Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Functional brain correlates for premovement planning and compensatory adjustments in rapid aimed movement
(USC Thesis Other)
Functional brain correlates for premovement planning and compensatory adjustments in rapid aimed movement
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
FUNCTIONAL BRAIN CORRELATES FOR PREMOVEMENT PLANNING AND COMPENSATORY ADJUSTMENTS IN RAPID AIMED MOVEMENT Copyright 2000 by Beth Ellen Fisher 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 2000 Beth Ellen Fisher Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA 1HIGRADUATE SCHOOL UMVER9TY PARK L O B ANOBUS, CALVORMA 90009 77ns dissertation, written by ...... .§ g J{ l.H .l S l.F .i§ !3 0C ....................................... under the directum of hsr.— Dissertation Committee, and approved by alt its m em bers► has b e e n presented to and ac ce pte d by The Graduate School, in partial fulfillment of re quirements for die degree of DOCT OR OF PHI LOSOPHY D im a f Gnd»mt r St udi es DISSERTATION COMMITTEE A Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION ii To my parents, Marcy and Hal Stein, whose love and support has fueled every challenge undertaken, every goal pursued, every dream followed and To my beloved Richard whose everlasting presence in my heart has directed me on the journey. I will love you forever Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iii ACKNOWLEDGMENTS An arrow shot into the air is a symbol of unlimited aspiration. The archer says this to the sky with his arrow: “ However big and wide I can dream is however big and wide I can live.’ — author unknown— I am blessed in my life to have had the most remarkable people encourage me to always live and dream as ‘big’ and as ‘wide* as I can: Dr. Carolee Winstein, who has and always will be my teacher, mentor, and friend. Her relentless pursuit of knowledge is inspiring. I am grateful that I have been a witness to and a recipient of her commitment to excellence. Her genius is an ability to integrate information from countless sources and fields. Her boundless creativity enables her to take that integrated package to new levels of examination. Dr. Jim Gordon, through his patient teaching, he helped me’ to tell the story’ of my dissertation. I am continually amazed by his ability to make the most complex scientific concepts meaningful to both scientist and clinician. Dr. Sandy Howell encouraged and supported my graduate studies in every possible way. She made it possible for me to teach and complete my dissertation simultaneously. My dissertation committee, Dr. Stan Azen, Dr. Lucinda Baker, Dr. Nina Bradley, Dr. Jill McNitt- Gray who have provided insight and counsel throughout my studies. Kathy Sullivan: I am fortunate to have more than one person that I can call mentor. There is only one person that I call my hero. Kathy continually helps me to be the best I can be. Separately, we make an impact, as a team, we can change the world. My first professional mentors, Dr. Helen Hislop and Jacqueline Montgomery. They are both visionaries whose passion about physical therapy has always been contagious. Pan (Sompom) Onla-Or and Kathleen Ganley: they have taken turns holding onto the ’rope’ as I have 'skirted the edge of my abilities.' The countless times that they provided clarity and perspective helped me to continually move forward. Neither one of them know that along with being my best friends, they have been awe-inspiring teachers. My patients that truly have taught me the meaning of courage and have been my greatest teachers of ‘life.’ My colleagues, teachers, and friends at Rancho Los Amigos Medical Center: Karen Parker, Cheryl Resnik, Dee Thompson, Mary Ruth Velicki, Walt Weiss, Arlene Yang, and Cindy Zablotny. What a remarkable learning journey we have traveled together; what an amazing group of people I am blessed to call friends. Chris Powers, Jody Cormack, Steve Reischl, Robert Landel, and Mary Hudson-McKinney...from the moment I called these people colleagues, I have not stopped learning and being challenged by them. From the moment I called them friends, I have not been without their love and support. Posy and Michael McKinney, Libby and Mark Ward, and Peter Stein...my brothers and sisters...we have cried together in the worst of times and rejoiced in the best of times...they’re everything one dreams a family to be. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iv Brad Baker and Jennifer New...'non physical therapist’ friends who have been right there on the bench with me throughout my studies. They both continually fueled my efforts with laughter. I am thankful to all my physical therapy students as well as those seasoned professionals that I have had the pleasure to teach. They have encouraged me to put forward my ideas, consolidate complex scientific information and consider the possibilities for clinical application Florence and James Fisher...since my marriage to their son, our tragic loss of him, and to the present day, have loved and supported me like a daughter. I am thankful to call them mom and dad. While he will never appreciate the acknowledgment, I must thank Sporty George, my dog. This noble creature was the reason I got up and got through the day for a very long time. Lisette and Norman Ackerberg entered my heart as ’family’ almost from the moment that they entered my life. I have never experienced the magnitude of love, support and encouragement that these extraordinary people offer to those fortunate to be a part of their lives. They have helped me to understand that the challenges of my work and of my life are exciting opportunities for growth and evolution. What I have read in books has given me information, what I have learned from Lisette has given me wisdom. Last but not least, Roger Phillips...he has only known me in the context of a PhD student but he has been committed to traveling the path with me regardless of how rocky it has gotten. He has, more than anyone, helped me to be excited about the 'journey.' He has more than anyone given me the energy to keep going. Two of the most amazing gifts I have ever received were given to me from Roger: knowledge that I can have love in my life again, and his son....Max. I thank Roger every day and will for the rest of my life for helping me to see again, the beauty in this world. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V TABLE OF CONTENTS Dedication jj Acknowledgments iii List of Tables vi List of Figures vii Abstract xvi Chapter 1 INTRODUCTION 1 Chapter 2 EXPERIMENT 1: DEFICITS IN COMPENSATORY 21 TRAJECTORY ADJUSTMENTS AFTER UNILATERAL SENSORIMOTOR STROKE Chapter 3 EXPERIMENT 2: INCREASED TASK COMPLEXITY 58 OF AIMED ARM MOVEMENTS AFFECTS PLANNING AND UPDATING OF RESPONSE PARAMETERS FOR INDIVIDUALS WITH UNILATERAL SENSORIMOTOR DAMAGE Chapter 4 EXPERIMENT 3: THE ROLE OF THE CEREBELLUM 120 IN THE PLANNING OF AIMING RESPONSE PARAMETERS Chapter 5 CONCLUSIONS 163 BIBLIOGRAPHY 176 APPENDIX 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vi TABLES Chapter 2: Table 1. Lesion Characteristics 30 Table 2. AE for Predictable and Unpredictable Conditions 42 Table 3. Average MT and Rise Time 44 Chapter 3: Table 1. Subject and Lesion Characteristics 67 Table 2. Impairment and Functional Limitation Measures of Stroke Subjects 71 Table 3. Individual Stroke Subject Data 94 Table 4. Velocity Rise Times for Stroke Subjects by Target 118 Chapter 4: Table 1. Subject and Lesion Characteristics 131 Table 2. Impairment and Functional Limitation Measures of Cerebellar Subjects 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii FIGURES Chapter 2: Figure 1. TRP paradigm demonstrating presentation of movement cue between 32 the third and fourth tone. In this example, the stimulus is given £ 300 ms before the 4th tone and is calling for movement to target three. The S-R interval is the time period between stimulus presentation and the beginning of movement. The S-T interval is the time period between stimulus presentation and the 4th tone. Figure 2. a) Example of an acceptable single peaked velocity profile for a 45° 35 flexion movement, b) 45° flexion movement with 'premovement' extension movement related to movement preparation. Onset of movement was defined as a change in velocity from zero in the appropriate direction of movement or beginning negative velocity. Offset was defined as return to zero velocity, c) 45° flexion movement in which the subject initiated a second response prior to zero velocity. This ‘double response" trial was accepted for analysis as velocity returned to less than 1 0 % of peak velocity prior to the initiation of the second response, d) 45° flexion movement in which the subject initiated a second response prior to reaching zero velocity. This ‘double response” trial was rejected as velocity did not return to less than 1 0 % of peak velocity prior to the initiation of the second response. Figure 3. Representative data from a single trial (45° flexion) showing key 36 kinematic and temporal variables: a = Ony - Offy = Displacement; average velocity (°/sec) from the onset of movement b until peak velocity c; position rise time = c - b (ms) d = Off, - On, = Movement time; e = S-R interval; f = S-T interval. Figure 4. Illustrates modified statistical model of the determinants of final 38 position that are tested. According to the model, the target can influence the final position through two paths. In the planned trajectory path, the target influences the final position (Y) through its effect on the average to peak velocity (dY/dt). The squared correlation coefficient (r^ ,) represents the proportion of the variance in final position that can be explained by variation in average to peak velocity (dY/dt). Therefore, r*y, reflects the degree to which final position is explained by premovement planning. The second path represents the corrective influence of the target on final position through the implementation of compensatory adjustments to the trajectory. The additional variance accounted for by the target (through the compensatory adjustment pathway) is equal to the difference between that proportion of the variance accounted for by the combination of average velocity from onset to peak and target and that accounted for by the velocity measure alone (R2 ™ - r2^). Figure 5. Stack bar plot of percent variance in displacement for predictable trials 43 by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left and the stroke group is to the right Data are averaged across 6 subjects/group and 4 targets. SEM error bars are shown for each component of the stack bar plot. Figure 6 . a) Stack bar plot of percent variance in displacement for unpredictable 46 trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. viii left and the stroke group is to the right. Data are averaged across 6 subjects/group and 4 targets. SEM error bars are shown for each component of the stack bar plot, b) Bar graph representing percent variance in displacement due to compensatory adjustments. Data were averaged across groups, and response direction accuracy (correct, wrong). The shortest preparation interval (0-99 ms) was significantly different from the two longest intervals (200*299 and 300*400 ms); p = .001 and .006 respectively, c) percent variance in final position explained by compensatory adjustments for each group as a function of preparation interval. Control group = squares; stroke group = diamond. Group x preparation interval interaction, (p = .13). Note the relatively high SEM for the control group, d) The total variance (i.e., that due to both planning and compensatory adjustments) in the unpredictable condition is compared to the percent variance in final position due to planning alone for both stroke (diamond) and control (square) groups. Figure 7. The left four stack bars are the percent variance due to planning and 48 compensatory adjustments for the control group across preparation intervals. Depicted is the composite variance in final position consisting of that due to planning and compensatory adjustments for correct direction responses to flexion targets. The middle four stack bars represent data from one control subject (subject #2). Data from a second control subject (subject #5) is presented in the right four stack bars. Note a greater increase in proportion of variance due to compensatory adjustments across the earliest three preparation intervals for control subject #5 compared to the group. Figure 8 . Comparison of composite variance due to planning and compensatory 50 adjustments between two unilateral stroke subjects. The left four stack bars represent the stroke group data. Subject #10 had the smallest SM lesion = 2.2 cm3 ; Subject #11 had the second largest SM lesion of the stroke group = 64.2 cm3 . Note greater percentage of variance explained by compensatory adjustments across preparation intervals for subject # 1 0 compared with subject #11. Chapter 3: Figure 1. Lesion location for each stroke-group subject from magnetic 69 resonance imaging (MRI) or computer tomography (CT) images. Each section shows the lesion-site of greatest extent using the functional atlas of Domasio and Domasio (1989). Figure 2. Illustrates the six target locations with the lever positioned at the 5t h 73 target. A replication of the right hand display was presented on the computer monitor for subjects using their right arm to perform the task; the left hand display was viewed on the monitor for subjects using their left arm. Numbers were not actually displayed, but are included for clarification of the text. Figure 3. a)The written feedback displayed when movement was initiated at 78 various times throughout the trial sequence. This example shows a total timing window of 75 ms before and 75 ms after the 50 ms tone (total timing window = 200 ms), b) The modified timed-response paradigm (TRP). Movement initiation was to be synchronized with the last in a series of four tones (1-4). The movement cue was presented at variable times between 400 and 0 ms before the fourth tone. In this example, the stimulus is given z 300 ms before the 4th tone and is calling for movement to target four. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix stimulus-tone (S-T) interval, the stimulus-response (S-R) interval, and the timing error [(S-R interval) - (S-T interval)] are shown. Figure 4. Representative data from a single trial (45s flexion) showing key kinematic and temporal variables: a = Ony • Offy = Displacement; average 83 velocity (°/sec) from the onset of movement b until peak velocity c; position rise time = c - b (ms) d = Off, - On, = Movement time; e = S-R interval; f= S-T interval. Figure 5. a) Bar graph of mean AE (iSEM ) for responses within each S-R interval for the unpredictable (crossed bars) and predictable (striped bars) 8 8 conditions. Data are averaged across 12 subjects and 6 targets, b) Symbols represent group means for AE (±SEM) to each target amplitude in the predictable condition. The control group is represented by the triangle and the stroke group by the squares. Data are averaged across S-R interval 6 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0001). c) Stack bar plot of percent variance in displacement for predictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left and the stroke group is to the right. Data are averaged across 6 subjects/group, 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. Figure 6 . a) Bar graph of group mean AE (iSEM ) within each S-R interval for 89 unpredictable correct direction responses. Data are averaged across 6 subjects/group and 6 target amplitudes. Control group = striped bars; stroke group = dark bars, b) Symbols represent group means for AE (iSEM ) to each target amplitude for correct direction, unpredictable responses. The control group is represented by the triangle and the stroke group by the squares. Data are averaged across S-R interval 6 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0001). Figure 7. Scatter plot of constant error (CE) across S-R intervals for a full 9 1 complement of correct direction, unpredictable condition responses. Subject 8 (a), and subject 1 1 (b) are stroke subjects and subject 6 (c) is a control subject. Regression lines are run through each target amplitude. A thicker horizontal line at 0s represents the target. For S8 , the default position was the short target extent. For S11 and S6 the default position was close to the middle target amplitude. Figure 8 . a) Stack bar plot of percent variance in displacement for predictable 95 trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left, the high- adjustment stroke group is in the middle and the low-adjustment stroke group is to the right Data are averaged across 6 subjects for the control group and 3 subjects per stroke sub-group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot b) Bar graph of correct -direction, unpredictable-condition responses representing percent variance in displacement due to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X compensatory adjustments over S-R interval. Data were averaged across 9 subjects (6 control and 3 high-adjustment stroke subjects). The shortest preparation interval (0-99 ms) was significantly different than the middle two preparation intervals (i.e., 100-199 ms, and 200-299 ms), P = .001and .009, respectively, c) Percent variance in final position explained by compensatory adjustments for each group as a function of preparation interval. Control group = dark bar; high-adjustment stroke group = open bar, and low-adjustment stroke group = striped bar. There was no group x preparation interval interaction for either the control and high-adjustment stroke group (P = .36) or the control and low adjustment stroke group (P = .61). Figure 9. Individual stroke subject scatter of percent variance in displacement 96 due to compensatory adjustments as a function of AE. Each stroke subject is represented by a different symbol. Data were correct direction responses and averaged across 6 targets and 4 S-R intervals. Note separation of stroke subjects into 2 groups with subjects 9,10, and 1 1 demonstrating lower AE and higher percent variance than subjects 12, 7, and 8 . Figure 10. Ensemble averages (±SEM) of the displacement, velocity, biceps, 98 and triceps profiles for (n=18) responses to the 15° extension target in the shortest preparation interval. Like trials (same S-R interval and target) over the 15 unpredictable trial blocks were averaged. Averages were calculated over 1000 ms. For the displacement profile, the 15s target is shown (dot-dash line) in order to show the averaged displacement in relation to the actual target. Mean displacement, velocity, biceps and triceps are represented by the thin black lines in between two thicker lines which represents ± standard error or the mean (SEM). Note: Low SEM related to responses demonstrates the consistency of the responses. Figure 11. Ensemble averages of movement velocity, biceps, and triceps for a 99 representative control subject (S2). Left panel are responses to the 30° extension (default) target; right panel are responses to the 45° extension target. The first, fourth, and seventh row of signals for both target amplitudes are averaged velocity, biceps and triceps profiles for the longest S-R interval (300- 400 ms) in the unpredictable condition. The second, fifth, and eighth row of signals are averaged velocity, biceps and triceps profiles for a middle S-R interval (100-199 ms) in the unpredictable condition. The third, sixth and ninth row of signals are averaged velocity, biceps and triceps profiles from the shortest preparation interval (0-99 ms) in the unpredictable condition. The number of averaged responses for the 30° target were 29,19, and 12 from most to least preparation time. For the 45° target, response number was 30,17, and 1 1 respectively. Offset of movement, defined as return to zero velocity is represented as the range of offsets for the three preparation intervals. Note: averaged AE for the three preparation intervals for both target amplitudes. Arrows shown for 45° target, triceps EMG represent the only two preparation conditions in which the second agonist burst begins before the first agonist burst returns to baseline. Figure 12. Ensemble averages of movement velocity, biceps, and triceps for a 100 representative low-adjustment stroke subject (S8 ). Left panel are responses to the 15° extension target (default); right panel are responses to the 45° extension target. The first, fourth, and seventh row of signals for both target amplitudes are averaged velocity, biceps and triceps profiles for the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xi longest S-R interval (300-400 ms) in the unpredictable condition. The second, fifth, and eighth row of signals are averaged velocity, biceps and biceps profiles for a middle S-R interval (100-199 ms) in the unpredictable condition. The third, sixth and ninth row of signals are averaged velocity, biceps and triceps profiles from the shortest preparation interval (0-99 ms) in the unpredictable condition. The number of averaged responses for the 15° target were 27,16, and 18 from most to least preparation time. For the 45° target, response number was 25,20, and 17 respectively. Figure 13. Ensemble averages of velocity, biceps, and biceps for a 101 representative high-adjustment stroke subject (S9). Left panel are responses to the 30° (default) flexion target; right panel are responses to the 45° flexion target. The row designation is as described in Figure 12. For both target amplitudes, responses for three S-R intervals are shown. The number of averaged responses for the 30° target were 23,21, and 19 from most to least preparation time. For the 45° target, response number was 22,17, and 16 respectively. In the longest preparation interval, the onset of the antagonist is delayed relative to the agonist burst (arrow), similar to what was seen to the default. Figure 14. Symbols represent group means for AE (±SEM) to each target 104 amplitude for wrong direction, unpredictable responses. The control group is represented by the triangle and the stroke group by the squares. Data are averaged across S-R interval 6 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0001). Figure 15. a) Bar graph of mean AE (±SEM) to the virtual target for wrong 105 direction responses within each S-R interval. Data are averaged across 6 stroke subjects for each target amplitude (short, medium, and long). Wrong direction responses did not converge on the correct target extent as preparation time increased. Across preparation intervals, wrong direction responses were positioned near the short target extent (-20°). b) Bar graph of mean AE (±SEM) to the virtual target for wrong direction responses within each S-R interval. Data are averaged across 6 control subjects for each target amplitude (short, medium, and long). The control group demonstrated a significant decrease in AE to the virtual long-amplitude target between the two middle preparation intervals (i.e, 100-199 and 200-299) and between the 200-299 ms interval and the longest preparation interval (300-400 ms). Note that no data are presented for the short amplitude target in the longest preparation interval. Only 2 out of 6 control subjects had wrong direction responses for this target/S-R interval. Figure 16. a) Frequency of direction errors as a percent of the total responses 106 within each S-R interval for the stroke and control groups in the unpredictable condition, b) Stack bar plot of percent variance in displacement for unpredictable, wrong direction responses by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left, the high-adjustment stroke group is in the middle and the low-adjustment stroke group is to the right. Data are averaged across 6 subjects for the control group and 3 subjects per stroke sub-group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot, c) Individual Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xii stroke subject scatter of percent variance in displacement due to compensatory adjustments as a function of AE. Each stroke subject is represented by a different symbol. Data were wrong direction responses and averaged across 6 (virtual) targets and 4 S-R intervals. Chapter 4: Figure 1. A) MRI or CT horizontal axial sections from the rostral to the caudal 129 cerebellum. The sagittal view indicates the section’s level. The horizontal levels progress from a-f (rostral to caudal cerebellum). Taken from: Amarenco, P. (1991). The spectrum of cerebellar infarctions. Neurology, 41:973-979. B) Lesion location for each cerebellar-group subject from magnetic resonance imaging (MRI) or computer tomography (CT) images. Each section shows the lesion-site of greatest extent using one of 6 horizontal axial sections taken from the rostral to the caudal cerebellum. The letter to the left of each lesion represents the horizontal section of Amarenco, 1991. Figure 2. A) Bar graph of group mean AE (±SEM) within each S-R interval for 140 predictable condition. Data are averaged across 5 subjects/group and 6 target amplitudes. Control group = striped bars; cerebellar group = dark bars. B) Symbols represent group means for AE (±SEM) to each target amplitude for predictable condition responses. The cerebellar group is represented by the triangle and the control group by the squares. Data are averaged across S-R interval 5 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0 0 0 1 ). C) Stack bar plot of percent variance in displacement for predictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The cerebellar group is to the left, the control group is to the right. Data are averaged across 5/group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. Figure 3. Bar graph of group mean AE (±SEM) within each S-R interval for 142 unpredictable -correct direction responses. Data are averaged across 5 subjects/group and 6 target amplitudes. Control group = striped bars; cerebellar group = dark bars. Figure 4. A) Bar graph of mean AE (±SEM) for unpredictable-correct direction 144 responses within each S-R interval. Data are averaged across 5 cerebellar subjects for each target amplitude (short, medium, and long). For the cerebellar group, extent specification largely involved improved performance to the long target amplitude There was a significant decrease in AE for the cerebellar group across all four preparation intervals (post-hoc P values between intervals ranged from .0001 to .03). B) Bar graph of mean AE (±SEM) for unpredictable-correct direction responses within each S-R interval. Data are averaged across 5 control subjects for each target amplitude (short, medium, and long). For the control group, both long and middle amplitude targets were specified between 0 and 299 ms with no further improvement in accuracy beyond 299 ms. Absolute error to the short target remained relatively stable across preparation intervals for the control group. Figure 5. Symbols represent group means for AE (±SEM) to each target 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xiii amplitude for unpredictable-correct direction responses. The cerebellar group is represented by the triangle and the control group by the squares. Data are averaged across S-R interval and 5 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0001). Figure 6 . Scatter plot of constant error (CE) across S-R intervals for a full 146 complement of correct direction, unpredictable condition responses. Regression lines are run through each target amplitude. A thicker horizontal line at 0° represents the target. For cerebellar subject 7 (A), and control subject 5 (B), the short (15°) target amplitude was the default. For cerebellar subject 8 (C) and control subject 1 (D), the default position was the medium target (30°) amplitude. Note regression line for medium target for S8 is superimposed on 0°. Figure 7. Stack bar plot of percent variance in displacement for unpredictable- 147 correct direction trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The cerebellar group is to the left, the control group is to the right. Data are averaged across 5 subjects/group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. Figure 8 . Ensemble averages of movement velocity, biceps, and triceps muscles 149 for a representative control subject (S2). Left panel are responses to the 30° extension target (default); right panel are responses to the 45° extension target. The first, fourth, and seventh row of signals for both target amplitudes are averaged velocity, biceps and triceps profiles for the longest S-R interval (300- 400 ms) in the unpredictable condition. The second, fifth, and eighth row of signals are averaged velocity, biceps and triceps profiles for the shortest S-R interval (0-99 ms) in the unpredictable condition. The third, sixth and ninth row of signals are averaged velocity, biceps and triceps profiles from predictable condition responses. The number of averaged responses for the 30° target was 29,19, and 12 from first to third row. For the 45° target, response number was 30,17, and 11 respectively. Offset of movement, defined as return to zero velocity is represented as the range of offsets for the three preparation intervals. Note averaged AE for the three preparation intervals for both target amplitudes. Figure 9. Ensemble averages of movement velocity, biceps, and triceps for a 150 representative cerebellar subject (S9). Left panel are responses to the 15° extension target (default); right panel are responses to the 45s extension target. The row designation is as described in A, The first two signals for both target amplitudes are responses for the longest and shortest S-R intervals in the unpredictable condition. The 3r d signal are averaged predictable condition responses. Arrows in the left panel, signal-rows 6 and 9 (predictable condition biceps and triceps) identify that the EMG activation pattern appears to be triphasic. The number of averaged responses for the 15° target was 26,19, and 16 from top down. For the 45° target, response number was 25,20, and 16, respectively. Figure 10. Bar graph of correct-direction, unpredictable-condition responses 152 representing percent variance in displacement due to compensatory adjustments over S-R interval. Data were averaged across 10 subjects (5 control and 5 cerebellar subjects). The shortest preparation interval (0-99 ms) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xiv was significantly different than the two longest preparation intervals (i.e., 200-299 ms, and 300-400 ms), P =.001 and .009, respectively. Figure 11. A) Bar graph of mean AE (±SEM) to the virtual target for wrong 155 direction responses within each S-R interval. Data are averaged across all 10 subjects for each target amplitude (short, medium, and long). Wrong direction responses did not converge on the correct target extent as preparation time increased. Across preparation intervals, wrong direction responses were positioned between the short and middle target amplitudes (-22°). B) Frequency of direction errors as a percent of the total responses within each S-R interval for the stroke and control groups in the unpredictable condition. The time course for direction specification was significantly different for the two groups (X2 = 78.40, p < .00001). C) Stack bar plot of percent variance in displacement for unpredictable, wrong direction responses by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The cerebellar group is to the left, and the control group is to the right. Data are averaged across 5 subjects for the cerebellar group and 5 subjects for the control group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. Chapter 5: Figure 1. A) Bar graph of group mean AE (±SEM) for predictable condition 1 7 0 responses. Data are averaged across group, S-R interval and 6 target amplitudes. CB group (Experiment 3) = striped bar - averaged across 5 subjects; Low-Adjustment SM area stroke group (Experiment 2) = black bar - averaged across 3 subjects; High-Adjustment SM area stroke group (Experiment 2) = black bar • averaged across 3 subjects. B) Bar graph of group mean AE (iSEM ) for unpredictable condition - correct direction responses. Data are averaged across group, S-R interval and 6 target amplitudes. CB group (Experiment 3) = striped bar - averaged across 5 subjects; Low-Adjustment SM area stroke group (Experiment 2) = black bar • averaged across 3 subjects; High- Adjustment SM area stroke group (Experiment 2) = black bar • averaged across 3 subjects. Figure 2. Scheme (adapted from Allen & Tsukahara, 1974) showing proposed 1 7 5 roles of several brain structures in premovement planning and response updating. It is proposed that basal ganglia (not addressed in this study) and the neocerebellum are involved with association cortex in the planning, programming of volitional movement. Solid- thick black lines and arrows represent efferent and afferent connections of the planning pathway: 1) Association cortex (Assn Cx) known to have a role in planning includes; Supplementary motor area (SMA); Premotor (PM); Prefirontal cortex (PF) and Posterior parietal; 2) Cerebrocerebellum includes the lateral cerebellar hemisphere with the dentate as the primary output nucleus to the cerebellar thalamus. The cerebellar thalamus refers to the histologically identified nuclei which receive cerebellar afferents and send efferent projections to motor cortex and back to association cortex. Listed for the purposes of this study are the ‘motor’ thalamic nuclei of the cerebellar thalamus: VPLo = the pars oralis of the ventral posterolateral nucleus; VLc = ventral lateral, pars caudalis; VIps = ventral lateral, pars postrema; X = nucleus X. For the purposes of this study the adjustment pathway (depicted by broken lines and arrows) includes: 1) Motor cortex - specifically primary motor cortex (M1); 2) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XV the spinocerebellum which includes the intermediate cerebellar hemisphere with the interposed nuclei as the primary output nucleus to the cerebellar thalamus with efferent projections to motor cortex. The 3 different symbols placed along various lines represent sites that with damage, could potentially produce behaviors identified for each of the 3 groups: Low-adjustment SM area stroke group; High-adjustment SM area stroke group and Cerebellar group. The symbols do not represent specific lesion locations of the subjects in the 3 experiments presented. