correlated
with a decrease in physical disability.
The relationship between UE impairment and physical disability was also examined.
Higher levels of physical disability were associated with contralateral UE motor severity (R2 = .31)
and contralateral UE dexterity (R2 = .24). However, a larger proportion of the variance in physical
disability was explained by contralateral UE motor severity, contralateral grip strength, and
ipsilateral grip strength (R2 = .63) than either of the univariate comparisons. These findings are
significant because they reveal that the factors contributing to physical disability are multivariate
and cannot be simply correlated with isolated clinical measures. Clinically meaningful impairment
measures that are associated with changes in function are required to knowledgeably inform
rehabilitation strategies (Duncan, 1994). These results begin to investigate the relationship
between post-stroke impairments and physical disability. The results highlight the importance of
multivariate methods to investigate the complicated relationships that contribute to physical
disability, and suggest that impairment treatment interventions which improve movement
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
172
selectivity and increase strength may be clinically meaningful impairment outcomes that decrease
the level of disability.
Effects of lesion location and size
In this study, the severity of contralateral UE motor impairment was directly related to
large lesions within the distribution of the MCA. However, multivariate regression analysis
demonstrated that lesion location accounts for a larger proportion of the variance in contralateral
UE motor severity when both lesion location and extent are known. Large volume lesions usually
involve infarcts that include both cortical and subcortical structures. It appears that motor
impairment severity is more highly related to subcortical damage since lesions that involved the
striatocapsular region had a greater impact on motor impairment severity regardless of volume.
These findings support the work of Kunesch et al. (1995) who found more severe contralateral
sensorimotor involvement with striatocapsular lesions. In addition. Binkofski et al. (1996) found
that a major determinant of hand motor recovery was pyramidal tract sparing. These authors did
not find a correlation between lesion size and motor impairment and concluded that lesion
location had a greater impact on motor severity and recovery than did lesion size. Figure 5
illustrates the functional neuroanatomy that accounts for this finding as it shows the corticospinal
tract from its origins in the precentral gyrus to its destination in the spinal cord. It is easy to
visualize how a small infarct in the internal capsule would have a greater impact on motor
impairment severity because of the cortical expanse affected by white matter tracts disrupted in
this area. In addition, these findings suggest that the FMMS is a valid indicator of pyramidal
(corticospinal tract) function.
Interestingly, while pathology (i.e., lesion location and volume) was associated with motor
impairment there was no association between pathology and physical disability. Obviously,
neurovascular pathology results in the impairments associated with stroke. However, it appears
that a constellation of impairments and the impact these impairments have on function contribute
more significantly to physical disability than the pathology itself.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
173
Pr acaneaJgyr u*
Cor ona ra d i u s
PoUanor l« W >
i r t a m al cao auX :
'
Figure 5. Pyramidal tract origin and course. Adapted from: (Young & Young, 1997).
Limitations
The overarching purpose of this study was addressed in that a clinical disablement model
appears to be a feasible and informative method to investigate the multivariate contributors to
physical disablement However, there were limitations to this study that need to be addressed in
future work. First the individuals who participated in this study had chronic stroke (median=38
mo; range 6-147 mos). The advantage of studying chronic stroke is that the results are less likely
to be affected by spontaneous recovery or other factors that may impact physical disability during
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
174
the acute post-stroke phase. There is ample evidence that spontaneous recovery occurs within
the first 1-3 mos post-stroke (Bonita & Beaglehole, 1988; Duncan et al., 1992; Duncan, Goldstein,
Homer, Landsman, Samsa, & Matchar, 1994; Olsen, 1990; Wade et al., 1983). It is likely that the
contributors to physical disability during the acute phase may be different from those that impact
physical disablement in later phases. Future research is needed to investigate the relationship
between impairment functional limitation, and physical disability across the stroke recovery
continuum.
Second, the impairment measures used were limited to UE sensorimotor function. Lower
extremity (LE) recovery (either as a sole predictor or in combination with a total UE and LE
FMMS) has been identified as a variable that is highly predictive of functional status (Duncan et
al., 1992; Chae et al., 1995; Gowland, 1984; Olsen, 1990). Future work should include both UE
and LE impairment measures. However, the significance of the UE findings should not be
overshadowed. Separate analyses for the UE and LE may identify separate contributors to
physical disablement Understanding the impact of sensorimotor function in the UE from that of
the LE may identify key intervention strategies that differ for UE and LE post-stroke rehabilitation.
Summary
This study used a clinical disablement model to investigate factors that contribute to post
stroke physical disability. There is a strong relationship between level of physical assistance
required in ADL’s and physical disability; validating the use of the FIM as a clinical measure for
predicting the level of physical disability post-stroke. The multivariate impact of motor
impairments on physical disablement was also demonstrated suggesting that complex
interrelationships exist between the post-strcke impairments that impact physical disability
outcome. Causal modeling can be an effective method for testing these complex
interrelationships and has been used effectively for developing a theoretical basis for clinical
intervention in other disciplines (Boynton De Sepulveda & Chang, 1994). There is a need for
theoretically-driven research in physical therapy. The multivariate relationships identified in this
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
175
study suggest that the Nagi scheme for disablement may be effective as a conceptual model for
physical disablement research in physical therapy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
176
Bibliography
Aizawa, H., Inase, M., Mushiake, H., Shima, K. & Tanji, J. (1991). Reorganization of activity in
the supplementary motor area associated with motor learning and functional recovery.
Experimental Brain Research. 84. 668-671.
Aminoff, M.J., Greenberg, D.A., & Simon, R.P. (1996). Clinical Neurology. 3rd Ed. Appleton
LangerStamford, C t
Arbib, M.A. (1981). Perceptual structures and distributed motor control. In: V.B. Brooks (Ed.),
Handbook of phvsioloov. Sec 1. Vol 2. (pp. 1449-1480). Bethesda, MD: American
Physiologic Society.
Berglund, K. & Fugl-Meyer, A.(1986). Upper extremity function in hemiplegia: a cross-
validation study of two assessment methods. Scandinavian Journal of Rehabilitation.
18: 155-157.
Binkofski, F„ Seitz, R.J., Arnold, S., Classen, J., Benecke, R. & Freund, H-J. (1996). Thalamic
metabolism and corticospinal tract integrity determine motor recovery in stroke.
Annals of Neurology. 39: 460-470.
Bjork, R.A. & Bjork, E.L. (1992). A new theory of disuse and an old theory of stimulus
fluctuation. In: A.F. Healy, S.M. Kosslyn. & R.M. Shiffrin (Eds.), From learning
processes to cognitive processes: Essavs in honor of William K . Estes. (Vol. 2, pp. 35-
67). Hillsdale, NJ: Eribaum.
Bondi, M.W., Kaszniak, A.W., Rapcsak, S.Z. & Butters, M.A. (1993). Implicit and explicit
memory following anterior communicating artery aneurysm rupture. Brain and
Cognition. 22, 213-229.
Bonita, R. & Beaglehole, R. (1988). Recovery of motor function after stroke. Stroke. 19. 1947-
1500.
Bonita, R., Solomon, N., & Broad, J.B. (1997). Prevelance of stroke and stroke-related
disability. Stroke. 2 8 .1898-1902.
Boyton De Sepulveda, L.I. & Chang, B. (1994). Effective coping with stroke disability in a
community setting: The development of a causal model. Journal of Neuroscience
Nursing. 26. 193-203.
Brinkman, C. & Porter, R. (1979). Supplementary motor area in the monkey: Activity of
neurons during performance of a learned motor task. Journal of Neurophvsioloav. 42,
681-709.
Brooks, V.B. (1979). Motor programs revisited. In: R.E. Talbot & D.R. Humphrey (Eds.),
Posture and Movement (pp. 13-48). New York: Raven Press.
Brooks, V.B., Hilperath, F., Brooks, M., Ross, H-G., & Freund, H-J. (1995). Learning “ What”
and “How” in a human motor task. Learning and Memory. 2. 225-242.
Brooks, V.B., Kennedy, P.R. & Ross, H.-G. (1983). Movement programming depends on
understanding of the behavioral requirements. Physiology and Behavior. 31. 561-563.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
177
Brooks, V.B. & Watts, S.L. (1988). Adaptive programming of arm movements. Journal of Motor
Behavior. 20. 117-132.
Buckner, R.L., Petersen, S.E., Ojemann, J.G., Miezin, F.M., Squire, L.R. & Raichle, M.E.
(1995). Functional anatomic studies of explicit and implicit memory retrieval tasks.
Journal of Neuroscience. 1 5 .12-29.
Chae, J., Johnston, M„ Kim, H„ & Zorowitz, R. (1995). Admission motor impairment as a
predictor of physical disability after stroke rehabilitation. American Journal of Physical
Medicine and Rehabilitation. 74. 218-223.
Chollet, M.D., DiPiero, V., Wise, R.J.S., Wise, M.D., Brooks, D.J., Dolan, M.D. & Frackowiak,
R.S.J. (1991). The functional anatomy of motor recovery after stroke in humans: A
study with positron emission tomography. Annals of Neurology. 29. 63-71.
Colebatch, J.G., Deiber, M.-P., Passingham, R.E., Friston, K.J. & Frackowiak, R.S.J. (1991).
Regional cerebral blood flow during voluntary arm and hand movements in human
subjects. Journal of Neurophvsioloav. 6 5 .1392-1401.
Corkin, S. (1968). Acquisition of motor skill after bilateral medial temporal-lobe excision.
NeuropsvcholoQia. §. 255-265.
Crossman, E.R.F.W. (1959). A theory of the acquisition of speed-skill. Ergonomics. 2, 153-
166.
Cushman, L. & Caplan, B. (1987). Multiple memory systems: Evidence from stroke. Perceptual
and Motor Skills. 64. 571-577.
Damasio, H. (1995). Human Brain Anatomy in Computerized Images. New York: Oxford
University Press.
Damasio, H. & Damasio, A.R. (1989). Lesion analysis in neuropsychology. New York: Oxford
University Press.
Darling, W.G. & Cooke, J.D. (1987). Changes in the variability of movement trajectories with
practice. Journal of Motor Behavior. 19. 291-309.
Darling, W.G., Cooke, J.D. & Brown, S.H. (1989). Control of simple arm movements in elderly
humans. Neurobioloav of Aging. 10.149-157.
Data-Pac III (Version 1.59) [Computer software], (1997). Laguna Hills, CA: RUN
Technologies.
Deiber, M.-P., Passingham, R.E., Colebatch, J.G., Friston, K.J., Nixon, P.D. & Frackowiak,
R.S.J. (1991). Cortical areas and the selection of movement A study with positron
emission tomography. Experimental Brain Research. 84. 393-402.
Deweer, B., Ergis, A.M., Fossati, P., Pillon, B., Boiler, F., Agid, Y. & Dubois, B. (1994). Explicit
memory, procedural learning and lexical priming in Alzheimer’s disease. Cortex. 30.
113-126.
Donoghue, J.P. & Sanes, J.N. (1994). Motor areas of the cerebral cortex. Journal of Clinical
Neurophvsioloov. 11. 382-396.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
178
Doyon, J., Gaudreau, R., Laforce, R., Castonguay, M., Bedard, P.J., Bedard, F. & Bouchard, J-
P. (1997). Role of the striatum, cerebellum, and frontal lobes in the learning of a
visuomotor sequence. Brain and Coonifa'on. 34. 218-245.