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xvi ABSTRACT FUNCTIONAL BRAIN CORRELATES FOR PREMOVEMENT PLANNING AND COMPENSATORY ADJUSTMENTS IN RAPID AIMED MOVEMENT The distinction between pre-movement planning and feedforward adjustments has been considered crucial to an understanding of the neural control of goal-directed movement. Rapid and accurate responses without manifest discontinuities are controlled by parallel neural processes that concurrently provide for trajectory planning and feedforward updating as the response unfolds. However, the functional neural substrates associated with these parallel 'paths' have yet to be determined. The purpose of this dissertation was to provide evidence for a functional neural distinction between these two motor control processes. Numerous reports identify a role for the cerebellum in premovement planning, while sensorimotor (SM) areas appear to support the execution of ongoing adjustments to the movement trajectory. To test these hypotheses, the performance of six subjects with unilateral SM area stroke were compared to that of matched control subjects and 5 subjects with unilateral cerebellar stroke were compared to that of matched control subjects under conditions of a timed-response movement paradigm. Subjects rapidly flexed or extended the forearm to targets presented in either a fixed (predictable condition) or a random sequence (unpredictable condition). Subjects with SM area stroke used the limb ipsilateral to the side of the SM lesion. Subjects with cerebellar stroke used the limb that was contralateral to the side of the cerebellar lesion. Time to prepare the response was manipulated by varying the time of target presentation relative to an auditory cue for movement initiation. Kinematic analysis and multiple regression were used to determine the effect of premovement planning and trajectory updating on end-point accuracy. Both the subjects with SM area strokes and those with cerebellar strokes were significantly less accurate than their matched controls. Overall, the source of the inaccuracy for individuals with SM area strokes was largely the result of a deficiency in the performance of feedforward compensatory adjustments. In contrast, deficits in premovement planning accounted for the inaccuracy for individuals with cerebellar strokes. These results suggest a role for SM areas in feedforward updating while the cerebellum has substantial involvement in the planning of goal-directed aiming tasks. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 CHAPTER 1: INTRODUCTION Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in 1899, Woodworth recognized that the very thing that makes a movement useful and purposive is its accuracy (Woodworth, 1899). In 1949, Brown and Slater-Hammel reasoned that determining how accurate movements are performed is essential in light of the fact that “ when an attacking aircraft is suddenly perceived the guns must be slewed around to the target with the least possible delay” (Brown & Slater-hammel, 1949). Although the specific interest of investigators in studying the accuracy of aiming movements might be motivated by the era in which they have worked, a common thread through research over time is the considerable interest to understand the operations by which the central nervous system (CNS) both transforms target information into response parameters and executes movement to a target. Direction and extent as spatial and dynamic ‘ingredients’ or parameters of goal-directed aiming movements have been studied in an effort to understand the planning of movement prior to execution as well as the modifications to the trajectory during movement that are necessary for accurate outcomes. The distinction between pre-movement planning and adjustments or modifications as the trajectory unfolds has been considered crucial to an understanding of the neural control of goal-directed movement (Gordon & Ghez, 1987a). The purpose of the three studies presented here was to provide evidence of the neuro-anatomical basis for these two motor control operations. Therefore, behavioral performance measures functionally associated with pre-movement planning and ongoing modifications were compared in individuals with known damage to central nervous system (CNS) areas presumed to subserve these two mechanisms. In addition, an experimental paradigm that manipulates the time that subjects have to prepare upper extremity aiming movements to targets varying in both direction and amplitude was used in order to partition those response components attributed to pre-movement planning and those attributed to response modifications. Thus, the purpose of the three studies to follow was to determine the neural substrates identified with premovement planning and compensatory adjustments. As such, the following discussion will focus on both behavioral and neurophysiological evidence for these two motor control operations. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 Pre-movement Planning and Trajectory Modifications: Past and Present Historically, simple limb movements to a target were thought to result from the coupling of two actions. Early models of motor control advanced the idea that virtually all aiming movements proceed in a two-stage fashion, with an initial ballistic phase followed by a corrective phase (Brooks, 1979; Meyer, Abrams, Komblum, Wright, & Smith, 1988; Woodworth, 1899). The ballistic phase is choreographed by pre-movement instructions to an effector. Given that the ballistic phase cannot be altered once initiated and does not use feedback information and error- detection processes, it is presumably regulated by open-loop control processes. In the case of a single-joint movement to a target (e.g. movement of a manipuiandum about the elbow joint), the kinematic signature of the pre-planned action is a stereotypical bell-shaped velocity profile (Brooks, 1979; Latash & Anson, 1996; Ojakangas & Ebner, 1991). Such responses have been termed continuous because the trajectories show no overt deflections indicating the presence of corrections (Brooks, 1979). Most of the distance to the target is covered as the limb is rapidly propelled by the initial impulse governed by the prior plan Once executed, it was assumed that very little modification would occur in the movement for the next few hundred milliseconds or so (Desmedt & Godaux, 1978). Then in the corrective phase, feedback or closed-loop control is utilized for computation and correction of errors. Thus, accurate target-directed movement is assured as discrepancies between hand and target positions are assessed. It was generally considered that very rapid responses have only a single ballistic phase. Presumably, there is no time for peripheral feedback-based corrections in these rapid responses. It was also considered that trajectory corrections can only be implemented at the completion of the ballistic phase (Desmedt & Godaux, 1978). However, within the last 25 years, it has become evident that even rapid movements may show overt corrections during their course. Because the corrections occur too soon to be accounted for by peripheral feedback, they are presumably triggered by early or initial trajectory 'errors’ that are detected by internal monitoring of neural signals (Cooke & Diggles, 1984; Evarts, 1971; Higgins & Angel, 1970; Vicario & Ghez, 1984). Cooke and Diggles (1984), observed very rapid corrections in movement trajectories as healthy Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 subjects performed a visual tracking task. Given the rapidity with which these corrections occurred, it was suggested that they could not be based on information from the moving limb but must have been prepared and implemented while the effects of the pre-planned action were unfolding. Thus it is thought that the CNS internally simulates the dynamic behavior of the motor system in planning and control by utilizing internal feedback (Evarts, 1971) or feedforward control (Ghez, Hening, & Gordon, 1991). By providing advance information rather than feedback generated by the response itself, feedforward control can be used to adjust the controlled variables before events occur that would influence them. Thus, this type of control potentially solves a fundamental problem in motor control: the long conduction delays in most sensorimotor loops make feedback control too slow for the rapid corrections that have been observed. Feedforward control appears to be a parsimonious way for the system to manage an unexpected or unpredictable change (perturbation) in task constraints. This is particularly true if there has been previous experience with the range of values the upcoming perturbation could assume. Premovement Planning Behavioral Evidence Reaction Time as an Index of Premovement Planning The concept of preplanning of motor responses refers to a mode of control in which commands are structured before a movement sequence begins (Klapp, 1977). Preplanning or programming of a response has been inferred from a relation between reaction time (RT), the speed with which one reacts to a signal, and the nature of the response which follows. Evidence that reaction time (RT) reflects premovement planning or central programming was provided in an experiment by Henry and Rogers (1960). The investigators used a simple-RT paradigm in which subjects knew on any given trial which response was to be made. However, in different series of trials, subjects performed 3 movements that varied in complexity. The first movement involved merely lifting a finger from a key. The second movement required that the subject lift the finger from the key, move the arm forward and upward to grasp a tennis ball suspended on a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 string. The third movement required that the subject lift the finger from the key, move forward and upward to strike the tennis ball with the back of the hand, move forward and downward to push a button, and then strike a second suspended ball. The reaction time increased significantly from the first to the second and from the second to the third movement. Given that the stimulus for movement was kept constant and that there was only one possible response for each of the three movements, the only variable that could account for the increase in reaction time was the increasing complexity of the movements. These results suggested that as the movements increased in complexity more central nervous system (CNS) processing was required in order to preplan or program the movement (Henry & Rogers, 1960). The increase in reaction time that is observed when simple- is compared with choice-RT has been explained in a similar way. As opposed to the simple RT paradigm, in a choice setting the subject does not know in advance what movement will be required when a signal to respond occurs. Instead, the subject must either program the movement to execute a response after receiving the response signal or must simultaneously prepare all possible responses prior to the ‘go" signal and then select the appropriate response. In either case, the CNS processing of signal information for construction of the plan or selection of the appropriate plan is a time- consuming process observable as an increase in choice RT. Thus, the delays between target presentation and movement onset seen in both the simple- and choice RT paradigms have been attributed in part, to signal processing in the CNS needed to prepare an adequate response. Further confirmation of the existence of preplanning or preparatory activity has been provided by experiments in which advance information about the upcoming movement is provided before the target comes on. In this variation of the choice RT paradigm called the precuing technique (Rosenbaum, 1980), a precue is given before the presentation of the reaction signal, which provides advance information about some, none, or all of the values of parameters that will need to be specified for an appropriate response. For example, if a precue involves the illumination of targets on one side of a central location, then the subject has advance information regarding direction of the upcoming movement. The assumption is that the subject will make use Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 of the precue by specifying in advance the precued values. Given advance preparation, it is assumed that those values specified during the precue will not have to be specified during the RT interval. Conversely, those values that were not specified in advance will have to be specified during the RT interval. The differences in RT for a given movement will then depend on the number and types of values that were preplanned (Rosenbaum, 1980). Studies utilizing such precues have found reductions in RT, which implies that at least some aspects of movement planning have been advanced in time and have occurred before presentation of the actual target (Rosenbaum, 1983). Trajectory Indices of Premovement Planning According to a preplanning or programming view, movements with different trajectories are produced by scaling prestructured force-time functions in magnitude and duration (Bock, 1994). Experimental support for the preplanning concept has been derived largely from studies of hand trajectories during aimed arm movements. A number of invariant kinematic and electromyographic characteristics have been found suggesting that the movements were brought about by stereotyped force-time functions. Electromyography (EMG1. In agreement with the concept of central preplanning, EMG studies have documented certain regularities in the pattern of muscle activation during movements as well as during rapid targeted force pulses (Gordon & Ghez, 1984). Thus, a consistent tri-phasic EMG pattern is exhibited in fast arm movements. In a rapid elbow-flexion action, a burst of the agonist muscle (biceps) occurs first, then the agonist is turned off and the antagonist (triceps) is turned on presumably to bring the limb to a stop. The antagonist muscle is then turned off, and the agonist comes on again near the end of the response. There are two lines of evidence that this pattern of activation is centrally planned. For one, the movement is so rapid that it would not be possible for the system to use feedback from the responding limb to trigger the end of activity in one muscle and initiate activity in the other. The movement will have finished before a feedback based process could be completed. Secondly, the triphasic burst pattern was preserved in a patient incapable of utilizing proprioceptive feedback secondary to a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 severe pan-sensory neuropathy (Hallett, Shahani, & Young, 1975; Hallett, Berardelli, Matheson, & Marsden, 1991) Kinematic index of preplanning. It has been shown that human subjects performing rapid isotonic arm movements (Freund & Budingen, 1978) or isometric force pulses (Gordon & Ghez, 1984,1987a) as well as cats applying forces to a lever (Ghez & Vicario, 1978) or reaching for food in horizontal target wells (Martin, Cooper, & Ghez, 1995) accurately capture targets at different amplitudes by adjusting the rate of a muscle contraction in such a way that the contraction time remains approximately constant across the different target amplitudes. Thus, duration invariance (Bullock & Grossberg, 1988) in the peak rise of isometric tension or in the peak displacement of isotonic movements is maintained while subjects respectively, modulate or scale the rate of force rise or the velocity of isotonic movements in order to vary the amplitude of the response. Gordon and Ghez (1987), called this strategy of amplitude modulation of an underlying stereotyped waveform, pulse-height control. On the basis of two key results, Gordon and Ghez concluded that this pulse-height strategy appears to constitute a critical component of the central plan necessary to achieve accurate matching of force impulses to targets. The finding that early dynamics of the trajectory (in particular, peak acceleration, which occurs at about 25% of the force rise time) were highly predictive of the ensuing peak force and were scaled to the magnitude of target displacements, indicates that response amplitudes were largely preplanned. Further, the degree to which the initial peak acceleration predicts the peak force achieved represents a measure of the contribution of preplanning to trajectory formation. Secondly, while duration invariance was observed both in responses performed under Fast (“ make the force impulse as brief as possible” ) and Accurate (“ be as accurate as possible without regard to rise time” ) conditions, force rise time was more consistently regulated around a constant value when subjects were attempting to be as accurate as possible (Gordon et al., 1987a). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 A Default Response is Preplanned Under Certain Task Constraints Traditionally in the RT literature, the amount of information available to the performer in advance of a motor response was considered one of the most significant factors in determining the extent to which the response could be completely preplanned. When the task goal is to move rapidly and accurately to a single target (simple RT condition), the performer can presumably specify all of the details of the movement in advance and then essentially, trigger the action at the appropriate time (Rosenbaum, 1983). How does the nervous system resolve the problem of planning an action when little information is available to the performer? How does the performer prepare a response which may be subjected to an unexpected perturbation? How does the performer contend with having to respond before sufficient information is available? When uncertainty limits the degree to which the action can be entirely preplanned, the nervous system appears to resolve the problem of planning the action by assigning the response a value that is distributed around the center of the range of possible values. Thus, across a number of different experimental paradigms, subjects appear to prepare a predictive motor plan or default response. The establishment of such predictive behavior appears to be based upon information from the results of past responses (Ghez, 1979). In other words, previous experience with the range of possible target values, enables a subject to manage an unexpected or unpredictable change (perturbation) in task constraints by providing a default value near the center of the range of target values. When more information is available to the system with respect to perturbation parameters an appropriate response is obtained by simply increasing or decreasing responsiveness or gain around the default response. In their initial studies, Ghez and Gordon (1987) examined optimally prepared responses. In other words, subjects could respond when presumably all sensorimotor processing was completed. The trajectory invariances that were observed over a wide range of amplitudes suggested that response amplitude was governed by a centrally generated pulse height control policy (Gordon & Ghez, 1987a). The next series of experiments (Favilla, Hening, & Ghez, 1989; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 Favilla, Gordon, Hening & Ghez, 1990; Favilla & De Cecco, 1996; Ghez, Hening, & Favilla, 1989; Hening, Favilla, & Ghez, 1988; Hening, Vicario, & Ghez, 1988) were designed to ask a key question regarding trajectory formation: what aspects of trajectory control are impaired when subjects are required to respond without adequate time to accurately specify response features? As with the earlier studies, the task involved generating isometric forces around the elbow in order to produce an impulse of force that would match a target level presented on a screen. At the beginning of each trial, the subject aligned their force line with the target line. At an unpredictable time, the target was stepped up to a new level on the screen. Following the visual stimulus, subjects would attempt to generate a peak force to match the new level of the target. Trials were given in blocks in which successive target steps were either predictable (simple condition) or unpredictable (choice condition). The time available to the subjects for processing target information was varied in two ways. First, two different instructions were added to a conventional reaction time paradigm. Subjects were either to respond as soon as possible (ASAP condition) or when they felt fully prepared to respond, which was the when ready (WR) condition. Second, an experimental paradigm that controlled more precisely the time between target presentation and response onset (the Stimulus-Response or S-R interval) was developed. In this timed response paradigm (TRP), subjects performed the same task as detailed earlier, however, now they were also presented with an auditory stimulus with which they were to synchronize the initiation of their responses. Subjects heard a series of 4 tones and were to synchronize the onset of their response with the fourth tone. At an unpredictable time in the 500 ms interval between the third and fourth tone, the target was stepped to a new level. Thus, the amount of time that subjects had to prepare the appropriate response was systematically varied from 0 to 500 ms. As such, responses were made at S-R intervals both shorter and longer than the minimum latency needed to fully specify a response to the target. Both the choice responses initiated ASAP to unpredictable targets, as well as responses specified at S-R intervals of < 250 ms in the TRP paradigm were significantly less accurate than Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 responses initiated WR or responses specified at S-R intervals of > 250 ms. For the purposes of the present discussion, the form of this inaccuracy is the most critical result. In choice responses initiated ASAP to unpredictable targets, and responses specified at S-R intervals of < 250 ms, there was a systematic distortion in scaling of the entire set of responses to the different targets. The range of peak forces was markedly constricted around a modal value near that of the middle- sized target. The fact that both sets of responses showed the stereotyped configuration of trajectory profiles with duration invariance demonstrates that the amplitude of the responses continued to be governed by the same overall pulse-height control strategy. Thus, it appears that prior to target presentation, a centrally generated default response is prepared that is near the center of the range of target amplitudes previously experienced. In two of the studies of this series in addition to target amplitude, direction was added as a response parameter. Thus subjects produced isometric flexion and extension force pulses to three target amplitudes in either direction. Under these experimental constraints, subjects prepared a single default amplitude, but selected direction at random (Favilla et al., 1989; Favilla et al., 1990). In summary, the presence of trajectory invariances such as duration invariance, regularities in the pattern of muscle activation, and the finding that early dynamics of the trajectory (i.e., peak acceleration) are highly predictive of the ensuing peak force or displacement, infers the degree to which the responses are governed by a prior plan. This interpretation is in accordance with the purpose of preplanning an action: the need for the nervous system to adopt simplifying strategies and thus reduce the number of variables that must be controlled. Preplanning appears to be the specification of a default response when uncertainty limits the degree to which the action can be entirely preplanned. The default response embodies the trajectory invariances which index preplanning. However, prior to an unpredictable target presentation the default response is prepared near the center of the range of target amplitudes previously experienced. In this way, the performer can contend with uncertainty by assigning the response a value that is distributed around the center of the range of possible values and then Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 simply scale the response up or down when more information is available to the system with respect to response parameters. NeuroDhvsioloaical Evidence Single cell recording techniques have been extensively used to measure neuronal activation in different brain areas as non-human primates perform upper extremity choice reaction time tasks. Commonly, the choice RT paradigm is modified to include an instructional delay period (IDP). The IDP occurs after a preparatory signal (PS) which informs the animal completely, partially, or not at all about parameters of the forthcoming movement and before a response signal (RS) that calls for execution of the requested movement. In this way, investigators can identify those neurons whose activity is modulated with movement preparation, and those whose activity is modulated with movement execution. The neuronal activity during the delay period has become known as ‘ ‘set’ ’ related activity (Evarts, 1971). Motor “ set” or preparation is the state of readiness to make a particular movement. There exists substantial support for the interpretation that premotor cortex (PM) activity during the IDP reflects motor “ set.” Premotor cells have been observed to be differentially active during the delay periods after instruction signals specifying movements to be made in different directions (Fu, Suarez, & Ebner, 1993; Fu, Flament, Coltz, & Ebner, 1995; Kurata, 1993; Riehle & Requin, 1989,1993,1995; Riehle, Mackay, & Requin, 1994; Tanji, Okano, & Sato, 1988; Weinrich & Wise, 1982). Given that a reaching movement is a vector in space that can be described fully by its direction and amplitude, it is not surprising that there exists an abundance of studies that have identified preparatory neural coding of direction (Georgopoulos, 1995). In addition to PM, directionally selective, preparatory-related changes in neuronal activity have been found in a number of other cortical structures, such as primary motor (Ml) cortex (Alexander & Crutcher, 1990a; Ashe et al., 1993; Georgopoulos, Crutcher, & Schwartz, 1989; Riehle & Requin, 1989; Tanji & Evarts, 1976), supplementary motor area (Alexander & Crutcher, 1990b) and the parietal cortex (Crammond & Kalaska, 1989). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 The directional selectivity of motor cortical cells was studied by training rhesus monkeys to move a manipulandum toward visual targets on a planar working surface. The electrical activity of single cells in the forelimb area of the contralateral motor cortex were recorded extracellularly while the monkeys performed the task (Georgopoulos, Kalaska, Caminiti, & Massey, 1982). The main finding was that the activity of single motor cortical cells was broadly tuned around a “ preferred direction.” The cell discharge was highest with movements in that direction and decreased gradually with movements made in directions farther away from the preferred one. The broad directional tuning indicated that a given cell participates in movements of various directions and that a movement in a particular direction will involve the activation of a set or population of cells. Georgopoulos and colleagues thus proposed the population-vector hypothesis which regards a population of cells as an ensemble of vectors. Each vector represents the contribution of a directionally tuned cell. It points in the cell’s preferred direction and is weighted according to the change in cell activity associated with a particular movement direction. This weighted vector sum of these neuronal contributions is the neuronal population vector. The vectorial representation of directional information at the motor cortical level is a distributed code in the sense that each neuron carries only part of the information, whereas the neuronal population determines uniquely the direction of reaching (Georgopoulos, 1994,1995). From the population vector analysis there is irrefutable evidence that the population vector in Ml points in the direction of the planned movement during the RT and during an IDP. However, it is important to realize that motor cortex is the site of convergence from a large number of other areas. The fact that input structures to the motor cortex, such as the cerebellum (Fortier, Kalaska, & Smith, 1989; Schmahmann & Pandya, 1997; Smith, Dugas, Fortier, Kalaska, & Picard, 1993), PM (Caminiti etal., 1991), and parietal cortex (Kalaska & Crammond, 1992), demonstrate distributed population coding suggests that the discharge patterns of motor cortical cells are generated through this convergence. In addition, across the studies that have compared neuronal activity changes in Ml with other motor cortical regions during the preparatory period, it is consistently observed that a far greater percentage of preparatory neurons are Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 located in PM and SMA as opposed to Ml (Alexander & Crutcher, 1990a; Riehle & Requin, 1989, 1993,1995; Tanji et al., 1988). The Cerebellum: A Neural Substrate of Premovement Planning In that the involvement of a number of motor structures in movement preparation is well documented, it is reasonable to suppose that many structures cooperate in this process. However, neurophysiological investigations have placed an emphasis on the discharge characteristics of premotor neurons that precede movement execution during an imposed dely period. As such, a substantial involvement of PM in movement preparation or set has been identified (Alexander & Crutcher, 1990b; Weinrich & Wise, 1982). The premotor cortex is a major cortical target of the lateral cerebellar hemisphere. This anatomical subdivision of the cerebellum corresponds to a functional region of the cerebellum known as the cerebrocerebellum (Ghez & Thach, 2000). The cerebrocerebellum is thought to play a special role in the planning and initiation of movement. The cerebrocerebellum projects to the dentate nucleus, which sends fibers through the superior cerebellar peduncle to the ventral lateral nucleus of the thalamus. From the ventral lateral nucleus, the dentate nucleus influences motor and premotor regions of the cerebral cortex. It has been observed that the activity of the dentate neurons during voluntary movement tends to slightly precede that of motor cortical neurons and so it is perhaps through this pathway that an action planned in the cerebrocerebellum is relayed to premotor cortex. The results of several studies suggest that the cerebrocerebellum participates in preparing the motor cortex for self-paced movements as well as trained movements in response to an external stimulus (Sasaki, Gemba, & Hashimoto, 1981). This has largely been indexed by modification of the excitability of motor cortex as the result of cerebellar lesions (Di Lazarro et al., 1994). To determine the influence of cerebellar involvement on the preparatory state of the cerebral cortex for voluntary movements, the movement-related cortical potentials preceding sequential and goal-directed finger and arm movements in patients with cerebellar dysfunction Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 were studied (Tarkka, Massaquoi, & Hallett, 1993; Wessel, Verleger, Nazarenus, Vieregge, & Kompf, 1994). Movement-related cortical potentials can be recorded on the scalp and analyzed with electroencephalographic (EEG) averaging. Two of these potentials, the bereitschaftspotential (BP), and the negative slope are recorded as early as 1 second and 400 ms, respectively, before the onset of movement and as such, are likely to represent the general preparation for voluntary movement. The amplitude of movement-related cortical potentials close to movement onset was reduced in cerebellar patients suggesting inadequate cerebellar activation of the motor cortex. Thus, a strong input from the cerebellum seems to be crucial for the generation of a normal motor potential close to the movement onset. Additionally it was found that the distribution of movement-related cortical potentials was more diffuse and bilateral in the cerebellar subjects compared with controls. It was suggested that this finding may be related to a more extended cortical activation as a sort of compensation for the greater difficulty in executing the movement (Tarkka et al., 1993; Wessel et al., 1994). The diffuse activation of the primary motor cortex in the patients could be due to impaired cerebellar input to the primary motor cortex. Neurons from the dentate area of the cerebellum discharge in relation to movement intent before the onset of movement, providing important input for the motor cortex to initiate the intended movement. Finally, measures of regional cerebral blood flow (rCBF) using Positron Emission Tomography (PET) demonstrated decreased activation of PM in individuals with cerebellar damage compared with controls performing a self-paced sequential finger opposition task. It was suggested that this was most likely a consequence of decreased cerebellar input via the thalamus (Wessel, Zeffiro, Lou, Toro, & Hallett, 1995) The behavioral manifestation of the decreased movement-related cortical potentials referenced above, is possibly the delay in the initiation of movement commonly observed with lesions of the cerebrocerebellum. Two mechanisms have been proposed to account for the delay in the initiation of movement. First, the dentate nuclei might provide background facilitation to either cortical or subcortical neurons so that, after dentate lesions, commands to initiate movement could bring the motor neurons to fire only after an increased period of summation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 Second, the dentate nucleus might participate in or in fact, convey the commands initiating movement (Ghez & Thach, 2000). A study by Brooks and Thach (1981), provided some insight as to which of these two alternatives might explain delays in movement initiation. The patterns of activity of neurons in the motor cortex of monkeys were recorded while the animals moved the contralateral arm in response to a visual cue. These patterns were then compared before and after reversible inactivation of the dentate. Inactivation was achieved be cooling through a probe inserted into the dentate nucleus. Both the discharge of motor cortex neurons associated with the movement and the onset of the movement itself were delayed. If the dentate was merely providing background excitation, the change in activity of neurons in the motor cortex should have occurred at the normal time, but more time should have elapsed before the onset of movement. Instead, the results suggest that the dentate nucleus provides important commands for initiating movement in the motor cortex (Brooks & Thach, 1981). Additional behavioral evidence that the cerebellum participates in the preplanning of voluntary action has been found. In a study by Bonnefoi-Kyriacou and colleagues (1995), patients with cerebellar dysfunction and age-matched controls performed pointing movements with the arm. In one condition, the task was a simple RT movement directed toward a spatially defined target. The other two conditions involved choice tasks in which the amplitude and direction of movement were varied. The reaction times and movement times were significantly longer in the cerebellar patients than in controls. Thus, delay in initiation and execution of the movement was found. Again, it was suggested that this might be the result of loss of excitatory feedback from the cerebellum to the motor cortex (Bonnefoi-Kyriacou, Trouche, Legal let, & Viallet, 1995). As discussed earlier, the triphasic EMG burst pattern is a behavioral index of preplanning. It has been found that the triphasic pattern that characterizes a ballistic movement in normal subjects, is impaired in cerebellar patients. The main alterations observed are a prolonged agonist burst and a prolonged acceleration time (Hallett et al., 1991). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 Compensatory Adjustments Behavioral Evidence As discussed earlier, it was considered that very rapid aiming responses have only a single ballistic phase and that presumably, there is no time for peripheral feedback-based corrections in these rapid responses. If trajectory corrections were required to assure accuracy, they could only be implemented at the completion of the ballistic phase (Desmedt & Godaux, 1978). However, it has become evident that even rapid movements may show overt corrections during their course. Because the corrections occur too soon to be accounted for by peripheral feedback, they are presumably triggered by early or initial trajectory 'errors’ that are detected by internal monitoring of neural signals. (Cooke & Diggles, 1984; Evarts, 1971; Higgins & Angel, 1970; Vicario & Ghez, 1984). Reaction Time and Kinematic Indices of Compensatory Adjustments Early evidence of corrective adjustments from the RT literature came from the work of Rosenbaum (1980). In his precue RT study, Rosenbaum found that reaction times were decreased minimally when extent was precued. The fact that arm and direction precues facilitated reaction times more than extent precues led Rosenbaum to suggest that precues for direction and arm allowed for advance specification of these parameters whereas precues about extent may not facilitate motor preparation to the same degree. Based upon two findings it appeared that at least some decisions about extent were made after decisions about arm and direction. First, a significant increase in movement time for far targets compared to near targets when extent remained to be specified, suggested to Rosenbaum that extent decisions could be made during movement time. Second, less errors were made when extent remained to be specified. The explanation for this finding, coupled with the longer reaction times for arm and direction specification was that extent errors could be corrected during movement time, whereas arm and direction errors could not Reaction times for arm and direction thus reflect the difficulty of correcting errors during movement times. Less time preparing and specifying extent is needed given that it can be completed during the movement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 Measurements of error correction time (ERT), an RT specific for amending an incorrect response suggested that error detection could rely on internal feedback or feedback of a central rather than peripheral origin (Higgins & Angel, 1970). Subjects engaged in a pursuit tracking task and were instructed to superimpose the cursor directly over the target and to restore the alignment as rapidly as possible whenever the target jumped to a new position. The subjects additionally were told that if an incorrect move was performed, it should be corrected as soon as possible. Measurements of ERT were compared with measurements of proprioceptive reaction time (PRT). Proprioceptive reaction time was the time between the subject feeling movement or force applied to the joy stick and the application by the subject of a resisting force. In every case, the mean ECT was less than the mean PRT. It thus appeared that the subjects were able to monitor their behavior internally, comparing the actual motor commands with some reference value. The errors appeared to be amended by a central mechanism operating more rapidly than sensory feedback. Additional evidence for this idea came from a study in which subjects performed a visual tracking task (Cooke & Diggles, 1984). In addition to error correction time, error-related EMG activity was measured. The ECT results were comparable to those of Higgins and Angel (1970). For error movements, corrections occurred more rapidly than could be accounted for by proprioceptive feedback eliciting a reaction time movement. Suppression of the EMG activity in the muscle producing the error movement started as early as 20-40 ms after the initiation of the error-elated EMG activity and as much as 50 ms before any overt sign of limb movement. It was concluded that the first step in the error correction, suppression of drive to the muscle producing the error movement, could not be based on information from the moving limb, but rather on central monitoring of the commands for movement. Finally, Henry (1953) showed that subjects could make changes in position and changes in tension that were so small that the subject could not perceive them. Standing, blindfolded subjects pushed against a handle with their hand. The handle was attached to a mechanical device that could alter the position of the handle. In one condition subjects were to keep the pressure against the handle fixed. When the handle pushed against the subject, the correct response was to ease up on the handle so that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 tension was held constant. In a second condition, the goal was to maintain a constant position of the handle which meant that the subject had to increase or decrease pressure exerted against the handle as the handle pressure changed. A third condition assessed the conscious perception of change in pressure. Subjects held their arm immobile and reported when a change in the pressure exerted by the apparatus was sensed. When the subject was asked to report a conscious change, a force of .559 dynes was required for detection. However, when the subject responded to changes in tension and position, forces of only .296 and .029, respectively, were successfully detected and responded to by the muscular system. The results suggest that internal, unconscious mechanisms generated the corrections that were made (Henry, 1953). Gordon and Ghez (1987) formulated a statistical index of compensatory adjustments that occur too soon to be accounted for by peripheral feedback. They relied upon an analysis of the variability of targeted force pulses in which the contribution of compensatory adjustments to the overall variance in peak force could be determined. It was found that 4 to 29% of the variance in peak force achieved was explained by compensatory adjustments. When subjects are forced to respond ASAP to unpredictable targets, they initially preplan the peak force that will be achieved by the modulation or scaling of an underlying stereotyped waveform. The initial peak acceleration, which occurs early in the rising phase of the force impulse, indicated the accuracy of the early phase of the trajectory and provides an operational measure of the preplanned scaling of responses. If the preplanned specification of the force impulse was correctly matched to the target then all of the variability in peak force will be accounted for by the initial peak acceleration. If there is a mismatch between the actual initial peak acceleration and the initial peak acceleration necessary to accurately capture the target, then an overshoot or undershoot of the target will occur. While this mismatch did occur, the correct peak force was still achieved by the implementation of corrective adjustments to the later phase of the trajectory. The corrective adjustments take the form of adjustments to agonist and antagonist activity that effectively changes the force “ deceleration.’ Compensatory adjustments then, are deviations of the force rise time from the regulated value. Because the adjustments are present so early in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 trajectory and even modify the first agonist burst, they are unlikely to depend upon afferent feedback. The data suggests that, in addition and in parallel with the planned action, the nervous system specifically prepares to process information derived from internal feedback in order to correct initial trajectory errors. Neurophvsiolooical Evidence In their initial analysis of neural activity of cells in Ml and PM, Fu and colleagues (1993) determined that during the premovement period the activity of approximately half of the cells was related to both direction and extent of the movement. The activity of single neurons was recorded as two rhesus monkeys performed reaching movements in a horizontal plane. There were 48 targets placed at 8 different directions and 6 amplitudes. Both direction and extent for the upcoming movement were unpredictable. The discharge during the premovement period was significantly correlated with direction while the extent contribution overall during this period was less than direction. The correlation of discharge with extent significantly increased during the movement, suggesting that as a movement evolves, information about amplitude is continually updated. In all of the studies in which neuronal activity is recorded in several areas of the cerebral cortex as monkeys perform movement-precuing reaction time tasks, the activity of the majority of cells in the primary motor cortex is correlated with movement execution (Alexander & Crutcher, 1990a; Riehle & Requin, 1989,1993,1995; Riehle, Mackay, & Requin, 1994b; Tanji et al., 1988). Recordings in motor cortex include many units whose discharge begins after response onset. As suggested by Vicario and Ghez (1984), the late activity may reflect the analysis of information about the ongoing response to determine whether immediate corrective action is needed and for the comparisons required to test the accuracy of the current internal model. Changes in targeted limb trajectories produced by the reversible inactivation of motor cortex and red nucleus were studied as cats performed a tracking task with their forearm (Martin & Ghez, 1988; Martin et al., 1995;). The normal stereotyped form of isometric force trajectories was unaffected by inactivation of either site. However, while response amplitude was unchanged Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 during red nucleus inactivation, inactivation of the motor cortex resulted in hypometric responses. Thus motor cortex inactivation lead to a deficit in the animal's ability to scale the size of the response to an expected range of targets, but did not affect response initiation. Additionally, cerebellar and motor cortical activity were compared in monkeys responding to predictable perturbations of a hand-held object In contrast to cerebellar neurons, there was a relative absence of preparatory responses in the motor cortex. The activity of the neurons in the motor cortical region was related to muscles active during grasping. Smith et al. (1993), concluded that the motor cortex may be more intimately involved with compensatory increases in grip force after perturbations and have less involvement with preparing appropriate anticipatory strategies (Smith et al., 1993). Summary The distinction between pre-movement planning and ongoing adjustments or modifications has been considered crucial to an understanding of the neural control of goal- directed movement (Gordon & Ghez, 1987b). It has been suggested that rapid responses without manifest discontinuities are controlled by parallel neural processes that concurrently provide for trajectory planning and error correction as the response unfolds. However, the functional neural substrates associated with these parallel 'paths' have yet to be determined. The purpose of this dissertation is to provide evidence for a functional neural distinction between these two motor control operations. Numerous reports identify a critical role for the cerebellum in the generation of the plan, while SM cortex appears to have a critical role in the execution of ongoing adjustments to the movement trajectory. As such, subjects with cerebellar damage would be capable of implementing compensatory adjustments but demonstrate deficits in the ability to plan the response in advance. Conversely, individuals with unilateral SM area damage would show deficits in the ability to generate compensatory adjustments and update the response while premovement planning capability should be spared. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 CHAPTER 2: Experiment 1: DEFICITS IN COMPENSATORY TRAJECTORY ADJUSTMENTS AFTER UNILATERAL SENSORIMOTOR STROKE* * Fisher, BE., Winstein, CJ., Velicki, MR. (2000). Deficits in compensatory trajectory adjustments after unilateral sensorimotor stroke. Experimental Brain Research. http:Wdx.doi.Org/10.1007/S002219900316 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 2 Abstract A previous study demonstrated that the time-course for amplitude specification of goal- directed aiming movements is similar for individuals with and without a unilateral sensorimotor (SM) area lesion (Velicki, Winstein & Pohl, 2000). However, subjects with a SM lesion performing with the arm ipsilateral to the side of the brain lesion were significantly less accurate than control subjects in an unpredictable condition. The unpredictable condition requires that subjects both formulate an initial plan for movement as well as adjust the response later as additional information about the target (i.e., the goal) is gained (Gordon & Ghez 1987a, 1987b). Gordon and Ghez (1987) demonstrated that premovement planning and compensatory adjustments are the processes contributing largely to accuracy in targeted isometric force responses. They described a statistical model which partitioned response trajectories into the planned and compensatory adjustment components. The purpose of this study was to apply the statistical model to the Velicki, et al. data to determine if the difference in accuracy in those with unilateral stroke was due to a deficit in premovement planning, compensatory adjustments, or a combination of these two factors. We compared the performance of six subjects with unilateral stroke to that of matched control subjects participating in a timed-response movement paradigm. Subjects rapidly flexed or extended the forearm in order to capture a short (20°) or long (45°) target presented in either a fixed (predictable condition) or a random sequence (unpredictable condition). For individuals with stroke the limb used was that which was ipsilateral to the side of the SM lesion. Time to prepare the response was manipulated by varying the time of target presentation relative to an auditory cue for movement initiation. Velocity was derived from the displacement data and multiple regression was used to determine the effect of premovement planning and compensatory adjustments on end-point accuracy. In the predictable condition, premovement planning contributed to final position more for the subjects with stroke [M (SEM) = .50 (.02)] than for the control subjects [M (SEM) = .36 (.03)]. In the unpredictable condition, there were no differences between groups in percent variance due to planning [M (SEM) = .54 (2.1)] for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 the stroke group and [M (SEM) = .45 (2.8) for the control group]. This suggests that the ipsilateral (i.e. intact, undamaged) SM hemisphere significantly participates in the premovement planning of an aiming action. In contrast, for both predictable and unpredictable conditions, compensatory adjustments accounted for a smaller percentage of the variability in final position for the subjects with stroke compared with control subjects [M (SEM) = .09 (2.2) for the stroke group and [M (SEM) = .25 (4.8) for the control group in the unpredictable condition]. Therefore, the less accurate responses for the stroke group can be explained by deficits in the compensatory adjustment component. This suggests a substantial role for SM areas in the preparation and implementation of corrective actions while the effects of the pre-planned action are unfolding. In particular, we discuss the role of the ipsilateral SM areas in relation to parallel feedforward processing in unimanual aiming. Key Words: Motor control, Arm movements, Stroke, Feedforward, Reaction time, Human Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Introduction For over a century, investigators have endeavored to understand how accurate upper extremity aiming movements are performed. Since the early 1800's, the idea that functionally distinct behaviors can be localized in the cerebral cortex has been considered (c.f. Finger, 1994). In a previous study (Velicki et al., 2000), we combined these lines of inquiry and examined the contribution of the sensorimotor areas (SM) to motor control of aiming movements. Six subjects with unilateral sensorimotor area strokes and six age/hand matched control subjects participated in a modified timed-response paradigm (TRP). Subjects with stroke used the upper extremity ipsilateral to the side of the brain lesion to adjust a lever and capture one of four targets presented on a visual screen. Two targets were arranged on either side of a home position such that specification of target position entailed one of two directions (flexion or extension) and one of two amplitudes (short or long). In the TRP, reaction time (RT) is manipulated as the independent variable rather than measured as the dependent variable (Hening et al. 1988). As such, the time course of direction and extent specification of targeted upper extremity movements can be evaluated over varying preparation intervals or imposed RTs. Response accuracy for both subjects with stroke and healthy controls was high and independent of preparation time in a predictable condition in which targets appeared in a fixed (predictable) sequence. In unpredictable trials in which the targets appeared in a random sequence, both stroke and control subjects were equally accurate to the short amplitude targets regardless of preparation interval. For both groups responses to the long amplitude targets increased in accuracy as preparation time increased. These findings suggested that the short target amplitude was programmed in advance as the default and the long target amplitude was specified from the default amplitude as preparation interval increased. However, the control subjects were significantly more accurate to the long amplitude targets than stroke subjects despite the fact that their movement times were faster. The conclusion was that the stroke group was as capable as the control group in specifying the default amplitude in advance of the stimulus, but showed deficits in scaling that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 default response to the long amplitude target. One remaining question is what underlying motor control processes represented in SM areas affected by the stroke contribute to accuracy. The Control of Accurately Aimed Movements Early models of motor control advanced the idea that virtually all aiming movements proceed in a two-stage fashion, with an initial ballistic phase followed by a corrective phase (Brooks, 1979; Meyer et al. 1988; Woodworth, 1899;). The ballistic phase is choreographed by pre-movement instructions to an effector, does not rely on feedback information and error- detection processes, and is presumably regulated by open-loop control processes. In the case of a single-joint movement to a target (e.g. movement of a manipulandum about the elbow joint), the kinematic signature of the pre-planned action is a continuous bell-shaped velocity profile (Brooks, 1979; Latash & Anson 1996; Ojakangas & Ebner 1991) with no overt deflections indicating the presence of corrections (Brooks, 1974). In the corrective phase, feedback is utilized for computation and correction of errors. Thus, accurate target-directed movement is assured as discrepancies between hand and target positions are assessed. While current models of motor control adhere to a two-stage process in the performance of upper limb aiming movements, an important addition has been made. Within the last 25 years, it has become evident that trajectory adjustments may be present even in the absence of overt corrections. Because these corrections occur too soon to be accounted for by peripheral feedback, they are presumably triggered by early or initial trajectory ‘errors’ that are detected by internal monitoring of neural signals (Evarts, 1971; Cooke & Diggles, 1984; Higgins & Angel, 1970; Vicario & Ghez, 1984). Cooke and Diggles (1984), observed very rapid corrections in movement trajectories as healthy subjects performed a visual tracking task. Given the rapidity with which these corrections occurred, it was suggested that they could not be based on information from the moving limb but must have been prepared and implemented while the effects of the pre-planned action were unfolding. While the corrective adjustments observed in these studies occurred at latencies too short after movement onset to be based on feedback Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 6 from the movement itself, the adjustments were overt corrections being made during the course of the movement. In order to investigate whether trajectory adjustments are present even in the absence of overt corrections, Gordon and Ghez (1987a, 1987b) conducted a series of studies examining trajectory control in targeted isometric force responses. In the course of these experiments a statistical method which partitioned the trajectories into the planned and compensatory (i.e. corrective) components was developed and a causal model including both of these factors was tested. It was demonstrated that the variance in final force achieved for targeted isometric force responses is largely determined by two factors: premovement planning and compensatory adjustments. The observation that the very first measurable feature of the response, the initial peak acceleration, (d2F/dt2 ) largely predicted the response, allowed for a quantitative assessment of the contribution of the planned action to the final force achieved. The fraction of the variance of force that can be explained by the peak d2F/dt2 was determined as the squared correlation coefficient (r2 ) of the relationship between peak d2F/dt2 and peak force. It was determined that the calculated fraction accounted for between 70 and 96% of the variance in peak force across individual subjects under two sets of instruction: 'respond as soon as possible' (fast condition) versus 'respond when ready’ (accurate condition) (Gordon & Ghez, 1987a, 1987b). Thus, the mechanism contributing largely to accuracy in a simple isometric response is development of a preplanned trajectory operating through a ‘pulse-height control policy.' First described by Bahill, Clark, and Stark in 1975, pulse-height control enabled subjects to achieve impulses of different sizes by linear scaling of the amplitude of a common waveform. According to pulse height control the amplitude of the responses are varied through scaling of the rate of rise of force to the targets, while the duration of the force rise time remains invariant. The fact that between 4 and 30% of the variance in peak force was not accounted for by the peak d2F/dt2 motivated Gordon and Ghez to ask whether adjustments to the later phase of the trajectory might compensate for errors in the initial trajectory that were evident in the peak d2F/dt2 . In order to determine these sources of the variance in peak force, response trajectories Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 were subjected to a multiple regression analysis (Gordon & Ghez 1987b). Using this statistical approach and the measure of the contribution of the planned trajectory (i.e., r2 ), the overall variance in peak force was partitioned into separate components determined by these two factors: pre-planned pulse height trajectory control and compensatory adjustments. This approach allowed Gordon and Ghez to test whether a causal model that includes both of these factors accounts for significantly more of the total variance in peak force than a model that only included the preplanned trajectory component. The statistical test for corrective components in the response was whether the target amplitude significantly increased the proportion of variance in peak force explained when it was added to the regression equation. Indeed, there was a significant added contribution of target amplitude to peak force which was independent of its indirect effect on peak force through the peak d2F/dt2 . Given that the variance in peak force was largely accounted for by these two components, the Gordon and Ghez model (1987b) provides a valid demonstration that the processes responsible for accurate goal-directed aiming include both a predetermined internal representation of a default trajectory and compensatory adjustments implemented in the interval between peak d2F/dt2 and peak force or amplitude achieved The Sensorimotor Area in Motor Control Gordon and Ghez (1987) thus suggested that rapid responses without manifest discontinuities are controlled by parallel neural processes that concurrently provide for trajectory planning and error correction as the response unfolds. However, the functional neural substrates associated with these parallel 'paths' have yet to be determined. With respect to the compensatory adjustment pathway, numerous reports identify a critical role for SM areas in the execution of ongoing adjustments to the movement trajectory. Studies in which neuronal activity is recorded in several areas of the cerebral cortex as monkeys and humans perform movement- precuing reaction time tasks, demonstrate that activity in the primary motor cortex is largely correlated with movement execution (Alexander & Crutcher, 1990a, 1990b; Richter, Anderson, Georgopoulos, & Kim, 1997; Riehle & Requin, 1989,1993,1995; Riehle et al. 1994; Tanji etal. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 1988; Tanji & Mushiake, 1996; Yoshino, Mikami, & Kubota, 1998 ). In a reaction time reaching task that required precise control of limb posture, the activation pattern observed in motor cortex related to both control of movement and the maintenance of posture. Crammond and Kalaska (1996) thus suggested that motor cortex played a role in the moment to moment control of the motor output. While the studies referenced above primarily recorded from contralateral SM areas, there exists numerous accounts of bilateral SM area activation in the execution of complex tasks (Chen, Gerloff, Hallett, & Cohen, 1997; Gerloff et al. 1998; Kawashima et al. 1998; Kitamura, Shibasaki, Takagi, Nabeshima, & Yamaguchi, 1993; Rao et al. 1993; Shibasaki et al. 1993; Tanji et al. 1988;). As was discussed, the individuals with unilateral SM area damage in our previous study (Velicki et al., 2000) showed deficits in the ability to generate compensatory increases in response parameters while the ability to generate the pre-planned default response appeared to be spared. That this reflects a role for ipsilateral SM areas in generating compensatory adjustments will be revealed by the model analysis. Pumose/HvDotheses The purpose of this investigation is to apply the statistical model of Gordon and Ghez (1987) to our previous data and to determine if the difference in accuracy in those with unilateral stroke was due to a deficit in premovement planning, compensatory adjustments, or a combination of these two factors. If ipsilateral SM areas do not have a substantial role in generating the planned action, than there will be no differences between stroke and control groups in the proportion of variance explained by the preplanned component of the model. The initial movement for both groups will be achieved through a pulse height control policy indicative of a centrally planned default response (Gordon & Ghez, 1987). If ipsilateral SM areas have a role in rapid updating of sensorimotor transformations to adjust aiming responses appropriately, the percentage of variance in displacement explained by compensatory adjustments will be less in stroke than in control subjects for both predictable and unpredictable trials and will not significantly change Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 across preparation intervals. Finally, if ipsilateral SM areas significantly contribute to the compensatory adjustment component, the percentage of variance in displacement explained by compensatory adjustments will be less for individuals with large SM area lesions than for those with smaller SM area lesions. While the idea that the central nervous system internally simulates the dynamic behavior of the motor system in planning and control is useful from a theoretical perspective, the existence and use of such an internal model is still under debate (Blakemore, Goodbody, & Wolpert, 1998; Wolpert, Ghahramani, & Jordan, 1995). It has been suggested that an internal model permits the preparation of predictive default parameters of aimed trajectories (Ghez et al. 1991). It has also been shown that this capability is spared in individuals with ipsilateral SM lesions (Velicki et al. 2000). Support for this position will be strengthened if this analysis reveals similar capabilities to produce default responses for stroke compared to that of control subjects. That the performance of subjects with stroke may reflect deficits in one or both paths of the Gordon and Ghez model, will provide evidence for the role of ipsilateral SM areas in motor control operations subserved by that path(s). Method The data used for analysis was collected in a previous study (Velicki et al. 2000). Since the apparatus, task and training have all been previously described in detail, we now describe these aspects of our methods more briefly. Subjects The data used for analysis was collected from 6 subjects with a unilateral SM area lesion secondary to a stroke involving the anterior circulation system and 6 control subjects without any reported central nervous system pathology. Three subjects with a right-sided lesion and 3 subjects with a left-sided lesion were matched to control subjects by arm used in testing, age, and hand dominance. Thus, subjects in the left stroke (LS) group used their left arm in testing and were matched with left control (LC) subjects using their left arm. Table 1 summarizes the demographic and lesion characteristics for each stroke subject (see Velicki et al., for brain 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. Lesion Characteristics Sub ID Sex Age (yrs) Hand Dominance Etiology Side (CVA) Lesion Size (cm*) Lesion Location 7 M 51 R Infarct R 11.5 Primary motor cortex with subcortical frontal extensions 8 F 53 R Infarct R 50.6 Striatocapsular + Cortex; primary sensorimotor cortical infarct with subcortical extensions; extensions into temporal association area 9 M 71 R Infarct R 83.7 Striatocapsular + Cortex; primary sensorimotor cortical infarct with subcortical extensions; some extension into prefrontal and temporal association areas 10 F 69 R Hemorrhage L 2.2 Posterior parietal cortex only 11 M 62 L Infarct L 64.2 Striatocapsular + Cortex; primary sensorimotor cortical infarct with subcortical extensions; minor extensions into temporal and prefrontal association areas 12 M 70 R Hemorrhage L 51.6 Striatocapsular + Cortex; primary sensorimotor cortical areas with subcortical frontal/parietal extensions; extension into parietal and temporal association areas. 31 section images including the lesion site area of greatest extent). The mean age of subjects was 62.7 ± 8.9 years for the stroke group, and 62.5 ± 8.0 years for the Control group. Instrumentation Data were acquired using an instrumented manipulandum. The manipulandum is a light weight aluminum lever which was mounted on the comer of a wooden table with a near frictionless vertical axle such that the lever was free to move in the horizontal plane above the table surface. A linear potentiometer was attached to the base of the vertical axle such that horizontal movement of the lever produced a change in voltage corresponding to the lever’s angular displacement. This analog signal was sampled at 250 Hz., and converted to digital units by an A-to-D board (Keithly/Metrabyte) installed in a 486-based, 50 MHZ speed IBM-compatible computer. Custom ASYST software (Vanman, 1992) was used for the control of data acquisition, data analysis and storage. Task To perform the task, subjects in the right stroke (RS) and left stroke (LS) groups used the ipsilateral arm and subjects in the RC and LC groups used the same arm as their yoked counterpart. Subjects in the RS and RC groups sat to the left of the lever and grasped the lever handle with their right hand. Subjects in the LS and LC groups used the opposite configuration. When the lever was placed in the home (or start) position the subject was positioned in approximately 45° of shoulder flexion and abduction and 70° of elbow flexion. The task involved moving the lever from the home position to one of four target locations that were differentiated by two direction and amplitude parameters. Subjects either flexed or extended the elbow in either a short arc of motion (2 0 ° from the home position) or a long arc of motion (45° from the home position). The onset of lever movement was to be timed with the last in a series of 4 tones. The computer presented a series of 4 audible tones, each with a progressively higher pitch. The tones each had a 50 ms duration, and were spaced by an inter-tone interval of 500 ms. Subjects faced a computer monitor displaying a graphic representation of the target locations. A yellow Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 arrow pointed to the correct target providing the movement cue. A red pointer provided information regarding the real-time position of the lever and provided feedback about the end position of the movement response. Procedures Training and data collection procedures have previously been described (Velicki etal., 2000). Details of the experimental paradigm are provided only for clarification of the analytic procedures employed. Figure 1 depicts the two independent variables used to reflect the control of preparation time. Movement Cue | 0 .1 M C c 3 * 1 2 I I I Tone I I I 1 I I | I I I I I L i 1 ■ S-T Movement Initiated j S ' * J I I L Written Feedback n rT Cue Presented ’Return to Home’ Figure 1. TRP paradigm demonstrating presentation of movement cue between the third and fourth tone. In this example, the stimulus is given z 300 ms before the 4th tone and is calling for movement to target three. The S-R interval is the time period between stimulus presentation and the beginning of movement. The S-T interval is the time period between stimulus presentation and the 4th tone. The stimulus-tone (S-T) interval was the time between movement cue presentation and the onset of the 4th tone. The movement cue was randomly presented between the third and the fourth tone within one of four preparation time bins (4 = 300-400 ms, 3 = 200-299 ms, 2 = 100-199 ms, and 1 = 0-99 ms). This was the preparation time forced by the paradigm. The stimulus-response (S-R) interval was the time between the movement cue presentation and the onset of movement. Subjects were not able to time their movements exactly on the 4th tone in all trials, so the S-R Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 interval reflected the preparation time actually used by the subject. The timing error was the difference between the imposed S-T interval and the actual S-R interval. A 'block' consisted of 64 trials that met predetermined criteria. For individual trials, movement initiation criteria controlled reaction time such that the onset of lever movement had to occur simultaneously with the 4t h tone. In order to ensure that movement could be initiated in concert with the 4m tone, a timing ‘window’ surrounding the 50 ms tone was adjusted for each subject. Therefore subjects were provided with a buffer of time within which movement of the lever had to be initiated. If this did not occur, the trial was not collected and the subject would receive written feedback on the computer screen indicating that they had initiated movement ahead of (Too early') or after (“ Too late") the 4m tone. Additionally, movement time criteria assured that responses were reflective of a programmed action. Movement time, reflecting the time between the onset and end of movement to the target was also adjusted for individual subjects to continually force subjects to complete their movement quickly. If the duration of a trial was longer than the time allotted by the program, the trial was not accepted and the subject would see Too slow' on the computer screen. Acceptable block criteria assured that for each of the 64 trials in a block, movement onset was not correlated with preparation interval. A significant correlation would occur if over the block of acceptable trials, the subject moved after the tone when preparation time was short, and ahead of the tone when preparation time was long. Eleven acceptable blocks were required of each subject: 3 blocks of a predictable condition in which targets of serial trials were presented in a fixed order (i.e., target 1,2,3,4); and 8 blocks of an unpredictable condition in which the targets of serial trials were presented in a pseudo-randomized order (e.g., target 2,1,4,3) Data Analysis Key kinematic variables were extracted using Datapac III version 1.59 by RUN Technologies Co ®. Raw position data were conditioned using a low-pass filter with a 20 Hz. cut off frequency. Velocity was then derived using a differentiation process. The amplitude value of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 the middle data point in each set of three points is replaced with the slope of the regression line calculated for a minimum interval. The minimum interval, known as the time constant is equal to 2 x the sample period. The sample period for this study was 4 ms. As such, the slope value is obtained over 8 ms. Data Screening Any single peaked velocity waveform within the defined event period (on - off) was considered for analysis (Figure 2a). In order to eliminate from analysis 'pre-movement' waveforms that are potentially related to movement preparation, movement onset was defined as a change in velocity from zero in the appropriate direction of movement. Figure 2b is an example of a velocity profile in which the subject moved the lever toward a flexion target. Onset was determined as a change from zero to beginning negative velocity (i.e., flexion) in order to eliminate from analysis the pre-movement extension (i.e., positive) waveform. As discussed earlier, compensatory adjustments or 'corrections' to the trajectory that are of interest for this analysis are those that are embedded within the initial trajectory and are not obvious corrections. As such, offset was determined as the point in time when velocity returned to zero. Beyond that point any obvious ‘corrections' were not considered part of the event period. Trials in which more than one peak was observed (double responses) in the velocity profile within the movement time interval were counted as follows: 1) Double response trials in which velocity returned to less than 1 0 % of peak velocity prior to the initiation of a second response were accepted for analysis, however, the offset was shifted to the end of the first peak (Fig 2c); 2) Double response trials in which the velocity did not slow to less than 10% of peak velocity prior to the initiation of the second response were rejected (Fig 2d). After the screening procedure, the full complement of analyzed trials consisted of 184- 191 trials/subject in the predictable condition (Total N = 2245) and 474-510 trials/subject in the unpredictable condition (Total N = 5953). Only 3% of both predictable and unpredictable trials were eliminated from analysis based on the above criteria. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 On Off Extension ' 0°/sec» _ Flexion 200 ms B j ^2 0 0 ° /se c / -42°/sec New Offset -445/sec Figure 2. a) Example of an acceptable single peaked velocity profile for a 45° flexion movement, b) 45° flexion movement with ‘premovement’ extension movement related to movement preparation. Onset of movement was defined as a change in velocity from zero in the appropriate direction of movement or beginning negative velocity. Offset was defined as return to zero velocity, c) 45° flexion movement in which the subject initiated a second response prior to zero velocity. This “ double response” trial was accepted for analysis as velocity returned to less than 10% of peak velocity prior to the initiation of the second response, d) 45° flexion movement in which the subject initiated a second response prior to reaching zero velocity. This “ double response" trial was rejected as velocity did not return to less than 1 0 % of peak velocity prior to the initiation of the second response. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 6 Dependent Measures and Statistical Analyses The position and velocity profiles for each trial were used to calculate the following variables (Figure 3): 1) final position achieved [Displacement = offy - ony (a)], 2) average velocity calculated over the time interval from the onset of movement (b) until peak velocity, (c) 3) velocity rise time = time of peak velocity (c) - onset of movement (b), 4) Movement time = off, - on, (d). Off On Position Movement Cue t Target Velocity 4th ^ Movement Cue 45° 2 0 0 °/sec h 2 0 0 m s m Time Figure 3. Representative data from a single trial (45s flexion) showing key kinematic and temporal variables: a = Ony - Offy = Displacement; average velocity (8/sec) from the onset of movement b until peak velocity c; position rise time = c - b (ms) d = Off, - On, = Movement time; e = S-R interval; f = S-T interval. To determine the proportion of variance in final position explained by preplanning and compensatory adjustments to the trajectory, multiple regression analyses were used. Separate regression analyses were performed for predictable and unpredictable conditions in order to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 explore the nature of compensatory adjustments under distinct sets of circumstances. Similar to the ‘when ready' condition of Gordon and Ghez (1987), in the predictable condition the subjects have correct information about target amplitude, but might produce a poorly scaled initial response. In this case, an adjustment would be based on internal or external feedback. Our unpredictable condition parallels the ‘fast’ condition of Gordon and Ghez (1987) in which an adjustment serves to modify an initial ‘guess' by the subject as additional information about the target is gained. This would be considered a feedforward compensatory adjustment since it is based on information about the target and not about errors in response amplitude. By performing separate regression analyses for the two conditions, it will be possible to determine the role of ipsilateral SM areas in these two types of corrective adjustments. In addition, separate regression analyses for flexion and extension responses within like-preparation intervals were performed due to trajectory shape differences for direction. Collapsing across target direction would introduce another source of variance. In summary, separate regression analyses were performed for each of the 4 preparation intervals nested within the two target directions (i.e., flexion and extension), and finally, nested within predictable and unpredictable conditions. Modifications to the Model The statistical model of the determinants of final position that were tested is illustrated in Figure 4. It is a modification of the Gordon and Ghez model given that the original experimental paradigm was adjusted by Velicki et al., (2000) to examine arm movements rather than isometric force pulses. As such, the causal model will test the statistical determinants of amplitude rather than peak force. Using the model illustrated in Figure 4, it will be possible to determine the percentage of variance in final position explained by planning as compared to compensatory adjustments to trajectories. In addition, the first derivative of displacement (i.e., velocity) rather than the second derivative (i.e., acceleration) was chosen as the kinematic index of the 'planned component.’ Specifically, average velocity measured from onset to peak velocity is the kinematic index of the ‘planned component.' In earlier analyses we attempted to use the same kinematic index of planning that had been used in the original Gordon and Ghez model, namely initial peak Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 acceleration. However, in our arm movement, the variability in initial peak acceleration was much greater than that found in the isometric force pulses used by Gordon and Ghez. As such, initial peak acceleration was a poor index of the planned component for our arm movement as it weakly predicted final position. We determined that velocity was the earliest trajectory variable that could be reliably measured. Indeed, three measures of velocity were found to be reliable predictors of final position: peak velocity, average velocity and average velocity measured from onset to peak velocity. Among these measures, there was less within-subject variability for average velocity from onset to peak velocity than for either peak or average velocity. As such, average velocity from onset to peak velocity was used to index planning. Given that this measure is proportionally related to peak acceleration, it corresponds more closely to the measure of planning used by Gordon and Ghez. Figure 4. Illustrates modified statistical model of the determinants of final position that are tested. According to the model, the target can influence the final position through two paths. In the planned trajectory path, the target influences the final position (Y) through its effect on the average to peak velocity (dY/dt). The squared correlation coefficient (r2^,) represents the proportion of the variance in final position that can be explained by variation in average to peak velocity (dY/dt). Therefore, r2 ^, reflects the degree to which final position is explained by premovement planning. The second path represents the corrective influence of the target on final position through the implementation of compensatory adjustments to the trajectory. The additional variance accounted for by the target (through the compensatory adjustment pathway) is equal to the difference between that proportion of the variance accounted for by the combination of average velocity from onset to peak and target and that accounted for by the velocity measure alone (f¥ Y 12 * ^ r» )- Target Planned Trajectory Comptnutory Adjustments to Trajectory Y Final Position Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 According to the model seen in Figure 4, the target can influence the final position through two paths. In the planned trajectory path, the target influences the final position (Y) through its effect on the average to peak velocity (dY/dt). The squared correlation coefficient, r2 , 2 , indicates the extent to which dY/dt itself is determined by the required target amplitude. Then, the squared correlation coefficient describing the relationship between dY/dt and final position (r> , in Figure 4) represents the proportion of the variance in final position that can be explained by variation in average to peak velocity (dY/dt). Therefore, r%, reflects the degree to which final position is explained by premovement planning. The second path represents the corrective influence of the target on final position through the implementation of compensatory adjustments to the trajectory. The independent effect of the target on final position is its additional contribution to the prediction of final position after controlling for the indirect effect of the target on final position through the average to peak velocity measure. The magnitude of the influence of preplanning and the influence of the target that is independent of the premovement plan was computed utilizing step-wise regression analysis. First, determination of the percentage of variance in final position explained by premovement planning yields the following simple regression equation: 1) Y = a + bfX, where Y = final position, X, = average to peak velocity (dY/dt), a = Y-intercept, b1 = is the regression coefficient expressing the change in Y for a given change in X,. The simple squared correlation coefficient (r% f) was computed for this relationship and reflects the proportion of variance in final position explained by the planned component. Next, target was added to the regression, yielding the following equation: 2) Y = a + b,Xt + bj<2 where Y = final position, a = Y-intercept, b1 2 = the regression coefficients for the respective independent variables, Xf = average to peak velocity (dY/dt), X2 = target. The squared multiple correlation coefficient (R2 ^ ) was computed for this relationship and reflects the proportion of variance in final position that can be explained by a linear combination of velocity and the target. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 0 The additional variance accounted for by the target (through the compensatory adjustment pathway) is equal to the difference between that proportion of the variance accounted for by the combination of average velocity from onset to peak and target and that accounted for by the velocity measure alone (R2 y ,2 The statistical test for the influence of compensatory adjustments on the response was if the direct path (Figure 4) from target to final position accounts for a statistically significant increment in the proportion of variance in final position (i.e., R2 * 1 2 -r^,). Group Analyses The difference between the control and stroke groups in the proportion of variance in final position explained by preplanning and compensatory adjustments was determined by a group (control, stroke) x preparation, or S-R intervals (4=300-400, 3=200-299,2=100-199,1=0- 99) mixed model Analysis of Variance (ANOVA). The dependent variables used in this analysis were the regression coefficients which represent percentage of variance explained by preplanning (rv,), and compensatory adjustments due to target (R*y ,2 -r2 /,). To test whether subjects were using a ‘pulse height’ control policy in generating the planned action, rise time invariance was tested for individual subjects for both predictable and unpredictable conditions. An earlier analysis revealed that there was no difference in rise time between 20° flexion and extension targets or 45° flexion and extension targets for each group. As such, the two short and the two long amplitude targets were combined for a one-way analysis of variance with target amplitude (short, long) the factor. The null hypothesis for this analysis, indicative of a pulse-height control policy, was that there would be no significant difference in rise time to the short (20°) and long (45°) target amplitudes. Stroke Group Analysis: Lame Versus Small Sm Cortical Lesion In order to determine whether lesion size was linearly related to the percentage of variance in final position due to planning and compensatory adjustments, simple linear regression was performed. The percentage of variance in final position due to premovement planning and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 the percentage of variance due to compensatory adjustments were separate dependent measures with lesion volume as the independent measure. For all statistical tests, significance was set at p < .05. A Students’s independent samples f-test was used for post-hoc comparisons to determine the locus of any significant interaction effects. Results Since the purpose of this study was to determine if differences in accuracy between control subjects and those with unilateral stroke was due to a deficit in premovement planning, compensatory adjustments, or a combination of these two factors, accuracy data from the initial study is presented for both predictable and unpredictable conditions. Group differences for proportion of variance in final position due to both paths of the model (i.e., premovement planning and compensatory adjustments) for the predictable condition are presented first, followed by the unpredictable condition. Rise time invariance is presented to reveal whether subjects used a pulse-height control policy to generate the planned component of the action. Finally, a within stroke group analysis of the effect of lesion size on the proportion of final-position-variance due to the separate components of the model is presented. Use of the statistical model as a valid framework for comparing group differences in premovement planning and compensatory adjustment capabilities was demonstrated empirically by the high percentage of variance in final position explained by the combined paths of the model. Across conditions, the proportion of variance in displacement explained by both premovement planning and comoensatory adjustments ranged from 26 to 85% for all subjects. Individual subjects in the stroke group demonstrated total variances as high as 93% (range = 23 to 93%). For subjects in the control group, the combination of both paths of the model accounted for 46 to 96% of the explanation of final position. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 2 Predictable Condition Accuracy In the predictable condition there were no differences in accuracy (measured as absolute error, AE) between stroke and control groups (Table 2.)- The stroke group demonstrated slightly higher extent error (-3 °) than the control group across preparation intervals, however, this difference was not reliable for response accuracy, (P = .12). Table 2. AE for Predictable and Unpredictable Conditions Predictablef Unpredictablef Stroke 8.5 ±1.9 11.4 ± .7 Control 5.8 ± .5 7.1 ± .4 fMean AE and SEM (in degrees) calculated across 4 preparation intervals and short and long amplitude targets. Premovement Planning The proportion of variance in displacement due to the preplanned trajectory was significantly greater for subjects with stroke [M (SEM) = .50 (.02)] than controls [M (SEM) = .36 (.03)], P < .01, (Figure 5, dark stack bar) in the predictable condition. Compensatory Adjustments The percentage of variance in displacement explained with the addition of target amplitude was significantly less for subjects with stroke [M (SEM) = .14 (.04)] than for control subjects [M (SEM) = .37 (.04)] P < .0001), (Figure 5, white stack bar). While the difference in total response accuracy between groups in the predictable condition was not reliable, the enhanced ability of the controls to make compensatory adjustments may have afforded them the slight advantage they demonstrated in accuracy. However, all subjects in the stroke group demonstrated the greatest percentage of variance due to adjustments in the predictable condition compared to the unpredictable condition (see below) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 3 suggesting that they were capable of using target information in advance in order to adjust their response. ■ Plan 9 ° r “ [ | Companaalory Adiustmanta C ontra ! Strata Figure 5. Stack bar plot of percent variance in displacement for predictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left and the stroke group is to the right. Data are averaged across 6 subjects/group and 4 targets. SEM error bars are shown for each component of the stack bar plot. Unpredictable Condition Accuracy Absolute error to the 45° targets (long extent) for the Stroke group was greater [M (SEM) = 15.4°(2.4)] than for the Control group [M (SEM) - 8.8°(1.5)], whereas there was no difference between groups in accuracy to the short targets (Velicki, et al., 2000). This finding suggests that subjects in the stroke group were as capable as control subjects of specifying the default extent in advance of the stimulus, but had difficulty adjusting that default response to the long extent within the imposed preparation interval. The relevance of the accuracy differences between the groups is enhanced by the fact that the control subjects were significantly faster than the subjects with stroke [M (SEM) = 318 ms (1.7)] for the stroke group [M (SEM) = 217 ms (1.3) for the control group; p < .0001] (Table 3). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 4 Premovement Planning Figure 6a shows that the stroke group demonstrated overall greater percentage of variance in displacement due to premovement planning than controls (dark stack bars). However, this was not a reliable group difference (p = .5). Table 3. Average MT and Rise Time Predictable Condition Movement Time f (ms) Rise Time f (ms) Stroke - Short 297 ±10.1 121 ±4.4 Stroke • Long 332 ±10.4 132 ±4.5 Control - Short 201 ± 7.3 86 ± 2.9 Control - Long 234 ± 7.4 99 ± 3.0 Unpredictable Condition Stroke - Short 301 ± 6.8 121 ±3.1 Stroke* Long 336 ± 7.2 133 ±3.3 Control • Short 198 ±4.3 83 ±1.4 Control • Long 236 ±4.9 96 ±1.7 * Mean and SEM calculated across 4 preparation intervals Compensatory Adjustments Control subjects showed a significantly greater percentage of variance related to adjustments than did subjects with stroke (Figure 6a - white stack bars). This difference was reliable as evidenced by a significant main effect of group in the proportion of variance in displacement due to compensatory adjustments (F(1,24) = 18.8, p =.0001). Additionally, the proportion of variance in displacement explained by the addition of target amplitude increased for both groups as preparation interval increased (Figure 6b). There was a significant main effect of preparation interval on the percentage of variance in final position due to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 5 compensatory adjustments (F (3,24) = 7.9, p = .0001). Thus, when the target was presented progressively earlier with respect to the 4t h tone, both groups of subjects could utilize target information to scale their responses. Post hoc analysis revealed that the proportion of variance due to compensatory adjustments in the shortest preparation interval (0-99 ms.) was significantly different than the two longest preparation intervals (i.e., 200-299 ms., and 300-400 ms). With the exception of the shortest preparation interval, the control group demonstrated greater adjustments across preparation intervals compared with the stroke group. However, there was not a significant group by preparation interval interaction (p = .13) most likely due to the variability observed in the control group responses (Figure 6 c ).1 Plan Plus Adjustments: Contribution of the two paths of the model The data previously presented are depicted in terms of the combined contribution of the two paths of the model (i.e., plan + adjustments) to more closely examine the behavioral differences between the stroke and control groups. Figure 6d illustrates that the percentage of variance in displacement explained by both premovement planning and compensatory adjustments was greater than that due to planning alone for both groups. A far greater increase for the control compared to the stroke subjects in the explanation of final position via the compensatory adjustment pathway is clearly depicted. 'Because subjects synchronize movement initiation with an auditory cue regardless of preparation time (i.e., target cue presentation), inaccuracy takes the form of both amplitude and/or direction errors. Of all trials analyzed 34% were responses to the wrong direction (N = 5953 total trials; n = 1517 wrong direction trials). The model analysis was employed in this study to provide insight into the processes responsible for enhanced accuracy for the control compared to the stroke group. As such, detailed discussion of wrong direction responses is not presented. However, it is of interest to note that the two groups demonstrated remarkably similar behavior when the response was made in the wrong direction, i.e., neither group showed an appreciable increase in proportion of variance due to the addition of target amplitude. This suggests that when direction was incorrectly specified, adjustment of the default extent was aborted. (Velicki, et al. 2000). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Control Strok* 0-99 100-199 200-299 300-400 S-R Interval Control Control 40 3 0 30 25 20 15 to 5 * 0-99 100-199 200-299 300-400 s 3 * S-R Interval 75 Stroke 66 61 54 47 40 Plan Fioure 6. a) Stack bar plot of percent variance in displacement for unpredictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left and the stroke group is to the right. Data are averaged across 6 subjects/group and 4 targets. SEM error bars are shown for each component of the stack bar plot, b) Bar graph representing percent variance in displacement due to compensatory adjustments. Data were averaged across groups, and response direction accuracy (correct, wrong). The shortest preparation interval (0-99 ms) was significantly different from the two longest intervals (200-299 and 300-400 ms); p = .001 and .006 respectively, c) percent variance in final position explained by compensatory adjustments for each group as a function of preparation interval. Control group = squares; stroke group = diamond. Group x preparation interval interaction, (p = .13). Note the relatively high SEM for the control group, d) The total variance (i.e., that due to both planning and compensatory adjustments) in the unpredictable condition is compared to the percent variance in final position due to planning alone for both stroke (diamond) and control (square) groups. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 Detailed in the following paragraph and demonstrated in Figure 7, is the general observation that the pattem(s) of change for both individual stroke and control subjects in percentage of variance due to planning and compensatory adjustments over preparation intervals was similar to the pattem(s) observed for each of their respective groups. The variances for the control subject in Figure 7 (subject #2) parallels the group performance in that the least percentage of variance due to compensatory adjustments was in the shortest preparation interval with progressively increasing explanation of final position due to compensatory adjustments as time to prepare increased. While the variances for control subject #5 (Fig. 7b), also demonstrated the same pattern of change over preparation intervals as the group, this individual subject demonstrated a greater increase in proportion of variance and contribution from compensatory adjustments across the earliest three preparation intervals compared to the group. An explanation for this difference was found in the timing criteria data for this subject. Recall that subjects were trained to initiate movement within a pre-determined amount of time (the timing window) surrounding the onset of the 4m tone. Of all the control subjects, the subject in Figure 7 had the longest timing window (± 50 ms. before and after the 4t h tone). Therefore, even in the shortest preparation interval in which target information is presented anywhere from 0 to 99 ms. prior to the onset of movement, this subject effectively ‘bought’ herself more time by delaying her response until after the tone but before the end of the acceptable timing window. In this way, the actual target could influence her response more than for others even when presented late relative to the 4t h tone. Rise Time Analysis: Pulse-Heioht Control The null hypothesis for whether or not the planned action was governed by a pulse- height control policy is that there would be no difference in rise time (i.e., time of peak velocity • onset time) between the long and short extent targets. Earlier analysis revealed no differences in rise time between flexion and extension targets and as such responses across target direction were combined. Therefore, the rise time analyses were performed for each individual subject as a 2 target amplitude (short/long) ANOVA within each condition (predictable/unpredictable). For Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 8 the purpose of the question at hand, the most important result was that no individual subject from either the stroke or the control group demonstrated rise time invariance in either condition. The time to peak velocity was significantly different for the short compared to long target amplitudes for all subjects (Table 3). This suggests that a pulse-height control policy did not govern the preplanned component of the response. That rise time invariance was not demonstrated in these data will be addressed in the discussion. a > E 8 a s 8 .1 o'* Control-Plan | | Control-Adiust 100 80 60 40 20 Sub2-P1an frd X Sub2-Adjust Sub5-Plan £ 2 Sub5-Adjust 0-99 100-199 300-298 300400 0-99 100-1 9 * 300-299 300400 S-R Interval 0-99 100-199 200-299 300-400 Figure 7. The left four stack bars are the percent variance due to planning and compensatory adjustments for the control group across preparation intervals. Depicted is the composite variance in final position consisting of that due to planning and compensatory adjustments for correct direction responses to flexion targets. The middle four stack bars represent data from one control subject (subject #2). Data from a second control subject (subject #5) is presented in the right four stack bars. Note a greater increase in proportion of variance due to compensatory adjustments across the earliest three preparation intervals for control subject #5 compared to the group. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Relation of lesion volume to variance in final position In order to determine whether the size of the CNS lesion was related to the percentage of variance in final position that was explained by planning and/or compensatory adjustments, individual stroke subject variances were subjected to simple linear regression. Two simple linear regression calculations with lesion volume as the independent measure were computed: The percent variance in final position due to premovement planning as one dependent measure and the percent variance in final position due to compensatory adjustments as the second dependent measure. There was a significant inverse relationship between lesion volume and percent variance due to premovement planning. Lesion volume accounted for 54% of the variance due to planning. Thus, the smaller the ipsilateral SM lesion the higher proportion of variance in final position is due to premovement planning. However, a categorization scheme for lesion location revealed that lesion location was predictive of ipsilateral limb aiming movement deficits (Winstein, Onla-or, Rose, Sullivan, & Pohl, submitted). This was particularly true for individuals with stroke categorized with a lesion in the striatocapsular plus cortex area. As such, it is unlikely that the above effect is due entirely to lesion volume alone. To control for the effect of lesion location, the regression was recalculated with data from only the four subjects with striatocapsular plus cortical lesions (Table 1). Lesion volume still accounted for 40% of the variance due to planning in this subset of subjects with a similar lesion location. While the lesion volume analysis did not reveal a significant relationship between the extent of the ipsilateral SM lesion and the proportion of variance in displacement due to compensatory adjustments, inspection of individual subject variances within the stroke group revealed that the two subjects with the smallest lesions (subject #10, lesion volume = 2.2 cm3 and subject #7, lesion volume = 11.5 cm3) did show greater percentage of variance in displacement due to compensatory adjustments than the other four stroke subjects with significantly larger lesions (e.g., subject #11, lesion volume = 64.2 cm3 (Figure 8). While subjects #10 and #7 had the smallest lesions they were also the only two without striatocapsular plus cortical lesions. As Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 such it is impossible to rule out lesion location as contributing to the spared capability of these two subjects to perform compensatory adjustments. < D 1 8 0 s * m Stroke-Plan | | Stroke-Adjust 100 80 60 40 20 0-99 100-199 200-299 300-400 SublO-Plan Subl 0-Adjust 0-99 10&-199 200299 300-400 S-R Interval Sub11-Plan X/A Sub11-Adjust V s 0-99 100-199 200-299 300400 Figure 8. Comparison of composite variance due to planning and compensatory adjustments between two unilateral stroke subjects. The left four stack bars represent the stroke group data. Subject #10 had the smallest SM lesion = 2.2 cm3 ; Subject #11 had the second largest SM lesion of the stroke group = 64.2 cm3. Note greater percentage of variance explained by compensatory adjustments across preparation intervals for subject #10 compared with subject #11. Discussion The modified statistical model of Gordon and Ghez (1987), allowed for an examination of motor control processes that may account for the differences in accuracy between individuals with SM lesions and healthy age-matched control subjects observed in our previous study (Velicki et al., 2000). Indeed, the components of the statistical model were previously found to be highly predictive of peak force achieved during targeted isometric force responses (Gordon & Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 Ghez, 1987). Now, we have demonstrated a valid application of this model to the study of arm movements and the underlying motor control processes governing those actions in persons with unilateral brain damage. The one exception with respect to parallel assumptions underlying the original model and the statistical model applied here is that it does not appear that the ‘planned* component was governed by a pulse-height control strategy for either the stroke or the control group. Therefore, subjects were not achieving different amplitudes by modulating the rate of rise of velocity while maintaining rise time constant. While our results were not dependent upon a pulse-height strategy underlying the planned component of the aiming action, the issue of control strategy is interesting with respect to motor control. As such, after discussing the results of the model application in relation to our hypotheses, this issue of control strategy will be addressed. Displacement was more strongly predicted by the preplanned component for subjects with ipsilateral sensorimotor area damage than non-lesioned controls in both predictable and unpredictable conditions. This suggests that an ipsilateral SM hemispheric lesion does not affect the ability to program an aiming action. The present analysis revealed a negative correlation between lesion size and the proportion of variance in displacement explained by premovement planning, suggesting that there may be an upper limit in terms of the amount of SM damage and planning capability. However, while the response more closely reflects the ‘plan* for the subjects with stroke compared with controls, the reduced accuracy suggests that the plan may be suboptimal (Winstein, Merian, & Sullivan, 1999). Thus, the presence of a lesion itself influences planning dependability. The poorer performance of the stroke subjects using the hand ipsilateral to the lesion might occur because normally there is bilateral activation in SM areas in the planning of complex tasks (Kitamura et al., 1993; Shibasaki et al., 1993; Winstein, Grafton, & Pohl, 1997). Earlier it was discussed that primary motor cortex may play a more crucial role in the ongoing modifications to the trajectory as opposed to the pre-movement planning of the action. Contralateral motor cortex inactivation in cats performing a tracking task with their forearm led to a deficit in the animal’s ability to scale the size of the response to an expected range of targets, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 but did not affect response initiation (Martin & Ghez, 1988). Additionally, preparatory neuronal activity in monkeys (Smith et al., 1993) and in humans (Macefield & Johansson, 1994) responding to perturbations of a hand-held object was relatively absent in the motor cortex. The activity of the neurons in the motor cortical region was related to muscles active during grasping. Smith et al., (1993), concluded that the motor cortex may have more involvement with compensatory increases in grip force or regulation of the already initiated response after perturbations than with preparing appropriate anticipatory strategies. However, it has been observed in several studies that have recorded the activity of single cells in the brain of behaving primates, that the activity in motor cortex is increased during the preparatory period prior to movement as well. A larger percentage of neurons showing preparation-related properties were observed in premotor cortex (PM) and supplementary motor area (SMA) than in primary motor cortex (M1), whereas a larger percentage of neurons in M1 demonstrate execution-related properties (Alexander & Crutcher, 1990; Riehle & Requin 1993; Riehle et al. 1994; Weinrich & Wise, 1982). The neuroanatomic correlates of motor preparation and execution in humans parallel the results of the primate studies. In several functional imaging studies using positron emission tomography, increases in regional cerebral blood flow has substantiated the participation of PM, SMA, parietal areas and M1 in the preparation for movement (Deiber, Ibanez, Sadato, & Hallett, 1996; Jahanshahi et al., 1995; Kawashima, Roland, O’Sullivan, 1994; Stephan et al., 1995;). However, the results of these studies suggest that the primary sensorimotor cortex has its major role in movement execution (Stephan et al., 1995). Schluter, Rushworth, Mills and Passingham (1999), studied the signal- set- and movement related activity in the human premotor and motor cortex by temporarily disrupting activation in those areas using transcranial magnetic stimulation (TMS). The results suggested set or preparatory activity in both premotor and motor cortices. However, TMS pulses over M1 were most disruptive when they were delivered around the time of response execution. So while there is evidence to suggest that motor cortex activation is increased relative to the planning of an action, there is clearly a more Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 substantial role for M1 in formulating compensatory changes of task parameters during movement execution. The above accounts of movement parameter modulation in SM areas during the movement corresponds to the adjustment pathway of the Gordon and Ghez model. As such, it was hypothesized that the individuals with stroke would demonstrate deficits primarily in producing compensatory adjustments. For subjects with stroke, compensatory adjustments were demonstrated by an increased percentage of variance in displacement due to the addition of target amplitude. However, compared to control subjects, the subjects with stroke were less proficient in updating the default response, as indexed by significantly less percent variance explained by compensatory adjustments in both predictable and unpredictable conditions. Earlier (see dependent measures and statistical analyses) we defined two types of corrective adjustments that could occur under these distinct conditions. In the unpredictable condition, the adjustment would be a feedforward compensatory adjustment as subjects modify their initial plan or best 'guess’ when target information becomes available. In the predictable condition, target information is known in advance but an error in response amplitude might be detected via internal feedback. As such, a feedback correction is implemented in order to appropriately scale the initial response. The results from our study suggest that ipsilateral SM areas contribute to both the feedforward and feedback processing of movement parameterization even for unimanual actions. However, each individual stroke subject was most proficient at generating compensatory adjustments in the predictable condition in which no differences in accuracy were found. Subjects with unilateral SM damage performing with the ipsilateral limb may be capable of generating internal feedback-based adjustments if this control process requires only contralateral cortical activation. It is well established that the planning and implementation of sufficiently complex tasks elicits bilateral neural activation patterns in various cortical areas (Culham et al., 1998; Gordon, Lee, Flament, Ugurbil, & Ebner, 1998; Winstein et al., 1997). One method for increasing task complexity is to manipulate the uncertainty of upcoming responses. A bilateral Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 neural activation pattern in parietal and frontal areas was observed when subjects moved a joystick under unpredictable conditions (Deiber et al., 1991). However, this bilateral pattern was not seen under conditions in which responses were made under predictable conditions. Even in the predictable condition, subjects performing in a TRP must contend with synchronizing the onset of their movement with an auditory cue. However, given the fact that target information is known by subjects ahead of the cue, it is unlikely that the task is sufficiently complex to require bilateral neural activation. Therefore, subjects with stroke can rely on the intact hemisphere for performance of predictable trials presumably utilizing internal feedback to compensate for an inappropriately scaled initial response. All the same, our data suggests that there is some spared capability to generate compensatory adjustments, as would be required in the unpredictable trials, in individuals with unilateral SM area damage. In the unpredictable condition, for both the stroke and control groups, the proportion of variance in displacement explained by compensatory adjustments significantly increased as preparation time increased. This suggests that the process of updating the default trajectory occurs as soon as response information is available and is not affected by unilateral sensorimotor area damage. However, while not shown to be a significant group difference, (p = .13) the control group responses appeared to constitute the majority of the compensatory adjustments that was seen (see figure 6 c ). Earlier it was stated that the planned component of the aiming action was not governed by a pulse height control strategy for either the stroke or the control group. Gottlieb, Corcos, and Agarwal (1989) proposed that accurate, single-joint movements are planned according to strategies of which there are at least two: a speed-insensitive (SI) and a speed-sensitive (SS) strategy. The choice between SI and SS depends on whether movement speed and/or movement time must be constrained to meet task requirements. Over a series of studies, the tasks, instructions, and variables which influence choice of strategy were determined (Corcos, Gottlieb, & Agarwal, 1989; Gottlieb et al. 1989; Gottlieb, Corcos, Agarwal, & Latash, 1990; Gottlieb, Chen, & Corcos, 1996; Pfann, Hoffman, Gottlieb, Strick, & Corcos, 1998). The Speed- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 insensitive strategy is one that subjects use to move to a visually defined target of fixed size over different distances under instructions to be accurate and fast. The term insensitive indicates that the control pattern is not specified according to either the speed or the movement time (Gottlieb et al., 1989). A speed-sensitive strategy is used when the subject exerts control over the speed at which the movement must be performed or over the movement time (Corcos et al., 1989). The regulation of movement speed occurs through modulation of pulse amplitude and as such movements of constant duration but varying distances are achieved through scaling the rate of rise of both the second and third derivates of displacement. Essentially, the speed-sensitive and pulse-height strategies are synonymous. In the TRP, subjects contend with several competing task constraints. They must move at a specified time whether they are ‘ready* to or not, they must try to be accurate, and they must complete the movement within a movement time criteria. The speed, however, at which to complete a 20° or a 45°movement (in this case) is not directly regulated. As such, the subjects performing the TRP are unlikely to invoke a pulse-height or speed-sensitive strategy to govern the initial impulse. Movements to different distances without explicit speed regulation are more likely to be controlled by the SI strategy which modulates pulse width. The characteristic feature of a pulse-width strategy is that the rate of increase in velocity and acceleration is constant as subjects perform movements of varying amplitudes, but movement duration is scaled to capture targets at different distances (Gottlieb et al., 1989). Two predictions of a pulse-width strategy that were tested a posteriori with these TRP data were: 1) that there would be no difference in velocity between the two target amplitudes and 2) that movement time and displacement would covary. Analysis of variance of the kinematic index of planning, namely average velocity from onset to peak velocity, revealed significant differences in the velocity measure between the short and long target amplitudes. Therefore, invariance of movement velocity was not found. However, for all subjects within each condition, both flexion and extension targets, and across preparation intervals significant correlations between movement time and displacement were found (r ranged from .26-.55). Therefore, only Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 the correlation analysis, weakly supported a pulse-w/cff/j strategy governing the preplanned component of the movement. In the unpredictable condition, subjects generated a default response close to that required for the small amplitude target. When the response actually required the long target, subjects would prolong the duration of their response to reach the required amplitude. In this respect, the implementation of compensatory adjustments produced apparent pulse-width control. That the absence of pulse-height control in this study was not merely a by-product of compensatory adjustments was demonstrated by the fact that rise time invariance was not found in the predictable condition. In the predictable condition the control strategy should not reflect feedforward adjustments based on evolving information about target amplitude. Instead, when subjects are able to or are instructed to maximize accuracy, rise time invariance will be most pronounced by producing responses with more slowly rising trajectories (Gordon & Ghez, 1987). While it is not possible to conclude that the subjects in this study used the same control strategy for both predictable and unpredictable conditions, neither condition elicited a pulse-height control strategy nor were there differences in rise time between predictable and unpredictable conditions (see Table 3). In discussing, the improbable existence of general limb control rules applicable to more than one experimental paradigm, Gottlieb et al., (1996) specifically stated that the two strategies (i.e., SI and SS) are not sufficient to describe how all single-joint flexion movements are controlled, particularly those performed under certain temporal constraints. The hallmark of the TRP is the temporal constraint of response initiation. Thus, while the task constraints of the TRP did not invoke a strict pulse-height control strategy, it was found that up to 96% of the variance in final position was determined by the two components of the model: premovement planning and compensatory adjustments. Using the statistical model, it was identified that both the planning of the action and the ability to generate compensatory adjustments were different for individuals with ipsilateral SM lesions compared with control subjects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 Summary It has been suggested that an internal model permits the preparation of predictive default parameters of aimed trajectories (Ghez et al. 1991). It has been demonstrated that this capability is spared in individuals with unilateral SM area lesions. The deficits observed appear to lie primarily in the compensatory adjustment pathway of the model. This suggests a substantial role of the ipsilateral SM areas in the preparation and implementation of corrective actions while the effects of the pre-planned action are unfolding, i.e., parallel feedforward processing. A deficit in feedforward control may create a compensatory increase in the degree to which subjects with stroke rely on their previously computed plan. Under predictable conditions this appears to be an acceptable and relatively effective strategy. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 CHAPTER 3 Experiment 2: INCREASED TASK COMPLEXITY OF AIMED ARM MOVEMENTS AFFECTS PLANNING AND UPDATING OF RESPONSE PARAMETERS FOR INDIVIDUALS WITH UNILATERAL SENSORIMOTOR DAMAGE Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Abstract Accuracy depends on how appropriately response parameters are planned or specified in advance and how precisely those parameters are adjusted with updated target information. To determine the role of sensorimotor (SM) areas in planning and updating goal-directed aiming, we used an aiming task of greater complexity than has previously been used. We compared the performance of six subjects with unilateral stroke to that of matched control subjects under conditions of a timed-response movement paradigm. Subjects rapidly flexed or extended the forearm in order to capture a short (15°), medium (30°), or long (45°) target presented in either a fixed (predictable condition) or a random sequence (unpredictable condition). Subjects with stroke used the limb ipsilateral to the side of the SM lesion. Time to prepare the response was manipulated by varying the time of target presentation relative to a start cue. Velocity was derived from displacement data and multiple regression was used to determine the effect of premovement planning and trajectory updating on end-point accuracy. Consistent with earlier results for a 4-target task, as preparation time increased, both stroke and control groups improved in accuracy in the unpredictable condition, and the stroke group was less accurate than the control group particularly to the long target amplitude. However, for this 6-target task, both planning and updating deficits accounted for the accuracy differences between groups, not just updating as with the 4-target task (Fisher, Winstein, & Velicki, 2000). However, within the stroke group two different strategies to contend with the increased complexity of planning and updating responses for the 6-target task were identified. The total percent variances between these two ‘subgroups' within the stroke group were nearly identical at - 73%. However, the allocation of the total into both aspects of accurate performance (i.e., response planning and updating) were in opposite directions. The percentage of variance in final position explained by the initial plan was high and similar to that of the control group for the 3 ‘low-adjustment’ subjects of the stroke group. In contrast, the percent variance explained by the ability to update the response was significantly lower for the low-adjustment stroke sub-group compared with controls. Absolute error was high for the 3 subjects within the low-adjustment stroke sub-group. A deficit in planning Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 capability for the 3 ‘high-adjustment’ stroke subjects was identified as significantly less variance in final position due to planning compared with controls. In contrast to our previous data (Fisher et al., 2000) a significantly greater percent variance in displacement due to compensatory adjustments was demonstrated by the high-adjustment sub-group compared with controls. Absolute error was low for the high-adjustment stroke subjects and similar to that of the control group. Therefore, a spared capability to generate feedforward compensatory adjustments appeared to largely compensate for a suboptimal plan and improve performance. These results suggest a substantial role for SM areas in the planning and updating of sufficiently complex goal- directed aiming tasks. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 INTRODUCTION Accurate goal-directed aiming actions depend on how appropriately response parameters (i.e., target direction and amplitude) are planned or specified in advance as well as how precisely those parameters are adjusted with updated target information. Goal-directed aiming movements of subjects with unilateral cerebral damage were compared with those of healthy age-matched control subjects to examine the role of SM areas in both aspects of accurate performance (i.e., response parameter planning and updating). Subjects with stroke- related unilateral cerebral damage moved with the arm ipsilateral to the lesion. Therefore, deficits in response parameter planning and/or updating could be used to infer the contribution of the damaged SM area to these motor control processes We used the timed-response paradigm (TRP) to investigate the time-course over which target parameters are specified during the preparation interval prior to the initiation of movement (i.e., the parameter specification process). In addition, we applied a modified statistical model developed by Gordon and Ghez (1987b) to determine the extent to which the movement end point could be explained by the ability to preplan the action and/or update the response (i.e., perform feedforward compensatory adjustments) under variable preparation times. By partitioning response trajectories into the planned and adjustment components, the model analysis allows insight into the central processes contributing to accuracy. Several neurophysiological studies have implicated SM areas in both movement parameter specification as well as parameter updating. Single cell recordings in the brain of behaving primates reveals increased activity in SM areas during the preparatory period prior to movement. These preparation-related neurons have been observed in primary motor cortex (M1), premotor cortex (PM) and supplementary motor area (SMA) (Alexander & Crutcher, 1990; Riehle & Requin, 1993; Riehle et a)., 1994; Weinrich & Wise, 1982). Observed cell activity modulation is correlated with distinct parameters of the forthcoming and/or ongoing movement as the position of a pointer is changed to match a target display. Thus, changes in neuronal firing have been correlated with movement direction (Fu etal., 1993b; Georgopoulos, 1994a, 1994b, 1994c; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 Georgopoulos, Caminiti, Kalaska, & Massey, 1983a; Georgopoulos, Kettner, & Schwartz, 1988; Georgopoulos, Crutcher, & Schwartz, 1989; Georgopoulos, Ashe, Smymis, & Taira, 1992; Tanji & Evarts, 1976), movement amplitude (Fu et al. 1993b, Fu et ai., 1995) and force (Kalaska, Cohen, Hyde, & Prud'homme, 1989; Kalaska & Crammond, 1992; Georgopoulos etal., 1992; Riehle & Requin, 1995). Additionally, numerous reports identify a critical role for SM areas in the execution of ongoing adjustments to the movement trajectory. In particular it has been suggested that M1 may play a more crucial role in the ongoing modifications to the trajectory during movement execution than in the pre-movement planning of the action (Alexander & Crutcher, 1990; Richter et al. 1997; Riehle & Requin, 1989,1993,1995; Riehle et al. 1994; Tanji et al. 1988; Tanji & Mushiake 1996; Yoshino et al. 1998). Fu and colleagues (1995) observed that amplitude modulation in M1 cells occurs after the time of peak velocity, and corresponds to updating the response. While the studies referenced above primarily recorded from contralateral SM areas, there exists numerous accounts of bilateral SM area activation in the execution of more complex aiming tasks (Chen et al., 1997; Gerioff et al., 1998; Kawahima et al., 1998; Kitamura et al., 1993; Rao et al., 1993; Tanji et a!., 1988; Shibasaki et al., 1993;). Thus, the contribution of the damaged hemisphere to the control underlying accuracy can be revealed as individuals with unilateral SM area damage perform goal-directed aiming actions with the ipsilateral arm (Haaland, Harrington, & Yeo, 1987; Winstein & Pohl, 1995; Winstein et al., 1999; Velicki et al., 2000) Previously we reported the contribution of ipsilateral SM areas to accurate specification of response parameters for aiming movements (Velicki et al., 2000). With two targets aligned on either side of a home position, accuracy of aiming movements required selection of both direction (flexion or extension) and amplitude (short or long). Time to prepare the response was manipulated by varying the time of target presentation relative to an auditory cue for movement initiation. In an unpredictable condition in which each of the four targets were presented in random order, the responses of both stroke and control subjects to the short targets were Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 relatively accurate regardless of preparation time. In contrast, absolute error (AE) to the long amplitude targets decreased for both groups as preparation time increased, but was significantly greater for the stroke than for the control group. For both groups then, the short target amplitude was planned in advance as the default and the long target amplitude was specified from the default amplitude. These results suggested that subjects with stroke-related brain damage were as capable as control subjects in specifying the default amplitude in advance of the stimulus but had difficulty updating that default response to achieve a similar degree of accuracy as control subjects (Velicki et al., 2000). Thus, it appeared as if the capability to plan or specify the initial response was spared in individuals with unilateral SM area lesions and the less accurate responses of the stroke group was due to deficient response updating. This hypothesis was supported by the results of a second study in which aiming response trajectories from Velicki et al., (2000) were subjected to a multiple regression analysis (Fisher et al, 2000) using the modified statistical model (Gordon & Ghez, 1987b). The variance in final position of the aiming response was partitioned into that attributed to premovement planning and compensatory adjustments. In the unpredictable condition, there were no differences between groups in explanation of final position due to planning. In contrast, compensatory adjustments accounted for a smaller percentage of the variance in final position for the subjects with stroke compared with control subjects. Therefore, the less accurate responses for the stroke group in the Velicki et al study (2000) could be explained by deficits in updating the default plan. This suggests a substantial role for ipsilateral SM areas in the preparation and implementation of corrective actions while the effects of the pre planned action are unfolding. While the results of the two studies provide insight into the role of the SM areas in response parameter planning and updating, discrepancies between the findings of Velicki et al., (2000) and the results of the original TRP studies of Ghez and colleagues raise new issues. In Velicki et al., (2000) subjects contended with specifying and updating response parameters in a four-choice aiming task (two-direction, two-target amplitude). In the original TRP studies, young Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 healthy subjects produced isometric force pulses in a six-choice force task (three target amplitudes in two directions) to one of three target amplitudes in two directions (Favilla et al., 1989; Favilla et al., 1990; Ghez et al., 1989; Hening et al., 1988). It is well known that the task complexity effect, formalized as Hick’s law, established that as the number of response alternatives increases there is a proportional increase in reaction time and information processing demand (Hick, 1952). Three task complexity factors pertinent to response planning and/or updating provided the motivation for this study. The first, “ default specification" concerns response planning. Similar to the results of Velicki et al., (2000) young, healthy subjects in the original TRP work produced a default response when preparation time was short and specification was minimal (Favilla et al., 1989; Favilla et al., 1990; Ghez et al., 1989; Hening et al., 1988). However, in contrast to Velicki et al., (2000), the default amplitude did not capture one of the three designated targets but instead was positioned in the mid-range of the three target amplitudes. Therefore the subjects in the Velicki et al. investigation may not have specified a default amplitude when preparation time was minimal but merely chose one of the two targets on either side of the home position. The increased complexity of computing the correct amplitude with additional targets would establish with greater certainty a similar ‘default specification” between stroke and control groups. The second task complexity issue pertains to response updating and involves “ wrong direction" responses. Regardless of preparation time, wrong-direction responses did not converge on the correct target amplitude for either groups in the Velicki et al., study (2000). No further ‘updating’ or adjusting of amplitude occurred when direction was incorrectly specified. In contrast, as preparation time increased in the original TRP investigations, response amplitude gradually converged on the target amplitude regardless of the chosen direction (Favilla et al., 1989). The discrepancy between studies for “ wrong direction" responses may reflect decreased processing demands of the four versus six target task. Thus, the subjects in the Velicki et al., study may have actually had greater processing “ resources" by which to identify a movement in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 5 the wrong direction and subsequently ‘abort’ the response. Comparing the results from the present 6-target arm-movement task with the previous 4-target task of Velicki et al., (2000) not only minimizes the discrepancies between both arm-movement studies and the original TRP studies, but also extends analysis of SM area function by examining parameter specification under conditions of increased complexity. The third task complexity issue concerns the implementation of different strategies to contend with a 4-target versus 6-target task. The Gordon and Ghez model (1987b) was originally described for a reaction-time paradigm in which healthy, young adults produced isometric force pulses to one of three target amplitudes in a single direction. The original proposal was that to accomplish this 3-target task, subjects would employ a ‘continuous’ strategy (Gordon & Ghez, 1987b). With three targets, an optimal strategy would be to select the middle target and then generate compensatory adjustments to 'cut' the response off to the short and prolong the response to the long target. The 4-target movement task of Velicki et al., (2000) may have allowed subjects to adopt a ‘discrete’ strategy i.e., guess big or small. Indeed both the stroke and control groups appeared to generate an initial response to the short target and then utilize a pulse-width control strategy to prolong the response if, in fact, the target was long (Fisher et al., 2000). We hypothesized that the addition of a third target might produce a strategic shift from a ‘discrete’ to a 'continuous' strategy. The purpose this investigation was to determine the role of SM areas in planning and updating goal-directed aiming movements using a six-choice instead of a four-choice aiming task. Sensorimotor areas have been implicated in both the specification and updating of accurate movements and as such it is hypothesized that subjects with SM stroke will demonstrate a greater magnitude of amplitude (extent) errors and a greater frequency of direction errors than healthy control subjects. Further, response planning deficits underlying these aiming errors will be evident in the default amplitude selection for the more complex 6-target task. Specifically, the stroke and control groups will not ‘plan’ a similar default amplitude when preparation time is limited. A ‘continuous strategy,' as evidenced by default amplitude selection in the mid-range of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 6 the three target amplitudes will be adopted by the control group. Individuals with stroke will demonstrate a ‘discrete strategy’ as seen in Velicki et al., (2000) by selecting the short target as the default. We expect response updating deficits for subjects with stroke through the model analysis and confirming earlier findings (Velicki et al., 2000). That is, the greater amplitude errors of the stroke group will be explained by significantly less variance in final position due to compensatory adjustments compared to the control group. METHOD Subjects Twelve adults participated in this study: six with a unilateral SM area lesion secondary to a stroke involving the anterior circulation system [M (SD) = 61.3 (10.5) yrs], and six age-matched neurologically healthy control subjects recruited from the Greater Los Angeles area [M (SD) 61.5 (8.9)]. Table 1 summarizes the characteristics of each subject as well as lesion information for each subject with stroke. Unilateral stroke was confirmed from MRI or CT scan reports. The site and size of the brain lesion was determined from the actual radiologic image. For the six subjects with stroke, a board-certified neurologist outlined the outer brain and lesion areas from the scans. To determine lesion location, the lesion site was mapped onto the appropriate view in the functional atlas of Domasio and Domasio (1989). Lesion locations were identified using a classification scheme modified from Kunesch, Binkofski, Steinmetz, & Freund (1995). 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). The section with the greatest extent of each lesion using the functional atlas of Damasio and Damasio (1989) is displayed in Figure 1. 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. Subject and Lesion Characteristics Sub ID Sub- Match Sex Age (yrs) Hand- Dominance Hand Used Etiology Side CVA Lesion Duration (yrs) %Lesion Volume Lesion Location Control 1 7 F 64 R R - - - - 2 8 M 66 R R - - - - 3 9 F 63 L R - - - - 4 10 F 48 R L - - - - 5 1 1 F 55 R L - - - - 6 12 M 73 R L - - - - Stroke 7 F 67 R R hemorrhage R 5 4.1 Striatocapsular + cortex; primary sensorimotor cortical hemorrhage with subcortical extensions into internal capsule and lentiform nucleus, extends to parietal, temporal association areas 8 M 63 R R infarct R 7 15.7 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 o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 9 F 64 L R 10 F 45 R L 11 F 54 R L 12 M 75 R L Table 1. Subject and Lesion Characteristics (continued) hemorrhage R hemorrhage L infarct L infarct L 13 6.5 Primary sensorimotor and temporal cortex with subcortical temporal extensions, portions of temporal association and medial temporal areas; no basal ganglia involvement 5 3.8 Striatocapsular + cortex; primary sensorimotor cortical hemorrhage with subcortical extensions into internal capsule, lentiform nucleus; extends to frontal, parietal subcortical white matter; portions of temporal association area 12 5.0 Striatocapsular + cortex; primary sensorimotor cortical hemorrhage with subcortical frontal extensions; extensions into internal capsule, lentiform nucleus and temporal association area 3 3.3 Striatocapsular + cortex; primary sensorimotor cortical infarct with white matter cortical extensions; extensions into lentiform nucleus and temporal association areas o > G O 69 Subject #7 Subject #8 Subject #11 Subject #12 Figure 1. Lesion location for each stroke-group subject from magnetic resonance imaging (MRI) or computer tomography (CT) images. Each section shows the lesion-site of greatest extent using the functional atlas of Domasio and Domasio (1989). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 The Scion Image ® program was used to obtain an index of lesion volume. As stated above, the perimeter of each brain slice from the MRI or CT scan as well as the perimeter of each lesion site was outlined. A routine is utilized by the program which checks if the pixel intensity of the outline is within the density slice range (inside = 1 or 0). The outlined region is captured and each brain slice area is calculated. Volume is calculated by multiplying area by slice thickness. By summing across all slices, total brain as well as total lesion volume were determined and the percent lesion volume calculated (Table 1). Total brain volume included the cortical and subcortical areas and excluded the cerebellum, brainstem, and ventricles. Each subject in the stroke group demonstrated some motor impairment that affected their functional mobility. With the exception of subject 12, each of the subjects with stroke demonstrated hemiparesis of the limbs contralateral to the side of the brain lesion. Subject 12 demonstrated minimal motor impairment of the right extremities which did not limit his functional use of that side (Table 2). All of the subjects with stroke were able to ambulate independently at a household or limited community level. Details of specific impairments and functional limitations are described below (see procedures) with assessment results provided in Table 2. 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 2. Impairment and Functional Limitation Measures of Stroke Subjects Impairment Functional Limitation Subject# Contralateral UEFM- Motor nContralateral UEFM-Sensory ‘Grip - C (lbs) “ Grip -1 (lbs) *FIM Motor 7 14 0 0 37 82 8 16 8 18 8 6 83 9 29 3 11 57 8 8 1 0 14 1 2 15 72 89 11 1 0 4 0 6 8 8 8 1 2 64 1 2 76 8 6 91 Contralateral upper extremity Fugl-Meyer motor subscore (total possible = 6 6 points) "Contralateral upper extremity Fugl-Meyer sensory subscore (total possible = 12 points) C rip - C = grip strength of contralateral hand (contralateral to side of brain lesion) “ Grip - 1 = grip strength of ipsilateral hand (ipsilateral to side of brain lesion) *FIM Motor = the motor sub-scale of the Functional Independence Measure (total possible = 91) 72 Instrumentation Data were acquired using an instrumented manipulandum. The manipulandum is a light weight aluminum lever which was affixed to a wooden support. By securing the wooden support to the edge of a table, the lever was effectively elevated to shoulder height. The lever rotates in a near frictionless vertical axle such that it was free to move in the horizontal plane. A hand-grip at the distal end of the lever was adjusted to accommodate the length of each subject’s forearm. A computer monitor was positioned in front of the subject at eye level at a distance of - 40 cm. All subjects could comfortably read the messages displayed on the monitor. A linear potentiometer was attached to the base of the vertical axle such that horizontal movement of the lever produced a change in voltage corresponding to the lever’s angular displacement. This analog signal was sampled at 2500 Hz., and converted to digital units by an A-to-D board (Keithly/Metrabyte) installed in a 486-based, 50 MHZ speed IBM-compatible computer. Each trial was 2000 ms in duration and consisted of A-D sampling of the lever potentiometer beginning 100 ms before the movement cue (see Task for definition of movement cue). Custom ASYST software (Nakai, 1997) was used for the control of data acquisition, data analysis and storage. Electromyographic data (EMG) were acquired simultaneously with the kinematic data. Surface EMGs were recorded from the biceps and triceps muscles. The two channels of EMG data were sampled at 2500 Hz. The signals were amplified (Data, Inc. differential amplifier) with the gain set between 3K and 5K for both muscle groups. Signal frequencies below 30 and above 500 Hz were filtered (attenuated) by means of both a 12-position high frequency and 12-position low frequency roll-off switch. Task The task involved moving the lever from the home position to one of six target locations that were differentiated by two direction and three amplitude parameters (Figure 2). Subjects flexed or extended their elbow to match one of three targets on either side of home position. Targets were spaced 15° apart such that target amplitudes were ± 15°, 30°, and 45° from home. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 3 Subjects initiated movement towards one of the six targets in synchrony with an auditory cue. Specifically, the onset of lever movement was to be timed with the last in a series of 4 tones. The computer presented a series of 4 audible tones, each with a progressively higher pitch. Pitch increased from 625 to 1300 Hz, corresponding to the ascending notes of an octave beginning with middle C and including E, G, and C. The tones each had a 50 ms duration, and were spaced by an inter-tone interval of 500 ms. Subjects faced a computer monitor displaying a graphic representation of the target locations as illustrated in Figure 2. A yellow arrow provided the movement cue for the goal target. A red pointer provided information regarding the real-time position of the lever and provided feedback about the end position of the movement response. Subjects also used the red pointer as feedback to guide the lever into the home position between trials. Written cues (e.g., “ Please return to home”) at the beginning of a trial, as well as written feedback (e.g., ‘Too late” ) following the completion of movement were also provided. Details regarding specific written messages are presented below. Fioure 2. Illustrates the six target locations with the lever positioned at the 5m target. A replication of the right hand display was presented on the computer monitor for subjects using their right arm to perform the task; the left hand display was viewed on the monitor for subjects using their left arm. Numbers were not actually displayed, but are included for clarification of the text .Extension Home Flexion Right Hand Display Left Hand Display Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 Procedures All subjects read, signed and received a copy of the Institutional Review Board approved informed consent form. To determine the homogeneity between groups (stroke, control) for visual, auditory, cognitive and motor function status, all subjects were assessed with the following measures: (1) Visual acuity (Rosenbaum Pocket Vision Screener), (2) peripheral vision of each eye (visual field), (3) hearing (4) attentional capabilities, (5) grip strength, and (6 ) manual dexterity (Box and Block test; Mathiowetz, Volland, Kashman, & Weber, 1985). To determine stroke severity, post-stroke functional limitations, and level of disability, the upper extremity Fugl- Meyer Assessment (Fugl-Meyer, Jaasko, Leyman, Olsson, & Steglind, 1975), the mobility sub scale of the Functional Independence Measure (Hamilton, Granger, Sherwin, Zielezny, & Tashman, 1987) and the physical functioning sub-scale of the SF-36 were used (Ware & Sherboume, 1992), respectively. Four subjects in the stroke group performed the task with their non-dominant arm while the remaining two subjects performed with the dominant arm. Similarly four control subjects performed the task with their non-dominant arm and two performed with the dominant arm (Table 1). Subjects in both groups that used the right arm to perform the task sat to the left of the lever and grasped the lever handle with their right hand. Subjects in both groups performing the task with their left arm used the opposite configuration. Seat height was adjusted for each subject such that when the lever was placed in the home (or start) position the subject was positioned between 85° and 90° of shoulder flexion, 45° of shoulder abduction and 70* of elbow flexion. As such, the axis of movement rotation was at the elbow joint and there was little if any contribution from shoulder internal and external rotation. This assured that the biceps brachii and the lateral head of the triceps brachii were the primary agonist and antagonist muscles for task performance. Subjects were then oriented to the lever movements and the corresponding visual display. For exposure to the basic requirements of the test task, subjects performed three practice tasks each successively approximating the test task. The first practice task ensured that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 subjects were able to accurately locate and move to the six target locations without any time constraints. At the beginning of a trial, the instruction: “ Please return to home" was displayed on the monitor. The subject used the red pointer for feedback to move the lever to the home position. When the subject remained within 2.5° of the home position for 1 second (s), the red pointer disappeared and a “ Ready” cue was displayed. Then the yellow arrow (i.e., the movement cue) appeared and pointed to the goal target position. Immediately after the subject completed their movement response, the red pointer reappeared providing feedback regarding the movement's end position. The subject could then compare the position of the yellow target arrow with the position of the red pointer indicating their movement response. Written feedback indicated a “ Hit” for an accurate response, or a “ Miss" for an inaccurate response. The criteria for a “ Hit” response was no greater than 2.5s from the target. Following this feedback, the written cue, “ Please return to home” was displayed and the next trial sequence began. Subjects controlled the intertrial interval by the rate at which they returned to the home position. In the second practice task, subjects were trained to initiate movement within a pre determined amount of time (the timing window) surrounding the onset of the 4th tone. A timing criteria of 60 ms before and after the tone was used to specify the timing window in which the movements of acceptable trials had to be initiated (Figure 3a). The 50 ms tone plus the 120 ms timing window meant that subjects had a total of 170 ms tolerance for movement initiation in this second practice task. Trials in the second practice task were identical to those of the first practice phase with the following exceptions. After the subject remained in the home position and the red pointer disappeared, a series of 4 tones began. The movement cue (i.e., the yellow arrow) was presented between the 3rd and 4th tone, exactly 200 ms before the 4th tone. The written feedback now provided information regarding when movement was initiated with respect to the criterion timing window. Figure 3a depicts the written feedback provided when movement was initiated at various times throughout the trial sequence. “ Stay Steady” was displayed when movement was initiated prior to the 3rd tone. Too Early” was displayed when movement was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 6 initiated after the 3rd tone and before the timing window. ‘Too Late’ was displayed when movement was initiated after the timing window. ‘Good” was displayed when movement was initiated within the timing window. The third practice task ensured that subjects were able to: (a) initiate movements within the timing window using variable preparation times, and (b) perform a rapid movement response. The trial sequence was identical to the second practice phase with the following exceptions. The yellow movement cue arrow was now presented at variable times between 0 and 400 ms prior to the 4th tone. In addition, movements had to be completed within a pre-established movement time (MT) criterion to receive the “ Good” feedback. A hierarchy of feedback was used such that, “ Too Slow” was provided after trials in which movement was initiated within the timing window but MT was greater than the MT criteria. The MT criteria was set at 300 ms to establish that subjects were to produce a rapid movement response with no overt corrections. Throughout the testing phase (see below) the subjects were challenged to move more rapidly as the criterion movement time was decreased. Orientation to the basic requirements of the task via the 3 practice tasks required s 30 minutes for all subjects. The goal of the practice tasks was to become familiar with moving the lever, locating the targets and beginning to contend with the timing demands of the TRP. As such, each individual subject determined when they had performed a sufficient number of trials within a practice task. After completion of the third practice task, subjects progressed to the testing task. The testing period trial sequence was identical to that of the third practice phase. The movement cue was randomly presented between the third and the fourth tone within one of four preparation time bins (1 = 300-400 ms, 2 = 200-299 ms, 3 = 100-199 ms, and 4 = 0-99 ms). For each day of data collection, electromyograms were recorded from the biceps muscle and the lateral head of the triceps using surface electrodes. To reduce the electrical resistance of the skin, the skin overlying the bellies of the biceps brachii and lateral head of the triceps brachii was lightly sanded and cleaned using 70% alcohol. The electrodes were passive, silver/silver chloride (Ag/AgCI) disks, one centimeter in diameter and were placed one centimeter Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 apart using a double differential electrode configuration. At the beginning of a session, maximal voluntary flexion and extension forces produced at the elbow were recorded, along with the associated EMG. With the arm supported on the wooden support, the shoulder positioned in 85° • 90° of flexion and the elbow and wrist in line with the shoulder, subjects were instructed to push in with their forearm as hard as they could against a metal bar fixed to the wooden table. Maximal biceps contraction was recorded as subjects held the flexion contraction for a 2 second period. The procedure was repeated for triceps as subjects pushed out against the metal bar. 