Doyon, J., Owen, A.M., Petrides, M„ Sziklas, V. & Evans, AC. (1996). Functional anatomy of
visumotor skill learning inhuman subjects examined with postron emission
tomography. European Journal of Neuroscience. 8, 637-648.
Doyon, J., Laforce, Bouchard, G., R., Gaudreau, D., Roy, J., Poirier, M„ Bedard, P.J., Bedard,
F. & Bouchard, J-P. (In press). Role of the striatum, cerebellum, and frontal lobes in
the automatization of a repeated visuomotor sequence of movements.
Neuropsvcholoqia.
Druckman, D. & Bjork, R.A. (1991). Optimizing long-term retention and transfer. In: D.
Druckman. & R.A. Bjork (Eds.), In the Mind’s Eve: Enhancing Human Performance.
(pp. 23-56). Washington, D.C.: National Academy Press.
Duncan, P.W. (1994). Stroke disability. Physical Therapy. 74. 399-407.
Duncan, P.W., Goldstein, LB., Homer, R.D., Landsman, P.B., Samsa, G.P., & Matchar, D.B.
(1994). Similar motor recovery of upper and lower extremities after stroke. Stroke. 25.
1181-1188.
Duncan, P.W., Goldstein, L.B., Matchar, D.B., Divine, G.W., & Feussner, J. (1992). Similar
motor recovery of upper and lower extremities after stroke. Stroke. 23. 1084-1089.
Duncan, P.W., Propst, M., & Nelson, S.G. (1983). Reliability of the Fugl-Meyer assessment of
sensorimotor recovery following cerebrovascular accident Physical Therapy. 63.
1606-1610.
Ewert J., Levin, H.S., Watson, M.G. & Kalisky, Z. (1989). Procedural memory during post-
traumatic amnesia in survivors of severe closed head injury. Archives of Neurology.
46, 911-916.
Fisk, J.D. & Goodale, M.A. (1988). The effects of unilateral brain damage on visually guided
reaching: Hemispheric differences in the nature of the deficit Experimental Brain
Research. 72, 425-435.
Fitts, P.M. (1954). The information capacity of the human motor system in controlling the
amplitude of movement Journal of Experimental Psychology. 47. 381-391.
Fitts, P.M. & Peterson, J.R. (1964). Information capacity of discrete motor responses. Journal
of Experimental Psychology. 6 7 .103-112.
Folstein, M.F., Folstein, S.E., & McHugh, P.R. (1975). Mini-mental state: A practical method for
grading the cognitive state of patients for the clinician. Journal of Psychiatric
Research. 12. 189-198.
Fowler, C.A. &Turvey, M.T. (1978). Skill acquisition: An event approach with special reference
to searching for the optimum of a function of several variables. In: G.E. Stelmach
(Ed.), Information Processing in Motor Control and Learning, (pp. 1-40). New York:
Academic Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
179
Franz, E.A., Eliassen, J.C., Ivry, R.B., & Gazzaniga, M.S. (1996). Dissociation of spatial and
temporal coupling in the bimanual movements of callosotomy patients. Psychological
Science. 7, 306-310.
Fu, Q.-G., Flament, J.D., & Ebner, T.J. (1995). Temporal encoding of movement kinematics in
the discharge of primate primary motor and premotor neurons. Journal of
Fu, Q.-G., Suarez, J.I., & Ebner, T.J. (1993). Neuronal specification of direction and distance
during reaching movements in the superior precentral premotor area and motor cortex
of monkeys. Journal of Neurophvsiologv. 70. 2097-2116.
Fugl-Meyer, A.R., Jaasko, L., Leyman, I., Olsson, S. & Steglind, S. (1975). The post stroke
hemiplegic patient I. A method for evaluation of physical performance. Scandinavian
Journal of Rehabilitation Medicine. 7,13-31.
Gabella, G. (1995). Cardiovascular. In: P.L. Williams et al. (Eds.), Gray’s anatomy (pp.1451-
1626). London: Churchill Livingstone.
Garcia, J.H. & Anderson, M.L. (1997). Circulatory disorders and their effects on the brain.
From: Davis, R.L. & Robertson, D.M. Textbook of Neuropathology, 3rd Ed. pp.715-
787. Williams & Wilkins: Baltimore.
Georgopoulos, A.P. (1991). Higher order motor control. Annual Reviews in Neuroscience. 14.
361-377.
Georgopoulos, A.P., Taira, M. & Lukashin, A (1993). Cognitive neurophysiology of the motor
cortex. Science. 260. 47-52.
Goodale, M.A. (1988). Hemispheric differences in motor control. Behavioural Brain Research.
30, 203-214.
Gordon, J. & Ghez, C. (1987). Trajectory control in targeted force impulses: II. Pulse-height
control. Experimental Brain Research. 67. 241-252.
Gowland, C. (1984). Predicting sensorimotor recovery following stroke rehabilitation.
Physiotherapy Canada. 36. 313-320.
Gottlieb, G.L., Chen, C-H., & Corcos, D.M. (1995). “ Adequate control theory* for human single
joint elbow flexion on two tasks. Annals of Biomedical Engineering. 23. 388-398.
Gottlieb, G.L., Chen, C-H., & Corcos, D.M. (1996). Nonlinear control of movement distnce at
the human elbow. Experimental Brain Research. 112. 289-297.
Gottlieb, G.L., Corcos, D.M., & Agarwal, G.C. (1989a). Strategies for the control of voluntary
movements with one mechanical degree of freedom. Behavioral and Brain Sciences.
12, 189-250.
Gottlieb, G.L, Corcos, D.M., & Agarwal, G.C. (1989b). Organizing principles for single-joint
movements: II. A speed-sensitive strategy. Journal of Neurophvsiologv. 62. 358-368.
Gottlieb, G.L., Corcos, D.M., & Agarwal, G.C. (1989c). Organizing principles for single-joint
movements: I. A speed-insensitive strategy. Journal of Neurophvsiologv. 62. 342-357.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
Grafton, S.T., Fagg, A.H., Woods, R.P., & Arbib, M.A. (1996). Functional anatomy of pointing
and grasping in humans. Cerebral Cortex. §, 226-237.
Grafton, S.T., Hazelti'ne, 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, R.S.J. & Phelps, M.E.
(1992). Functional anatomy of human procedural learning determined with regional
cerebral blood flow and PET. Journal of Neuroscience. 12. 2542-2548.
Grafton, S.T., Mazziotta, J.C., Woods, R.P. & Phelps, M.E. (1992). Human functional anatomy
of visually guided finger movements. Brain. 115. 565-587.
Green, R.E. & Shanks. D.R. (1993). On the existence of independent explicit and implicit
learning systems: An examination of some evidence. Memory and Cognition. 21. 304-
317.
Gresham. G.E., Duncan, P.W., Stason, W.B., etal. (1995). Post-Stroke Rehabilitation. Clinical
Practice Guideline, No. 16. Rockville, MD: U.S. Department of Health and Human
Services. Public Health Service, Agency for Health Care Policy and Research. AHCPR
Publication No. 95-0662.
Grober, E , Ausubel, R., Sliwinski, M. & Gordon, B. (1992). Skill learning and repetition priming
in Alzheimer's disease. Neuropsvcholooia. 30. 849-858.
Haaland, K.Y. & Harrington, D.L. (1990). Complex movement behavior Toward understanding
cortical and subcortical interactions in regulating control processes. In: G.E. Hammond
(Ed.), Cerebral control of speech and limb movements, (pp. 169-200). North-Holland:
Elsevier Science Publishers B.V.
Haaland, K.Y., Harrington, D.L & Yeo, R. (1987). The effects of task complexity on motor
performance in left and right CVA patients. Neuropsvcholoqia. 25. 783-794.
Haier, R.J., Siegel, B.V., MacLachlan, A., Soderiing, E , Lottenberg, S. & Buchsbaum. M.S.
(1992). Regional glucose metabolic changes after learning a complex
visuospatial/motor task: A postitron emission tomographic study. Brain Research. 570.
134-143.
Halsband, U. & Freund, H.-J. (1990). Premotor cortex and conditional motor learning in man.
Brain. 113. 207-222.
Halsband, U. & Passingham, R.E. (1985). Premotor cortex and the conditions for movement in
monkeys (Macaca fasicularis). Behavioural Brain Research. 18. 269-277.
Hamilton, B.B., Granger, C.V., Sherwin, F.S., Zielezny, M„ & tashman, J.S. (1987). A uniform
national data system for medical rehabilitation. In: M. Fuhrer (Ed.), Rehabilitation
outcomes analysis and measurement (pp. 137-147). Baltimore, MD: Brookes
Publication, Inc..
Hanlon, R.E. (1996). Motor learning following unilateral stroke. Archives of Physical Medicine
and Rehabilitation. 71. 811-815.
Hary, D. (1996). Template (Version 2.0) [Computer software]. Santa Monica: Integrated
Scientific Resources.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
181
Hebb, D.O. (1949). The Organization of Behavior A Neuropsychological Theory. New York:
Wiley.
Heindel, W.C., Salmon, D.P., Shults, C.W., Walicke, PA. & Butters, N. (1989).
Neuropsychological evidence for multiple implicit memory systems: A comparison of
Alzheimer's, Huntington’s, and Parkinson's Disease patients. Journal of Neuroscience.
2, 582-587.
Heller, A., Wade, D.T., Wood, V.A., Sunderland, A.. Langton-Hewer, R., & Ward, E. (1987).
Arm function after stroke: Measurement and recovery over the first three months.
Journal of Neurology. Neurosumerv. and Psychiatry. 50. 714-719.
Hoffman, D.S. & Strick, P.L. (1993). Step-tracking movements of the wrist III. Influence of
changes in load on patterns of muscle activity. Journal of Neuroscience. 13. 5212-
5227.
JAVA Video Analysis [Computer software]. (1989). Corte Madera, CA: Jandel Scientific.
Jeannerod, M. (1988). Models for the programming of goal-directed movements. In: D.E.
Broadbent J.L. McGaugh, N.J. Mackintosh etal. (Eds.), The neural and behavioral
organization of goal-directed movements. Oxford Psychology Series# 15. (dp. 1-40).
Oxford: Clarendon Press.
Jeannerod, M„ Arbib, M.A., Rizzolati, G. & Sakata, H. (1995). Grasping objects: The cortical
mechanisms of visuomotor transformation. TINS. 18. 314-320.
Jenkins, I.H., Brooks, D.J., Nixon, P.D., Frackowiak, R.S.J. & Passingham, R.E. (1994). Motor
sequence learning: A study with positron emission tomography. Journal of
Neuroscience. 14. 3775-3790.
Jette, A.M. (1994). Physical disablement concepts for physical therapy research and practice.
Physical Therapy. 74. 380-386.
Johnson, P.B. (1992). Toward an understanding of the cerebral cortex and reaching
movements: A review of recent approaches. In: R. Caminiti, P.B. Johnson. & Y.
Bumod (Eds.), Control of Arm Movements in Space. Experimental Brain Research
Series #22. (dp. 199-261). Berlin: Springer-Veriag.
Jones, R.D., Donaldson, I.M. & Parkin, P.J. (1989). Impairment and recovery of ipsilateral
sensory-motor function following unilateral cerebral infarction. Brain. 112.113-132.
Kalaska, J.F. & Crammond, D.J. (1992). Cerebral cortical mechanisms of reaching
movements. Science. 255. 155-158.
Kami, A., Meyer, G., Jezzard, P., Adams, M.M., Turner, R. & Ungerfeider, LG . (1995).
Functional MRI evidence for adult motor cortex plasticity during motor skill learning.
Nature. 377. 1517-1523.
Kawashima, R„ Roland, P.E. & O'Sullivan, B. (1994). Activity in the human primary motor
cortex related to ipsilateral hand movement Brain Research. 663. 251-256.
Kawashima, R., Roland, P.E. & O'Sullivan, B.T. (1994). Reids 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.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
182
Kuboto, K. & Komatsu, H. (1985). Neuron activities of monkey preffontai cortex during the
learning of visual discrimination tasks with GO/NO-GO performances. Neuroscience
Research. 3,106-129.
Kunesch, E., Binkofski, F., Steinmetz, H„ & Freund, H-J. (1995). The pattern of motor deficits
in relation to the site of stroke lesions. European Neurology. 35: 20-26.
Kwakkel, G., Wagenaar, R.C., Kollen, B.J., & Lankhorst, G.J. (1996). Predicting disability in
stroke: A critical review of the literature. Aae and Aaeina. 25: 479-489.
Lee, T.D. & Carnahan, H. (1990). Bandwidth knowledge of results and motor learning: More
than just a relative frequency effect The Quarterly Journal of Experimental
Psychology. 42, 777-789.
Lee, T.D. & Magill, R.A. (1983). The locus of contextual interference in motor skill acquisition.
Journal of Experimental Psychology Learning. Memory, and Cognition. 9, 730-746.
MacKenzie, C.L. & Marteniuk, R.G. (1985). Motor skill: Feedback, knowledge, and structural
issues. Canadian Journal of Psychology. 39. 313-337.
Marteniuk, R.G. & Romanow (1983). Human movement organization and learning as revealed
by variability of movement use of kinematic information, and Fourier analysis. In: R.A.
Magill (Ed) Memory and Control of Action (pp.167-197). Amsterdam: North Holland
Publishing.
Mathiowetz, V. (1990). Grip and pinch strength measurements. In: LR. Armstrong (Ed)
Muscle Strength Testing (pp.163-1771. New York: Churchill Livingstone.
Mathiowetz, V., Volland, G„ Kashman, N. & Weber, K. (1985). Adult norms for the box and
block test of manual dexterity. American Journal of Occupational Therapy. 39. 386-
391.
Mitz, A.R., Godschalk, M. & Wise, S.P. (1991). Learning-dependent neuronal activity in the
premotor cortex: Activity during the acquisition of conditional motor associations.
Journal of Neuroscience. 11. 1855-1872.
Moore, S. & Marteniuk, R.G. (1986). Kinematic and electromyographic changes that occur as
a function of learning a time-constrained aiming task. Journal of Motor Behavior. 18.
397-426.
Moscovitch, M. (1992). A neuropsychological model of memory and consciousness. In: L.R.
Squire. & N. Butters (Eds.), Neuropsychology of Memory. 2nd Edition, (pp. 5-22). New
York: Guilford Press.
Mowbray, G.H. & Rhoades, M.V. (1959). On the reduction of choice reaction times with
practice. Quarterly Journal of Experimental Psychology. 11. 16-23.
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.
Mutter, S.A., Howard, J.H. & Howard, D.V. (1994). Serial pattern learning after head injury.
Journal of Clinical and Experimental Neuropsychology. 16. 271-288.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
183
Nagi, S.N. (1991). Disability concepts revisited: Implications for prevention. In: A.M. Pope &
A.R. Tarlov (Eds), Disability in America: Toward a National Agenda for Prevention
(pp.309-327). Washington, D.C.: National Academy Press.
Newell, KM . (1985). Coordination, control and skill. In: D. Goodman, R.B. Wilberg. & I.M.
Franks (Eds.), Differing Perspectives in Motor Learning. Memory, and Control, (pp.
295-317). North-Holland: Elsevier Science Publishers B.V.
Newell, KM . (1991). Motor skill acquisition. Annual Review of Psychology 42. 213-237.
Nissen, M.J. (1992). Procedural and declarative learning: Distinctions and interactions. In: LR.
Squire. & N. Butters (Eds.), Neuropsychology of Memory. 2nd Edition, (pp. 203-210).
New York: Guilford Press.
Nissen, M.J. & Bullemer, P. (1987). Attentional requirements of learning: Evidence from
performance measures. Cognitive Psychology. 1 9 .1-32.
Nudo, R.J., Wise, B.M., SiFuentes, F., & Miiliken, G.W. (1996). Neural substrates for the
effects of rehabilitation training on motor recovery after ischemic infarct Science. 272.
1791-1794.
Nudo, R.J., Miiliken, G.W., Jenkins, W.M., & Merzenich (1996). Use-dependent alterations of
movement representations in primary motor cortex of adult squirrel monkeys. Journal
of Neuroscience. 16. 785-807.
Olsen, T.S. (1990). Arm and leg paresis as outcome predictors in stroke rehabilitation. Stroke.
2 1 247-251.
Pascual-Leone, A., Grafman, J., Clark, K, Stewart M., Massaquoi, S., Lou, J.-S. & Halletl M.
(1993). Procedural learning in Parkinson's Disease and cerebellar degeneration.
Annals of Neurology. 34. 594-602.
Pascual-Leone, A., Grafman, J. & Hallet M. (1994). Modulation of cortical motor output maps
during development of implicit and explicit knowledge. Science. 263. 1287-1289.
Passingham, R.E. (1988). Premotor cortex and preparation for movement Experimental Brain
Research. 70. 590-596.
Pavlides, C., Miyashita, E. & Asanuma, H. (1993). Projection from the sensory to the motor
cortex is important in learning motor skill in the monkey. Journal of Neurophvsiologv.
ZQ, 733-741.
Pew, R.W. (1966). Acquisition of hierarchical control over the temporal organization of a skill.
Journal of Experimental Psychology. 71. 764-771.
Pfann, KD., Hoffman, D.S., Gottlieb, G .L, Strick, P.L & Corcos, D.M. (1998). Common
principles underlying the control of rapid, single degree-of-freedom movements at
different joints. Experimental Brain Research. 118. 35-51.
Platz, T„ Denzler, P., Kaden, B. & Mauritz, K.-H. (1994). Motor learning after recovery from
hemiparesis. Neuropsvcholooia. 3 2 .1209-1233.
Pohl, P S. & Winstein (1998a). Motor control in the less affected upper extremity after stroke
and the effects of practice. Unpublished manuscript
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
184
Pohl, P.S. & Winstein (1998b). Age-related effects on temporal strategies to speed motor
performance. Journal of Aging and Physical Activity. §, 45-61.
Pohl, P.S., Winstein, C.J. & Fisher, B.E. (1996). The locus of age-related movement slowing:
Sensory processing in continuous goal-directed aiming movements. Journal of
Gerontology: Psychological Science.
Pohl, P.S., Winstein, C.J. & Onla-or, S. (1997). Sensory-motor control in the ipsilesional upper
extremity after stroke. NeuroRehabilitation. 2, 57-69.
Pope, A., & Tartov, A. (1991). Disability in America: Toward a National Agenda for Prevention.
Washington, D.C.: National Academy Press.
Rao, S.M., Binder, J.R., Bandetti, P.A., Hammeke, T.A., Yetkin, F.Z., Jesmanowicz, A., Lisk,
L.M., Moms, G.L., Mueller, W.M., Estkowski, L.D., Wond, E.C., Haughton, V.M. &
Hyde, J.S. (1993). Functional magnetic resonance imaging of complex human
movements. Neurology. 43. 2311-2318.
Remy, P., Zilbovicius, M.D., Leroy-Willig, A., Syrota, A. & Samson, Y. (1994). Movement- and
task-related activations of motor cortical areas: A positron emisssion tomographic
study. Annals of Neurology. 3 6 .19-26.
Roth, E.J., Heinemann, A.W., Lovell, L.L., Harvey, R.L., McGuire, J.R., & Diaz, S. (1998).
Impairment and disability: Their relation during stroke rehabilitation. Archives of
Physical Medicine and Rehabilitation. 79. 329-335.
Rothstein, J.M. (1994). Disability and our identity. Physical Therapy. 74. 375-378.
Rowland, R.P. (1991). Clinical syndromes of the spinal cord and brainstem. In: E.R. Kandel,
J.H. Schwartz, & T.M.Jessell (Eds.), Principles of Neural Science 3rd Ed. (pp. 711-
731). New York: Elsevier.
Roy, E.A., Kalbfleisch, L. & Elliott, D. (1994). Kinematic analyses of manual asymmetries in
visual aiming movements. Brain and Cognition. 24, 289-295.
Sadato, N., Campbell, G., Ibenez, V., Deiber, M-P, & Hallett, M. (1996). Complexity affects
regional cerebral blood flow chaneg during sequential finger movements. Journal of
Neruoscience. 16. 2693-2700.
Saint-Cyr, J.A. & Taylor, A.E. (1992). The mobilization of procedural learning: The "key
signature" of the basal ganglia. In: LR. Squire. & N. Butters (Eds.), Neuropsychology
of Memory. 2nd Edition, ( p p . 188-202). New York: Guilford Press.
Saint-Cyr, J.A., Taylor, A.E. & Lang, A.E. (1988). Procedural learning and neostriatal
dysfunction in man. Brain. 111. 941-959.
Salmoni, A.W., Schmidt R.A. & Walter, C.B. (1984). Knowledge of results and motor learning:
A review and critical reappraisal. Psychological Bulletin. 95. 355-386.
Sanes, J.N & Donoghue, J.P. (1992). Organization and adaptability of muscle representations
in primary motor cortex. In: R. Caminitii, P.B. Johnson, & Bumod, Y. (Eds.),
Experimental Brain Research Series 22. (pp. 103-127). Beriin:Springer-Verlag.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
185
Sanes, J.N, Donoghue, J.P., Thangaraj, V., Edelman, R.R., & Warach, S. (1995). Shared
neural substrates controlling hand movements in human motor cortex. Science. 286.
1775-1777.
Sasaki, K. & Gemba, H. (1981). Changes of movement field potentials in the cerebral cortex
during learning processes of visually initiated hand movements in the monkey.
Neuroscience Letters. 27. 125-130.
Sasaki, K . & Gemba, H. (1982). Development and change of corticai field potentials during
learning processes of visually initiated hand movements in the monkey. Experimental
Brain Research. 48. 429-437.
Schacter, D.L. (1987). Implicit memory: History and current status. Journal of Experimental
Psychology Learning. Memory, and Cognition. 13. 501-518.
Schlaug, G., Knorr, U. & Seitz, R.J. (1994). Inter-subject variability of cerebral activations in
acquiring a motor skill: A study with positron emission tomography. Experimental Brain
Research. 98. 523-534.
Schmidt R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review.
82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research
Quarterly for Exercise and Sport 56. 188-200.
Schmidt R.A. (1987). The acquisition of skill: Some modifications to the perception-action
relationship through practice. In: H. Hever & A.F. Sanders (Eds.), Perspectives on
Perception and Action, (pp. 77-103). Hillsdale, NJ: Lawrence-Eribaum Associates.
Schmidt R.A. f1988V Motor control and learning: A behavioral emphasis. 2nd. Champaign, IL:
Human Kinetic Publishers Inc.
Schmidt R.A. (1991a). Frequent augmented feedback can degrade learning: Evidence and
interpretations. In: J. Requin. & G.E. Stelmach (Eds.), Tutorials in Motor
Neuroscience, (pp. 59-75). Netherlands: Kluwer Academic Publishers.
Schmidt R.A. (1991b). Motor learning and performance: From principles to practice.
Champaign, IL Human Kinetics.
Schmidt R.A. (1991c). Motor learning principles for physical therapy. In: M.S. Lister (Ed.),
Contemporary management of motor control problems: Proceedings of the IISTEP
conference, (pp. 49-63). Alexandria, VA: Foundation for Physical Therapy.
Schmidt R.A. & Bjork, R.A. (1992). New conceptualizations of practice: Common principles in
three paradigms suggest new concepts for training. Psychological Science. 3, 207-
217.
Schmidt R.A. & Young, D.E. (1991). Methodology for motor learning: A paradigm for kinematic
feedback. Journal of Motor Behavior. 2 3 .13-24.
Seger, C.A. (1994a). Implicit learning. Psychological Bulletin. 115. 163-196.
Seger, C.A. (1994b). Criteria for implicit learning: Deemphasize concious access, emphasize
amnesia. Behavioral and Brain Sciences. 17, 421-422.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
186
Seitz, R.J., Roland, P.E., Bohm, C., Greitz, T. & Stone-Elander, S. (1990). Motor learning in
man: A positron emission tomographic study. Neuroreport 1,17-20.
Sekiya, H„ Magill, R.A., Sidaway, B . & Anderson, D.l. (1994). The contextual interference
effect for skill variations from the same and different generalized motor programs.
Research Quarterly for Exercise and Sport §5, 330-338.
Shanks, D.R. & St John, M.F. (1994). Characteristics of dissociable human learning systems.
Behavioral and Brain Sciences. 17. 367-447.
Shea, J.B. & Morgan, R.L. (1979). Contextual interference effects on the acquisition, retention,
and transfer of motor skill. Journal of Experimental Psychology: Human Learning and
Memory. 5, 179-187.
Shibasaki, H„ Sadato, N., Lyshkow, H., Yonekura, Y„ Honda, M., Nagamine, T„ Suwazono,
S., Magata, Y., Ikeda, A., Miyazaki, M., Fukuyama, H., Asato, R. & Konishi, J. (1993).
Both primary motor cortex and supplementary motor area play an important role in
complex finger movement Brain. 116. 1387-1398.
Smutok, M .A, Grafman, J., Salazar, AM., Sweeney, J.K., Jonas, B.S., & DiRocco, P.J.
(1989). Effects of unilateral brain damage on contralateral and ipsilateral upper
extremity function in hemiplegia. Physical Therapy. 69. 195-203.
Snoddy, G.S. (1926). Learning and stability. Journal of Applied Psychology. 1 0 .1-36.
Sonoda, S., Chino, N., Domen, K., & Saitoh, E. (1997). Changes in impairment and disability
from the third to the sixth month after stroke and its relationship evaluated by an
artificial neural network. American Journal of Physical Medicine and Rehabilitation. 76.
395-400.
Squire, L.R. (1986). Mechanisms of memory. Science. 232. 1612-1619.
Squire, LR. (1987). Memory: Neural organization and behavior. In: V.B. Mountcastle, F. Plum.
& S.R. Geiger (Eds.), Handbook of Phvsioloav: The Nervous System. Higher
Functions of the Brain. (Vol. V, pp. 295-371). Bethesda: American Physiology Society.
Squire, L.R. (1992). Declarative and nondeclarative memory: Multiple brain systems
supporting learning and memory. Journal of Cognitive Neuroscience. 4, 232-243.
Sullivan, R.J. (1997). GMP (Version 3.2) [Computer software]. Franklin, Wl: Author.
Swanson, L.R. & Lee, T.D. (1992). Effects of aging and schedules of knowledge of results on
motor learning. Journal of Gerontology: Psychological Sciences. 47, P406-P411.
Tanji, J., Okano, K. & Sato, K. (1988). Neuronal activity in cortical motor areas related to
ipsilateral, contralateral, and bilateral digit movements of the monkey. Journal of
Neurophvsioloav. 60. 325-343.
Tanji, J., Taniguchi, K. & Saga, T. (1980). Supplementary motor area: neuronal response to
motor instructions. Journal of Neurophvsioloav. 43. 60-68.
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.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
187
Tolman, E.C. (1932). Purposive behavior of animals and men. New York: Century.
Truwit, C.L. & Lempert, T.E. (1994). High Resolution Atlas of Cranial Neuroanatomv.
Baltimore: Williams & Wilkins.
Ungerteider, L.G. (1995). Functional brain imaging studies of cortical mechanisms for memory.
Science. 270. 669-775.
van Donkelaar, P. & Franks, I.M. (1991). The effects of changing movement velocity and
complexity on response preparation: Evidence from latency, kinematic, and EMG
measures. Experimental Brain Research. 83. 618-632.
Wade, D.T., Langton-Hewer, R., Wood, V.A., Skilbeck, C.E., & Ismail, H.M. (1983). The
hemiplegic arm after stroke: Measurement and recovery. Journal of Neurology.
Neurosurgery, and Psychiatry. 4§. 521-524.
Ware, J.E. & Sherboume, C.D. (1992). The MOS 36-item short-form health survey (SF-36):
Conceptual framework and item selection. Medical Care. 30. 473-483.
Wechsler, D.A. (1987). Wechsler memory scale-revised. Psychological Corporation. San
Antonio.
Weinrich, M. & Wise, S.P. (1982). The premotor cortex of the monkey. Journal of
Neuroscience. 2, 1329-1345.
Weiss, P. (1941). Self-differentiation of the basic patterns of coordination. Comparative
Psychology Monographs. 17, Whole No. 4.
Welford, A.T. (1985). Practice effects in relation to age: A review and a theory. Developmental
Neuropsychology. ±, 173-190.
Wickens, J., Hyland, B., & Anson, G. (1994). Cortical cell assemblies: A possible mechanism
for motor programs. Journal of Motor Behavior. 26. 66-82.
Willingham, D.B., Nissen, M.J. & Bullemer, P. (1989). On the development of procedural
knowledge. Journal of Experimental Psychology Learning. Memory, and Cognition. 15.
1047-1060.
Winstein, C.J. (1988). Relative freouencv of information feedback in motor performance and
learning. Unpublished doctoral dissertation, University of California, Los Angeles.
Winstein, C.J., Grafton, S.T., & Pohl, P.S. (1997). Motor task difficulty and brain activity:
Investigation of goal-directed reciprocal aiming using positron emission tomography.
Journal of Neurophvsioloov. 77. 1581-1594.
Winstein, C.J., Merians, A., & Sullivan, K.J. (Accepted with revisions). Motor learning after
unilateral brain damage. Neuropsvcholooia.
Winstein, C.J. & Pohl, P.S. (1995). Effects of unilateral brain damage on the control of goal-
directed hand movement Experimental Brain Research. 105. 163-174.
Winstein, C.J., Pohl, P.S. & Lewthwaite, R. (1994). Effects of physical guidance and
knowledge of results on motor learning: Support for the guidance hypothesis.
Research Quarterly for Exercise and Sport. 65. 316-323.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
188
Winstein, C.J. & Schmidt R A (1990). Reduced frequency of knowledge of results enhances
motor skill learning. Journal of Experimental Psychology Learning. Memory, and
CQaOifon, 1£, 677-691.
Wise, S.P. (1985). The primate premotor cortex: Past, present and preparatory. Annual
Review of Neuroscience. 3,1-19.
Wise, S.P. & Mauritz, K.-H. (1985). Set-related neuronal activity in the premotor cortex of
rhesus monkeys: Effects of changes in motor set Proceedings of the Roval Society of
London Series B: Biological Sciences. 223. 331-354.
Woodworth, R.S. (1899). The accuracy of voluntary movements. Psychological Review
Monograph Supplement. 3,1-114.
World Health Organization (1980). International Classification of Impairments. Disabilities and
Handicaps. Geneva, Switzerland.
Wulf, G., Lee, T.D. & Schmidt R A. (1994). Reducing knowledge of results about relative
versus absolute timing: Differential effects of learning. Journal of Motor Behavior. 26.
362-369.
Wulf, G. & Schmidt R A (1994). Feedback-induced variability and the learning of generalized
motor programs. Journal of Motor Behavior. 26. 348-361.
Wulf, G„ Schmidt R A & Deubel, H. (1993). Reduced feedback frequency enhances
generalized motor program learning but not parameterization learning. Journal of
Experimental Psychology Learning. Memory, and Cognition. 1 9 .1134-1150.
Wyller, T.B. (1998). Prevalence of stroke and stroke-related disability. Stroke. 29. 866-867.
Wyller, T.B., Bautz-Holter, E., & Holmen, J. (1994). Prevalence of stroke and stroke-related
disability in North Trondelag county, Norway. Cerebrovascular Disease. 4,421-427.
Yamashita, H. (1993). Perceptual-motor learning in amnesic patients with medial temporal
lobe lesions. Perceptual and Motor Skills. 7 7 .1311-1314.
Young, D.E. & Schmidt, R.A. (1990). Unit of motor behavior Modifications with practice and
feedback. In: M. Jeannerod (Ed.), Attention and Performance XIII. (pp. 763-795).
Hillsdale, NJ: Lawrence Ertbaum Associates Publishers.
Young, P.A. & Young, P.H. (1997). Basic Clinical Neuroanatomv. Baltimore:Wiiliams & Wilkins.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject R 1: Subject R2:
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject R 3: Subject R4:
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject R 5: Subject R6:
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject R 7: Subject R8:
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject R 9 : Subject R10:
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject L1: Subject L2:
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
195
S3
JO
(0
>
( 0
c
(0
o
C O
u
< D
S '
3
C O
c o
— I
O
< D
5*
3
C O
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject L 5 : Subject L6:
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject L 7 : Subject L8:
197
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subject L 9 : Subject L10:
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IMAGE EVALUATION
TE S T TA RG ET ( Q A - 3 )
1 5 0 m m
IIVA4GE. Inc
16 53 E ast M ain S treet
R o ch e ste r, NY 14609 USA
P h o n e : 716/482-0300
F a x: 716/288-5989
C -993. Applied Image. Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zed) Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9919109
C opyright 1999 by
S u lliv a n , K atherine Josephine
All rights reserved.
UMI Microform 9919109
Copyright 1999, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The role of the vasti in patellar kinematics and patellofemoral pain

PDF

Reactive synaptogenesis in the striatum of the adult rat: the role of dopamine

PDF

Satellite cell proliferation: two models and two anabolic stimuli

PDF

Experimental and computational analysis of microsatellite repeat instability in mismatch-repair-deficient mice

PDF

Functional brain correlates for premovement planning and compensatory adjustments in rapid aimed movement

PDF

This other darkness: exotic bodies and the gaze of romanticism

PDF

The role of host proteins in retroviral cDNA integration

PDF

Motor skill learning in individuals with Parkinson's disease: Consideration of cognitive and motor demands

PDF

Probabilistic analysis of power dissipation in VLSI systems

PDF

The effect of helmet liner density upon acceleration and local contact forces during bicycle helmet impacts

PDF

The role of light organ symbionts in signaling early morphological and biochemical events in the sepiolid squid Euprymna scolopes

PDF

The ecology and behavior of the California moray eel Gymnothorax mordax (Ayres, 1859) with descriptions of its larva and the leptocephali of some other east Pacific muraenidae

PDF

Studies on cholesterol metabolism in the rat

PDF

Alternate models of women's health care policy in the United States

PDF

Factors contributing to patellofemoral joint stress: a comparison of persons with and without patellofemoral pain

PDF

Connections of the dorsomedial hypothalamic nucleus in the rat

PDF

Phylogenetics of the cutlassfishes (Scombroidei: Trichiuridae)

PDF

Studies of Drosophila sex-lethal protein interactions and cloning of a novel interactor, SIN

PDF

Elevated inflammation in late life: predictors and outcomes

PDF

Spatial distribution of neuroendocrine motoneuron pools in the hypothalmic paraventricular nucleus

Asset Metadata

Creator Sullivan, Katherine Josephine (author)

Core Title The role of the sensorimotor cortical system in skill acquisition and motor learning: a behavioral study

School Graduate School

Degree Doctor of Philosophy

Degree Program Biokinesiology

Degree Conferral Date 1998-08

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

Tag biology, neuroscience,health sciences, rehabilitation and therapy,OAI-PMH Harvest,psychology, behavioral

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-397233

Unique identifier UC11350283

Identifier 9919109.pdf (filename),usctheses-c17-397233 (legacy record id)

Legacy Identifier 9919109.pdf

Dmrecord 397233

Document Type Dissertation

Rights Sullivan, Katherine Josephine

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