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. Possible Time For Movement Cue Presentation Acceptable Timing .Window 4| i f I— I B Movement Cue Movement Initiated Timing Error Written Feedback Cue Presented 'Return to Home" Figure 3. a)The written feedback displayed when movement was initiated at various times throughout the trial sequence. This example shows a total timing window of 75 ms before and 75 ms after the 50 ms tone (total timing window = 200 ms), b) The modified timed-response paradigm (TRP). Movement initiation was to be synchronized with the last in a series of four tones (1-4). The movement cue was presented at variable times between 400 and 0 ms before the fourth tone. In this example, the stimulus is given z 300 ms before the 4th tone and is calling for movement to target four. The stimulus-tone (S-T) interval, the stimulus-response (S-R) interval, and the timing error [(S-R interval) - (S-T interval)] are shown. ^ 0 0 79 Data acquisition Two independent variables were used to control preparation time (Figure 3b). The stimulus-tone (S-T) interval was the time between movement cue presentation and the onset of the 4th tone. This was the preparation time forced by the paradigm. The stimulus-response (S- R) interval was the time between the movement cue presentation and the onset of movement. Subjects were not able to time their movements exactly on the 4th tone in all trials, so the S-R interval reflected the actual preparation time. The timing error was the difference between the imposed S-T interval and the actual S-R interval. A “ block” consisted of 48 acceptable trials without timing or MT errors. Two different types of trial blocks were presented. In the predictable condition, the targets of serial trials were presented in a fixed order (i.e., target 1 ,2 ,3 ,4 ,5 ,6 ) and in the unpredictable condition, the targets of serial trials were presented in a pseudo-randomized order (e.g., target 2 ,5 ,1 ,6 ,4 ,3 ). Before each block of trials, subjects were informed as to whether the upcoming block of trials would be delivered randomly or in order. Twenty-three acceptable trial blocks were required of each subject: 8 blocks of the predictable condition and 15 blocks of the unpredictable condition. Fifteen blocks in the unpredictable condition assured that each subject completed a sufficient number of same target/S-T interval combination trials. The hallmark of the timed-response paradigm is that the amount of time that subjects have to prepare their response to the appropriate target is under experimental control. Regardless of the time of target presentation, the subject must move on the 4t h tone. As stated earlier, ‘moving on the 4m tone' initially meant that a subject had a 120 ms window of time surrounding the 50 ms tone in which to initiate movement and meet acceptable timing criteria. Subjects could successfully take advantage of this 170 ms cushion by initiating their response early (within the acceptable window of time) when the target was presented closer to the 3 r d tone and initiating their response later when the target was presented close to the 4t h tone. In order to monitor and regulate the degree to which the subjects could control their preparation time, a Pearson correlation of preparation time and timing error was computed after each successfully Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 completed block. A non-significant correlation (p s .23) indicated that the responses were being timed with the 4t h tone and not systematically varied with respect to preparation time. A significant correlation would occur if over the block of trials, the subject moved after the tone when preparation time was short, and ahead of the tone when preparation time was long. In addition to the timing correlation, the 48 acceptable trials were displayed in graph form on the computer monitor. A plot of preparation time (x axis) by timing error (y axis) was displayed. After each block, the plot and the timing correlation value were used to provide timing performance feedback to each subject. When the timing correlation was significant, subjects were encouraged to move on the 4* tone regardless of the target presentation time. In addition, after each block in which a significant correlation occurred and prior to the next block, the timing window was constricted by decreasing the allotted time before and after the tone. Subjects repeated blocks of trials until 15 unpredictable and 8 predictable blocks were obtained that had timing correlations similar to the lowest correlation achieved during the testing period (i.e., their best performance). Generally speaking, the blocks selected for analysis were those that were collected later rather than earlier in the testing period. Later in the testing period, subjects either became more skilled in synchronizing movement initiation with the 4W tone that the timing correlations were non-significant or the timing window was constricted to less than 1 0 0 ms (including the 50 ms tone). Once the timing window was less than 100 ms, all completed blocks (regardless of timing correlation) were accepted for analysis. Timing windows of £ 90 ms (including the tone duration) were achieved by 5 control subjects and 2 subjects with stroke. Subjects readily identified that a trial was considered acceptable if movement occurred within the acceptable time period (relative to the 4t h tone) regardless of response accuracy. Subjects were continually encouraged to be as accurate as possible. In addition, after each block was completed the subjects were shown a scatter plot of Absolute error (AE) by S-T interval. A high correlation indicated that responses were more accurate when target information was provided early and less accurate when target information was given later with respect to the 41 ” tone. For the subjects with stroke, 80% of the accepted unpredictable-condition blocks had Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 significant accuracy correlations of r > .20 compared with 90% for the control subjects. To complete data collection, subjects in the stroke group returned to the laboratory between 7 and 13 days. Subjects in the control group returned to the laboratory between 6 and 7 days. All subjects performed between 3 and 10 blocks/day. To produce a full complement of acceptable test trials (i.e., 1104 trials with specific S-T interval, target characteristics within the predictable and unpredictable conditions), each subject performed between 25 and 35 trial blocks. Data Analysis Data Management A block of trials consisted of 24 trial types resulting from 6 targets x 4 S-T intervals. Data were recombined such that ail predictable and unpredictable like trial types were processed and analyzed together. Therefore, at the completion of 15 unpredictable blocks, an individual subject had completed 30 trials with the same S-T and target combination, and 16 trials for every trial type of the predictable blocks. Following the procedure in which like trial types were combined, key kinematic variables were extracted using Datapac III for Windows95 version 2.0 by RUN Technologies Co ®. Raw displacement or position data were conditioned using a low-pass filter with a 20 Hz. cut-off frequency. Velocities were derived using a differentiation process. The amplitude value of the middle data point in each set of three points was replaced with the slope of the regression line calculated for a minimum interval. The minimum interval, known as the time constant is equal to 2 x the sample period. The sample period for this study was 400 psec. As such, the slope value is obtained over 800 psec. The EMG calibration data for each testing session was inspected for the presence of artifact and then a moving root mean square (RMS) calculation was performed. This smoothing operation replaced the amplitude value of each data point with the root mean square calculated for a 100 ms interval around that data point. The biceps and triceps EMG data for each testing session were then converted from raw EMG to a percentage of the calibration run for that session. As such all EMG was normalized to 100% of voluntary maximal contraction. EMG data Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 2 were then RMS smoothed using a 10 ms time interval. The normalization procedure was completed so that individual subject electromyograms could be combined across sessions. Data Screening For each of the 1104/subject trials both timing and kinematic criteria were applied in order to determine whether a trial would ultimately be used for analysis. If a subject initiated movement before the target cue was presented (i.e., the stimulus), a negative S-R interval would result (S-T interval + start time). These ‘anticipation’ error trials represented 2.6% of all trials for the stroke group and 1.9% of all trials for the control group and were eliminated from analysis. The kinematic algorithm detailed in Fisher et al., (2000) was applied to each individual trial. As such, all single peaked velocity waveforms were considered for analysis. Double response trials were rejected or salvaged based on the criteria outlined in Fisher et al., (2000). For the subjects with stroke, 2 % of both predictable and unpredictable trials were eliminated from analysis based on the kinematic criteria. Only 0.3% of both predictable and unpredictable trials were eliminated for the control group. After the screening procedure, the full complement of predictable condition trials consisted of 267-379 trials/subject (total n = 2139) for the stroke group and 360-378 trials/subject for the control group (total n = 2230). In the unpredictable condition, the number of analyzed trials for the stroke group was 507-865 trials/subject (total n = 4083) and 563-847 trials/subject (total n = 4246) for the control group. Dependent measures and statistical analyses The algorithm to determine movement onset and offset was a change in velocity from zero in the appropriate direction of movement. The position and velocity profiles for each trial within the above described event period were used to calculate the following variables (Figure 4): 1 ) final position achieved (Displacement = offy - or^ (a)], 2) average velocity from the onset of movement (b) until peak velocity (c), 3) movement time = off, - on, (d), 4) velocity rise time = c, - b,. Additionally, error was also computed as the difference between final position and the target. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. On Off Position S-R Movement Cue Velocity S-T Movement Cue 4th 83 45 200/sec 1 0 0 0 m* . . 200 ms .. M H Time Figure 4. Representative data from a single trial (45s flexion) showing key kinematic and temporal variables: a = Ony - Offy = Displacement; average velocity (°/sec) from the onset of movement b until peak velocity c; position rise time = c - b (ms) d = Off, - On, = Movement time; e = S-R interval; f = S-T interval. Parameter Specification Analysis The differences between the stroke and control groups in the time-course of parameter specification were marked by the following three dependent measures: 1) Constant error (CE) which is a signed value in degrees representing the difference between the final position achieved (a in Figure 4) and the target position. 2) Absolute error (AE) or the absolute value of CE. For each subject, the mean AE was calculated for acceptable trials with the same trial type (target-S-R Interval), block condition (predictable, unpredictable) and response direction accuracy (correct, wrong direction). 3) Direction error is the number of responses which are in the opposite direction of the goal target. The number of direction errors within each preparation interval was tallied. To evaluate the amplitude specification of responses in the correct direction, a mixed model ANOVA was used with AE as the dependent variable. The between-subject factor is Group (stroke, control), and the two within-subject factors are Target (1,2,3,4,5,6 ) and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 preparation interval (1= 300-400, 2 = 200-299,3 = 100-199,4 = 0-99). To evaluate direction specification, a stepwise logistical regression was used to evaluate the frequency of wrong direction responses as a function of preparation interval. Component Model Analysis The components of the statistical model that were used to determine the variance in final position due to premovement planning and compensatory adjustments are detailed in Fisher et al., (2000). Use of this model gauges the influence of the target on final position through two paths. Specifically, the percentage of variance in final position explained by planning (path 1) compared to compensatory adjustments to trajectories (path 2) was determined. Displacement was used as the dependent variable in a multiple regression analysis in which the average from onset of movement to peak velocity (dY/dt) was entered first followed by the target. The squared correlation coefficient describing the relationship between dY/dt and final position (r*y, ) reflects the degree to which final position is explained by premovement planning. It accounts for the indirect influence of the ‘predetermined’ but not yet specified target. Next, the second path, directly from target to final position was added to the regression. The squared multiple correlation coefficient {R?Y 1 2 ) was computed for this relationship and reflects the proportion of variance in final position that can be explained by a linear combination of velocity and the target. The additional variance accounted for by the target is equal to the difference between that proportion of the variance accounted for by the combination of average velocity from onset to peak and target and that accounted for by the velocity measure alone (& Y ,2 -r*Y 1 ). The increment in R2 obtained by this procedure will estimate the degree to which compensatory adjustments are correlated with the end position achieved. Separate regression analyses were performed for each of the 4 preparation intervals nested within the 2 target directions (i.e., flexion and extension), which in turn were nested within correct and wrong direction responses, and finally, nested within predictable and unpredictable conditions. The difference between the stroke and control groups in variance of final position explained by preplanning and compensatory adjustments was determined by a 2 group (stroke, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 control) x 2 response directions (correct, wrong) x 4 S-R interval (4=300-400, 3=200-299, 2=100- 199,1=0-99) mixed model Analysis of Variance (ANOVA). The dependent variables used in this analysis were the regression coefficients representing variance explained by preplanning (r\,), and compensatory adjustments due to target {R?y i2 -r2 ^). For all statistical tests, significance was set at p < .05. Post-hoc orthogonal contrasts with Bonferroni correction were performed to determine the locus of any significant main and interaction effects. Individual subject analyses individual subject agonist-antagonist EMG burst patterns were inspected to explore different strategies used in task performance. Over the 30 (approximately) like trials across blocks (i.e., same S-R Interval/Target/Response Direction/condition) biceps and triceps EMG ensemble signal averages were computed separately for each muscle. Averaged response profiles were time-locked to a change in movement velocity from zero in the appropriate direction of movement. EMG responses were calculated over an interval of 1000 ms around the same reference event that marked the beginning of the event period. A delay of -200 ms before the onset of the reference event was set in order to identify the earliest EMG response(s). As such, EMG signal averages were calculated within the time period of 200 ms prior to the reference event (i.e., movement onset) and 800 ms after onset. Electromyographic agonist-antagonist activation patterns were examined in order to additionally index response updating (i.e. compensatory adjustments). Response updating is most evident in correct direction responses in the unpredictable condition. Therefore individual subject activation patterns for correct direction responses were inspected. For each subject, 45° target responses in one direction (i.e., either flexion or extension) were compared with that subjects' default amplitude responses. These two target amplitude responses were compared within and between the longest and shortest preparation intervals. Additionally, long and default amplitude responses for a 'middle' preparation interval in which the subject demonstrated the greatest percent variance in final position due to compensatory adjustments was also examined. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 RESULTS Predictable condition responses are described first followed by a description of the unpredictable condition responses. Group differences in the unpredictable condition that highlight the contribution of the ipsilateral SM areas in response parameter planning and updating are presented for both correct and wrong direction responses. Comparison of means between same-amplitude flexion and extension targets revealed that AE was similar for targets of the same extent P > .05. Therefore, for all analyses, responses to each of the 3 similar-extent targets were combined. Target effects were addressed with respect to the short (15s ), medium (30°) and long targets (45°). Use of the statistical model as a valid ‘explanation* of final position was demonstrated empirically by the high percentage of variance in final position explained by the combined paths of the model. For unpredictable correct direction responses, the proportion of variance in displacement explained by both premovement planning and compensatory adjustments was 79% (± 2.1%) for the control group and 73% (± 2.0%) for the stroke group. Total percent variance increased to 93% ± 1.0% for the control and 89% ± .98% for the stroke group for predictable condition responses. Thus, variance in final position was largely predicted by the model for both groups. However, the differences in total percent variance between groups were significant (P = .05; P = .03 for unpredictable and predictable conditions, respectively). Predictable Condition Responses All Subjects The predictable condition provides a standard from which to gauge both the upper limits of accuracy and planning capabilities. Since task complexity is low and subjects can presumably plan their responses in advance, the predictable condition is a 'yardstick* for the highest degree of accuracy attainable as well as the highest explanation of final position due to planning. Across the three target amplitudes all subjects achieved their highest degree of accuracy in the predictable condition. As expected, all subjects preplanned their responses before the movement cue. As such, this degree of response accuracy, was not dependent on preparation Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 time (Figure 5a - striped bars). The capability under predictable conditions to plan the response in advance was confirmed by the model analysis. All subjects demonstrated the greatest percentage of variance due to adjustments in the predictable condition compared to the unpredictable condition (see below) suggesting they used target information in advance to plan their response. Group Analyses While the mean AE for the two groups was similar, the stroke group was (statistically) less accurate than the control group across all targets [AE M (SEM) = 4.8° (.09) for the stroke group; 4.5° (.09) for the control group, P = .03]. The stroke group was more accurate to the middle amplitude target than either the short or long amplitude targets, while the control group was more accurate to both the middle and long than the short amplitude targets (Figure 5b). This resulted in a significant target by group interaction (P < .0001). Post hoc analysis revealed that absolute error to the middle target was similar for the two groups. The stroke group was significantly more accurate to the short amplitude target than the control group (P = .014), but significantly less accurate than controls to the long amplitude target (P < .0001). There was a significant difference in premovement planning capability between the two groups. For the stroke group 80% ± 1.9% (mean ± SEM) of the variance in final position was explained by premovement planning compared with 85% ± 1.8 % for the control group (P < .05). There were no group differences in the ability to generate adjustments (P = .13). For the stroke group an additional 9% ±1.1% of the variance in end position was explained by adjustments compared with 7% ± 1% for the control group. (Figure 5c). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 8 15 1 2 9 6 3 0 100-199 a a 4 2 0 lo n g start S-R imarval (m i) Targst AmpMud* f I Com pansatory AdMtmaca Figure 5. a) Bar graph of mean AE (iSEM) for responses within each S-R interval for the unpredictable (crossed bars) and predictable (striped bars) conditions. Data are averaged across 12 subjects and 6 targets, b) Symbols represent group means for AE (iSEM ) to each target amplitude in the predictable condition. The control group is represented by the triangle and the stroke group by the squares. Data are averaged across S-R interval 6 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0001). c) Stack bar plot of percent variance in displacement for predictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left and the stroke group is to the right. Data are averaged across 6 subjects/group, 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. Unpredictable Condition Responses Response Planning: Correct Direction Amplitude Specification All subjects. As the S-R interval increased, all subjects were able to use movement cue information to improve response accuracy, P < .0001. Accuracy significantly improved across all four S-R intervals (P < .0001 for successive S-R intervals) (Rgure 5a - crossed bars). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 Group differences. Across all targets and preparation intervals, the stroke group was significantly less accurate [M (SEM) = 9.4° (.13)] than the control group [M (SEM) = 8.3° (.12), P < .0001]. The time course for extent specification was similar for both groups, P = 0.3. Figure 6 a demonstrates that accuracy progressively improved for each group as time to prepare increased. A ■ I Stroke Control 0-99 100-199 200-299 300-400 S-R Interval (ms) B 16 -| Stroke ~ 4 _ Control 12 - I U J < 3 ® 2 short medium long Target Amplitude Figure 6 . a) Bar graph of group mean AE (±SEM) within each S-R interval for unpredictable correct direction responses. Data are averaged across 6 subjects/group and 6 target amplitudes. Control group = striped bars; stroke group = dark bars, b) Symbols represent group means for AE (±SEM) to each target amplitude for correct direction, unpredictable responses. The control group is represented by the triangle and the stroke group by the squares. Data are averaged across S-R interval 6 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0 0 0 1 ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 Group differences in AE for each of the three target amplitudes reveals the magnitude of the accuracy differences between groups. A significant group by target interaction for AE (P < .0001) is illustrated in Figure 6 b. The control group was significantly more accurate to both the middle [M (SEM) = 5.5° (.23)] and long target amplitudes [M (SEM) = 9.7° (.22)] than the stroke group [M (SEM) = 7.0° (.22) for the middle target; M (SEM) = 15.0° (.23) for the long target]. Only for the short target amplitude was the stroke group [M (SEM) = 6.2s (.22)] significantly more accurate than controls [M (SEM) = 9.5s (.21), P < .0001]. Additionally, the stroke group was significantly less accurate as target amplitude increased whereas the control group was most accurate to the middle target with a similar degree of accuracy to the short and long targets. Indeed, for all of the control subjects, the default amplitude was close to the middle target ( - 25s ). Similarly, for three subjects in the stroke group (subjects 9,10 and 11) the default amplitude was close to the medium target. The remaining three subjects with stroke (7 ,8 , and 12) selected the short target as the default. The data from subject # 8 is depicted in Figure 7a. Many of the short-target responses for subject 8 have a constant error (CE) of approximately zero degrees in the earliest S-R intervals. In addition, the regression line through the 15s target responses is parallel to zero CE across preparation intervals. This suggests that subject 8 specified the short target amplitude as part of the premovement plan. Regardless of preparation time, CE for this default amplitude does not change. For the data from subjects 11 and 6 in Figure 7b and c (respectively), the default amplitude is closer to the middle target. Error for the middle target across the preparation period hovers around zero degrees for both of these subjects. Model Analysis: Premovement Planning There were no differences between groups in percent variance in final position due to premovement planning. Sixty-two percent ± 2.5% of the variance in final position was explained by premovement planning for the stroke group compared with 6 8 % ± 2 .6 % for the control group (P = .09). Thus, the accuracy differences between groups does not appear to result from a more accurate 'plan' for the control compared with the stroke subjects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 Stroke Svkjd m o > 9 e. UJ O ♦ > ♦ * J O ■ - 1 - 100 —I — 200 — I — 300 I 400 “ I 500 S-R Interval (ms) B Stroke S a l|«a in S-R Interval (ms) Control S a ltM l • S-R Interval (ms) Figure 7. Scatter plot of constant error (CE) across S-R intervals for a full complement of correct direction, unpredictable condition responses. Subject 8 (a), and subject 11 (b) are stroke subjects and subject 6 (c) is a control subject Regression lines are run through each target amplitude. A thicker horizontal line at 0° represents the target. For S8 , the default position was the short target extent For S11 and S6 the default position was close to the middle target amplitude. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 However, based on two separate measures related to specification of the initial plan and updating the response, the stroke group clearly separated into two distinct behavioral sub groups. As described earlier, first, the default amplitude for subjects 9,10, and 11 was close to the middle target amplitude while for subjects 7 ,8 , and 12 it was the short target amplitude. Second the percent variance in final position due to compensatory adjustments was higher for subjects 9,10, and 11 than for subjects 7 ,8 , and 12 (Table 3). In addition, subjects 9 and 1 1 demonstrated the lowest variances in final position due to planning. Based on the apparent differences with how these subject ‘subsets’ within the stroke group contended with the task, the model analysis was re-run comparing control subjects with each of these two stroke groups. Subjects 9,10, and 11 were considered the ‘high-adjustment’ stroke group and subjects 7, 8 , and 12 were considered the ‘low-adjustment stroke group.' The high-adjustment stroke group demonstrated significantly less percent variance due to planning [M (SEM) = 56% (4.1%)] than the control group [M (SEM) = 6 8 % (2.6%), P = .009]; whereas there were no differences between the low-adjustment stroke group [M (SEM) = 6 8 % (2.2%)] and the controls (Figure 8 a, filled stack). Response Updating: Correct Direction Recall that in Experiment 1 there was a significant effect of preparation interval across both stroke and control groups on the percent variance in final position due to compensatory adjustments. Thus, to determine the time course over which compensatory adjustments occur, the control group was separately combined with each stroke sub-group. The proportion of variance in displacement explained by the addition of target amplitude (i.e., adjustments) increased for the control and ‘high-adjustment’ stroke groups as preparation interval increased (Figure 8 b). There was a significant main effect of preparation interval on the percentage of variance in final position due to compensatory adjustments (P < .0001). Thus, when the target was presented progressively earlier with respect to the 4m tone, both groups of subjects could utilize target information to adjust their responses. Post hoc analysis revealed that the proportion of variance due to compensatory adjustments in the shortest preparation interval (0 -9 9 ms) was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 significantly less than the middle two preparation intervals (i.e., 100-199 ms, and 200-299 ms), P = .001 and .009, respectively. There were no differences between the longest and the other three preparation intervals in percent variance due to compensatory adjustments. There was not a significant effect of preparation interval on percent variance in displacement due to adjustments when the control group was combined with the ‘low-adjustment’ stroke group. Figure 8 c demonstrates the pattern of change in percent variance in adjustments for each of the three groups (i.e, control, high-adjustment and low-adjustment stroke groups). While there were no group by preparation interval interactions between the control and either of the stroke groups, the differences between groups in the pattern of change over preparation intervals represents different 'adjustment strategies' to be discussed later. For the present, both the control group and the high-adjustment stroke group demonstrated a pattern of change in which the percent variance in endpoint due to adjustments was highest in the two middle preparation intervals. The low-adjustment stroke group gradually increased from lowest to highest percent variance due to adjustments as preparation time increased. Only in the longest preparation interval did the low-adjustment stroke group approach a percentage of variance in final position due to adjustments similar to either the control or high-adjustment stroke group (Figure 8 c). The percent variance in final position due to adjustments was significantly different for the control group compared with each of the stroke groups, however, the differences were in opposite directions (Figure 8 a - open stack). The high-adjustment' stroke group had a significantly greater percent variance due to adjustments [M (SEM) = 18% (2.4%)] than the controls [M (SEM) = 11 % (1.3%), P =.002]. The ‘low-adjustment’ stroke group demonstrated significantly less percent variance from adjustments compared with the control group [M (SEM) = 5% (1.1%), P = .004]. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 4 Highest ▼ Lowest Table 3. Individual Stroke Subject Data r i i j > 1 IB i I ! % Variance I % Variance | Total Plan | Adjustments j %Variance S10 (73%) S10 (88%) S7 (11.1°) S11 (17%) S12 (10.0°) S7 (69%) S10 (15%) < ^ p £ > S11 (8.4*) S12 (63%) S12 (8%) | S7 (74%) | S7 (5%) j S12 (70%) S10 (4.8°) S11 (40%) S11 (57%) The table includes individual data for each stroke subject. All values shown represent means for correct direction responses across the three target amplitudes and four S-R intervals. Subjects are listed from top to bottom (rows) in order of highest to lowest value. Variables (columns) presented are as follows: First column = absolute error in degrees; second column = percent variance in displacement due to premovement planning; third column = percent variance in displacement due to compensatory adjustments; fourth column = total percent variance in displacement. Note separation of the stroke group into two subgroups: S9, S10, and S11 = High-Adjust; S7, S8 , and S12 = Low Adjust Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 B ConM wvmswk* uvU|Snk* 25 I 20 - 15 ' 10 - * 0-99 100-199 200-299 300-400 S-R Interval (m s) ■ Control CD Hign-AO|utt Strok* C Z 1 Low-*d(u»t Strok* o n 100-IM 20O-2M S-R Interval (ms) Figure 8 . a) Stack bar plot of percent variance in displacement for predictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left, the high-adjustment stroke group is in the middle and the low- adjustment stroke group is to the right. Data are averaged across 6 subjects for the control group and 3 subjects per stroke sub-group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot, b) Bar graph of correct -direction, unpredictable-condition responses representing percent variance in displacement due to compensatory adjustments over S-R interval. Data were averaged across 9 subjects (6 control and 3 high-adjustment stroke subjects). The shortest preparation interval C O - 99 ms) was significantly different than the middle two preparation intervals (i.e., 100-199 ms, and 200-299 ms), P - .00land .009, respectively, c) Percent variance in final position explained by compensatory adjustments for each group as a function of preparation interval. Control group = dark bar; high-adjustment stroke group = open bar, and low-adjustment stroke group = striped bar. There was no group x preparation interval interaction for either the control and high- adjustment stroke group (P = .36) or the control and low adjustment stroke group (P = .61). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 Response Updating and Accuracy The significance of the enhanced capability of the high-adjustment stroke group in performing compensatory adjustments is revealed in Figure 9 and Table 3. Figure 9 illustrates the relationship between percent variance due to adjustments and overall accuracy. Each of the subjects in the high-adjustment stroke group (i.e., subjects 9,10, and 11) were overall more accurate than the subjects in the low-adjustment stroke group (i.e., subjects 7, 8 , and 12). While the stroke group as a whole was less accurate than the control group, recall that the mean difference in AE across targets and preparation intervals was small. Clearly, the high-adjustment subset of the stroke group contributed to narrowing the error differences between stroke and control groups. 25 -| 1 20 - E § £ 15 - Q o 10 c s c C O •e < o > a ? 10 - 4 6 8 12 10 Mean AE (Deg) Figure 9. Individual stroke subject scatter of percent variance in displacement due to compensatory adjustments as a function of AE. Each stroke subject is represented by a different symbol. Data were correct direction responses and averaged across 6 targets and 4 S-R intervals. Note separation of stroke subjects into 2 groups with subjects 9,10, and 11 demonstrating lower AE and higher percent variance than subjects 12,7, and 8 . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 Individual Subject EMG Activation Patterns As described above, significantly greater variance in final position due to compensatory adjustments was evident for the 'high-adjustment' stroke group compared with the control group and for the controls compared with the ‘low-adjustment’ stroke group. Agonist-antagonist EMG patterns for one representative subject within each of these groups were examined for corresponding differences in response updating evident in the EMG activation patterns. Signal artifact prevented analysis of EMG for 3 of the 6 control subjects and 1 subject for each of the 2 stroke subgroups. Figure 10 demonstrates the averaged displacement, velocity, biceps, and triceps profiles with SEM over 18 trials for Subject 8 . The responses were to the 15° extension target in the shortest preparation interval (0-99 ms). The velocity, biceps, and triceps profiles were reassembled (Figure 11, Figure 12, and Figure 13) in order to compare EMG activation patterns both between a subjects default and the long target response as well as between preparation intervals within each target amplitude. A representative subject from the control group (subject 2, Figure 11) was the most accurate demonstrating the lowest AE across S-R intervals in the correct direction. Eighty-nine percent of the variance in end position was explained by a combination of planning (81% ± 2.9%) and updating (8.4% ± 1.8%) across preparation intervals. Across the three preparation intervals, a triphasic EMG burst pattern is evident when subject 2 moves to his default target (30s ) (Figure 11 - left panel). As expected, error is low to the default regardless of preparation time. The same activation pattern is seen to the long target amplitude when time to prepare is minimal (i.e., 0-99 ms). Averaged absolute error indicates that the subject actually approximated the default [45s (target) - 13°(AE) = 32s ]. Compensatory adjustments to the long target amplitude (Figure 11 - right panel) in the longest (300-400 ms) and middle (100-199 ms) preparation intervals appear to be implemented via the second agonist burst (triceps). Only in these two preparation conditions does the second agonist burst begin before the first agonist burst returns to baseline (see arrow) and appears somewhat prolonged to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 the end of the movement. Error was minimal for responses to both preparation intervals (AE = 2° for longest interval and 5° for middle interval). Subject 8 Onset Offset •200 ms 800 ms 100'/sec 20% 20% 100 ms Figure 10. Ensemble averages (±SEM) of the displacement, velocity, biceps, and triceps profiles for (n=18) responses to the 15s extension target in the shortest preparation interval. Like trials (same S-R interval and target) over the 15 unpredictable trial blocks were averaged. Averages were calculated over 1000 ms. For the displacement profile, the 15° target is shown (dot-dash line) in order to show the averaged displacement in relation to the actual target. Mean displacement, velocity, biceps and triceps are represented by the thin black lines in between two thicker lines which represents 1 standard error or the mean (SEM). Note: Low SEM related to responses demonstrates the consistency of the responses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 Control S2 S2 30' Extension Target 45’ Extension Target OffMt Onset Onset Offset AE 300-400 ms 100-199 ms Velocity 300-400 ms 100-199 ms Biceps 0-99 ms A A * 300-400 ms 100-199 ms Triceps ^ 0-99 ms ■ i 100 ms Figure 11. Ensemble averages of movement velocity, biceps, and triceps for a representative control subject (S2). Left panel are responses to the 30° extension (default) target; right panel are responses to the 45° extension target. The first, fourth, and seventh row of signals for both target amplitudes are averaged velocity, biceps and triceps profiles for the longest S-R interval (300-400 ms) in the unpredictable condition. The second, fifth, and eighth row of signals are averaged velocity, biceps and triceps profiles for a middle S-R interval (100-199 ms) in the unpredictable condition. The third, sixth and ninth row of signals are averaged velocity, biceps and triceps profiles from the shortest preparation interval (0-99 ms) in the unpredictable condition. The number of averaged responses for the 30° target were 29,19, and 12 from most to least preparation time. For the 45° target, response number was 30,17, and 11 respectively. Offset of movement, defined as return to zero velocity is represented as the range of offsets for the three preparation intervals. Note: averaged AE for the three preparation intervals for both target amplitudes. Arrows shown for 45° target, triceps EMG represent the only two preparation conditions in which the second agonist burst begins before the first agonist burst returns to baseline. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 Low-Adjustment S8 S8 1 5 'Extension Target 4 S > Extension Target Onset Otfut Onset OlltM AE 15 300-400 ms Velocity 100-199 ms 26 30 0-99 ms 300-400 ms 100-199 ms Biceps 0-99 ms 300-400 ms Triceps 0-99 ms 100 m s Figure 12. Ensemble averages of movement velocity, biceps, and triceps for a representative low-adjustment stroke subject (S8 ). Left panel are responses to the 15° extension target (default); right panel are responses to the 45° extension target. The first, fourth, and seventh row of signals for both target amplitudes are averaged velocity, biceps and triceps profiles for the longest S-R interval (300-400 ms) in the unpredictable condition. The second, fifth, and eighth row of signals are averaged velocity, biceps and triceps profiles for a middle S-R interval (100- 199 ms) in the unpredictable condition. The third, sixth and ninth row of signals are averaged velocity, biceps and triceps profiles from the shortest preparation interval (0-99 ms) in the unpredictable condition. The number of averaged responses for the 15° target were 27,16, and 18 from most to least preparation time. For the 45° target, response number was 25,20, and 17 respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 High-Adjustment S9 30' Flexion Target Onset ▼ Offsst ▼ S9 45' Flexion Target Velocity Biceps Triceps 100 ms Offset 100-199 ms V 0-9 300-400 ms 100-199 ms 0-99 ms 300-400 ms 100-199 ms 0-99 ms 300-400 ms I-99 ms Figure 13. Ensemble averages of velocity, biceps, and triceps for a representative high- adjustment stroke subject (S9). Left panel are responses to the 30° (default) flexion target; right panel are responses to the 45° flexion target. The row designation is as described in Figure 12. For both target amplitudes, responses for three S-R intervals are shown. The number of averaged responses for the 30s target were 23,21, and 19 from most to least preparation time. For the 45° target, response number was 22,17, and 16 respectively. In the longest preparation interval, the onset of the antagonist is delayed relative to the agonist burst (arrow), similar to what was seen to the default Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 Responses for a representative low-adjustment’ subject from the stroke group are seen in Figure 12. Subject 8 had the highest AE across preparation intervals and targets. In addition, compared with the other stroke subjects, this subjects’ percent variance in final position due to premovement planning was among the highest (72% - see Table 3). However, the percent variance explained by compensatory adjustments was the lowest (3%) of all subjects in the stroke group. Similar to that for control subject 2, there is a triphasic EMG burst pattern across all preparation intervals to the default target (15°) (Figure 12 - left panel). The activation pattern is nearly identical and accuracy is high. A nearly identical activation pattern is seen to the long target amplitude (45s ) for the middle (100-199 ms) and short (0-99 ms) preparation intervals, and final position is approximately the default amplitude (Figure 12 - right panel). It is only in the longest preparation interval that subject 8 ‘adjusts’ in order to capture the long target. Both the second agonist burst as well as the antagonist activation appear to be prolonged. The adjustment enables the subject to reach the middle (30s ) target but not the long (45s ) target. The total percent variance for a representative ‘high-adjustment’ subject from the stroke group (subject 9) is the same as for subject 8 , however, the apportionment is the reverse of that seen for subject 8 (Table 3). Subject 9 demonstrated one of the lowest percent variances due to premovement planning (55% ± 6 %), but had the highest percent variance due to compensatory adjustments (21% ± 4%) of all of the subjects in the stroke group. The implementation of compensatory adjustments appears to be through the timing and relative amplitude of the antagonist muscle (Figure 13). To the default target (30s ) across preparation intervals, an unopposed agonist burst (biceps) followed by co-contraction of agonist and antagonist (triceps) can be seen (Figure 13 - left panel). The same pattern is evident to the long (45s) amplitude target (Figure 13 - right panel) in the shortest (0-99 ms) preparation interval in which the subjects final position was approximately at the default. Accuracy is high to the long target amplitude for both the longest (300-400 ms) and middle (100-199 ms) interval responses. However, the time differences between those intervals appear to invoke different strategies. In the longest preparation interval, the onset of the antagonist is delayed relative to the agonist burst (arrow), Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 similar to what was seen to the default. The antagonist amplitude appears to be scaled in magnitude to the long target relative to the 30° target. For the middle preparation interval, increased antagonist amplitude is also evident and in addition co-contraction of agonist- antagonist is present from the very start of the response. Response Planning: Wrong Direction Extent Specification of Wrong Direction Responses Group differences. Similar to responses in the correct direction, the stroke group as a whole was significantly less accurate as target amplitude increased from short to long whereas the control group was most accurate to the middle target (Figure 14). This resulted in a significant group by target interaction (p = .0001). Thus it appears that default amplitude selection for each group is similar for correct and wrong direction responses. In addition, for the stroke group, wrong direction responses did not converge on the correct target extent as preparation time increased. Across preparation intervals, wrong direction responses were positioned near the short target extent (-20°) resulting in AE to the virtual target of 7.1° ± .38 (mean ± SEM) for responses to short targets, 10.0° t .39 for responses to the middle target and 21.3° i .38 for responses to long targets (Figure 15a). In contrast, control subjects scaled their response to the virtual long target amplitude as time to prepare increased. There was significant decrease in AE to the virtual long-amplitude target between the two middle preparation intervals (i.e, 100-199 and 200-299) and between the 200-299 ms interval and the longest preparation interval (300-400 ms), (P = .04 and P = .02). Although the control group scaled wrong-direction responses to the long target amplitude, that response was truncated (from -18° to - 1 0 ° between longest and shortest preparation interval) compared with that for correct direction responses (from ~17° to -5°). Mean AE to the virtual middle target amplitude remained relatively stable across S-R intervals at 8.3° 1 .49 (Figure 15b). Over the shortest to the second longest S-R interval (0 - 299 ms), mean AE to the virtual short target amplitude also remained relatively stable at -1 1 °. Upon Inspection of CE, it was identified that this was an 11 ° overshoot As such, final position was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 close to the default amplitude at - 26° when the 1 1 ° overshoot is added to 15° actual short target amplitude [11° (AE) + 15° (target) = 26°]. Mean AE to the short target for the longest preparation interval was not representative of the control group as a whole and as such, is not reported or shown in Figure 15b. Only 2 out of 6 control subjects that were inaccurate to the short target made up 91% of these wrong-direction responses in the longest preparation interval. Overall, the difference between the stroke and control group in extent specification of wrong direction responses resulted in a significant group by target by preparation interval interaction (p = .0 0 2 ). 25 i - a - Control 20 - & 3 15 - 10 - S lii < 5 © s short medium long Target Amplitude Figure 14. Symbols represent group means for AE (±SEM) to each target amplitude for wrong direction, unpredictable responses. The control group is represented by the triangle and the stroke group by the squares. Data are averaged across S-R interval 6 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0 0 0 1 ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 A Stroke 28 -| Control S-R Into m i (raa) Figure 15. a) Bar graph of mean AE (±SEM) to the virtual target for wrong direction responses within each S-R interval. Data are averaged across 6 stroke subjects for each target amplitude (short, medium, and long). Wrong direction responses did not converge on the correct target extent as preparation time increased. Across preparation intervals, wrong direction responses were positioned near the short target extent (-20°). b) Bar graph of mean AE (iSEM ) to the virtual target for wrong direction responses within each S-R interval. Data are averaged across 6 control subjects for each target amplitude (short, medium, and long). The control group demonstrated a significant decrease in AE to the virtual long-amplitude target between the two middle preparation intervals (i.e, 100-199 and 200-299) and between the 200-299 ms interval and the longest preparation interval (300-400 ms). Note that no data are presented for the short amplitude target in the longest preparation interval. Only 2 out of 6 control subjects had wrong direction responses for this target/S-R interval. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 Pkn 60 -| ■ Svota WA CbnM I I Compensatory AJutmems 0-99 100-199 200-299 300-400 S-R k m ra l pm) s 12 ■ 9 9 • 10 0 10 6 14 Ik a i V tru l AE (tkg) Figure 16. a) Frequency of direction errors as a percent of the total responses within each S-R interval for the stroke and control groups in the unpredictable condition, b) Stack bar plot of percent variance in displacement for unpredictable, wrong direction responses by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The control group is to the left, the high-adjustment stroke group is in the middle and the low- adjustment stroke group is to the right. Data are averaged across 6 subjects for the control group and 3 subjects per stroke sub-group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot c) Individual stroke subject scatter of percent variance in displacement due to compensatory adjustments as a function of AE. Each stroke subject is represented by a different symbol. Data were wrong direction responses and averaged across 6 (virtual) targets and 4 S-R intervals. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 Direction Specification Figure 16a shows the percentage of direction errors by preparation interval for each group. When preparation time was minimal (< 100 ms), the percentage of wrong direction responses was approximately 50% for both groups. When preparation time was greater than 300 ms, the frequency of wrong direction responses for all subjects was £15% of all responses within that S-R interval. The time course for direction specification was significantly different for the two groups (X2 = 15.20, p < .0001). The control group showed a 64% reduction in the frequency of direction errors between the 2"° longest (200-299) and longest (300-400) S-R intervals compared with 32% for the stroke group. The reduction in frequency of direction errors between the two middle S-R intervals (100-199 and 200-299) was 56% for the control and 46% for the stroke group. Only between the 100-199 ms interval and the shortest preparation interval (0-99) did the stroke group demonstrate a greater decrease in direction errors than controls (23% for the stroke group compared with 1 0 % for the control group). Model Analysis: Premovement Planning Group differences in the percent variance in final position due to premovement planning for wrong direction responses paralleled that for correct direction responses. The high- adjustment stroke group demonstrated significantly less percent variance due to premovement planning than the controls [M (SEM) = 51% (4.0%), M (SEM) = 64% (3.2%) respectively, P = .041]; while there was no difference between the control and low-adjustment stroke group [M (SEM) = 6 8 % (3.1%), P = .38] (Figure 16b - filled stack). Response Updating: Wrong direction Differences between the high-adjustment stroke group and controls in the percent variance in final position due to compensatory adjustments also paralleled that for correct direction responses. The ‘high-adjustment’ stroke group demonstrated a significantly greater proportion of variance due to adjustments than did the control group [M (SEM) = 11% (2.7%) and M (SEM) = 4.3% (.8 %), respectively, P - .008]. There were no differences between the low- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 adjustment stroke group and controls in percent variance due to compensatory adjustments (P = .24). An additional 2.5% ± 1.2% of variance due to adjustments explained final position achieved by the low-adjustment stroke group compared with the 4.3% ± .8 % for the control group (Figure 16b - open stack). Response Updating and Accuracy Direction as one parameter of the plan has clearly been incorrectly specified in wrong direction responses. Response updating and increased accuracy then refers to the degree to which the amplitude of the responses approximates the virtual target amplitude. Figure 16c demonstrates that the higher percentage of adjustments for subjects 9,10, and 11 (i.e., the high- adjustment stroke group) afforded a slight advantage in terms of approximating the virtual target amplitude compared with the remaining three low-adjustment stroke subjects. Absolute error across virtual target amplitude and preparation interval was lowest for the 3 high-adjustment subjects and reflects target-amplitude updating even when the response is in the wrong direction. DISCUSSION The importance of ipsilateral SM areas in response planning and updating was revealed here by decreased accuracy of aiming movements for individuals with stroke performing with the arm ipsilateral to the cerebral hemisphere lesion. This diminished accuracy was reflected by both greater errors in extent, (primarily to the long target amplitude), and a greater frequency of wrong direction errors compared to that of the control group. The statistically ‘significant’ accuracy differences between groups to the long target in the predictable condition was - 1 °, whereas the difference in the unpredictable condition was ~ 6 s and comparable to what was previously found using the 4-target task (Velicki et a!., 2000). Therefore, the 'meaningful' extent accuracy difference between groups was in the unpredictable condition where imperative planning was required. While the accuracy differences between groups paralleled those from our previous study (Velicki et al., 2000), there was a strategic shift in how subjects in both groups contended with the increased task complexity. These differences largely address the three task complexity factors Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 that prompted this investigation and will be discussed with respect to response planning and response updating. The role of ipsilateral SM areas within each of these control dimensions of accurate performance will be highlighted Response Planning One of the benefits of the TRP is that it enables inspection of the basic elements of the 'plan.' By forcing subjects to move when they have only partial or no target information, the response largely reflects what the subject has specified as a best ‘guess.’ In the unpredictable condition with minimal preparation time, the choice of default amplitude and direction is essentially what constitutes the ‘plan.’ The default amplitude is based on a subjects' prior experience with the range of target amplitudes and represents a central tendency bias or 'range effect’ (Poulton, 1981). Across a number of different experimental paradigms, subjects appear to prepare a predictive motor plan or default response based on prior knowledge. This default response allows for the preparation of key features prior to target presentation (Ghez et al., 1991). Providing a default value near the center of the range of target values and then tuning response magnitudes up or down as more information becomes available has been observed across a variety of motor tasks including: the production of isometric force pulses to match targets that vary in amplitude and direction (Gordon & Ghez, 1987a; Hening, Favilla, et al., 1988; Hening, Vicario, et al., 1988; Favilla et al., 1989; Favilla et al., 1990); responding to horizontal standing support-surface perturbations (Horak, Diener, & Nashner, 1989); and for grasping and lifting objects of unknown size and weight (Gordon, Westling, Cole, & Johansson, 1993) or objects whose weight unexpectedly changes (HSger-Ross, 1995; Johansson & Cole, 1994; Winstein, Horak, & Fisher, 2000). Regardless of task, when uncertainty limits the usefulness of predictive control strategies, it seems that the most efficient way for the nervous system to resolve the problem of performing an action is to assign the response a value that is distributed around the center of the range of possible values. When more information is available to the system with respect to perturbation parameters a more accurate response is obtained by simply increasing or decreasing responsiveness or gain around Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 the default response. The system will respond to deviations from the plan by selectively channeling sensory information which will determine gain and provide for rapid adjustments (Ghez et al., 1991). The control subjects responded to the increased complexity of the 6 -target task by generating an optimal plan. All subjects in the control group selected a default amplitude in the midrange of the three target amplitudes; a similar strategy was reported for the original TRP investigations (Favilla et al., 1989; Favilla et al., 1990; Ghez et al., 1997; Gordon & Ghez, 1987a; Hening, Favilla, et al., 1988; Hening, Vicario, et al., 1988). This is in contrast to the 4-target task in which all but 2 subjects selected the short amplitude target as the default (Velicki et al., 2000). Similarly, default amplitude for controls was in the midrange of the three targets even for wrong direction responses. This suggests that amplitude specification for the initial plan is independent of direction specification. For controls performing the 4-target task, the short target was again selected as the default (Velicki et al., 2000). The EMG activation patterns of the control subjects further substantiates selection of the middle target as the default amplitude. A clear triphasic EMG burst pattern indicative of premovement planning (Hallett et al., 1975; Wadman, Denier van der Gon, Geuze, & Mol, 1979) was evident in all of the preparation intervals in which the control subjects moved to their default (middle) target. Additionally, this same activation pattern was seen in the shortest preparation interval when the intended target was long but the actual response captured the default amplitude. Since there is too little time for updated information to affect the response, the response largely reflected what was planned in advance. The triphasic EMG burst pattern has been identified as reflecting the motor plan. However, it does not dissociate a plan constructed with ample preparation time versus one developed under time constraints. In the longest preparation interval, control subjects demonstrated this same activation pattern. In the longest unpredictable interval like in the predictable condition, the subject presumably had complete information about the target before Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 111 response onset. Thus, a primarily pre-planned EMG activation pattern should be characteristic of this longest preparation interval. Earlier it was hypothesized that the task complexity effect on planning would be evident in default amplitude selection and choice of strategy for contending with the 6 -target task. Support for this hypothesis was found in the control group data. A middle range default amplitude was planned and predicated a 'continuous' strategy of adjusting up or down to capture the long or short target amplitudes. It appears as though the stroke group as a whole was also able to take advantage of both prior knowledge of target as well as ample preparation time to develop a more accurate plan. As preparation time increased, the stroke group was able to use movement-cue information to improve response accuracy (Figure 6 a). However, deficits in premovement planning are particularly evident when preparation time was constrained. The degree to which the low-adjustment stroke group pre-planned their unpredictable condition responses was equivalent to that demonstrated by the control group (Figure 8 a, filled stack). There were no differences between control and low-adjustment stroke groups in percent variance in final position due to premovement planning. However, while the responses reflected the 'plan' for this subset of the stroke group, the plan clearly was suboptimal (Fisher et al., 2000). For this sub-group of the stroke subjects, the deficits in premovement planning pertain to the task complexity factor of 'default specification.’ Each of the subjects in the low-adjustment group selected the short amplitude target as the default. While these subjects became progressively more accurate to the long target as time to prepare increased, AE for the longest preparation interval was - 14° which roughly placed them at the middle target. A comparison to the - 5° AE for each target amplitude in the predictable condition strongly suggests that deficits in imperative planning accounted for the inaccuracy observed in the unpredictable condition. Absolute error to the short targets in Velicki et al., (2000) was similar between groups in predictable and unpredictable conditions. This suggested that subjects in the stroke group were as capable as control subjects of specifying the default extent in advance of the stimulus. An Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 alternative explanation is that the stroke group of Velicki et al., (2000) was merely as capable as controls at employing a 'discrete' strategy for the 4-target task that was used. The 6 -target task of the present study lends support for this alternative hypothesis. The low-adjustment stroke group were not as capable as controls of specifying the default amplitude. Selection of the short target as the default appears to have confined the low-adjustment stroke group to a 'discrete' strategy. After planning to the short target, the adjustments that this group then implemented, afforded improved performance to the middle target but not the long target. The low-adjustment stroke subjects appear to have ‘skirted’ the 6 -target task complexity effect by responding as if the task was only 4-targets. The deficits in premovement planning for the ‘high-adjustment’ stroke group were not related to default amplitude selection. Planning deficits for this sub-group of stroke subjects were manifested as 1 2 % less explanation of final position due to planning than was seen for the control group (Figure 8 a, filled stack). The EMG activation pattern demonstrated by the high- adjustment stroke subjects was generally one of agonist activation followed by agonist-antagonist co-contraction As discussed earlier, in the longest and shortest preparation intervals particularly to the default target, a pre-planned response would likely be represented as a triphasic burst pattern. As such, little evidence of preplanning was found in the EMG activation patterns of the high-adjustment stroke subjects However, the high-adjustment stroke group contended with the task by employing more of a 'continuous' strategy, comparable to the control group. All three of the high-adjustment stroke subjects specified a default around the middle range of possible target amplitudes. Similar to the control subjects, AE was low and relatively stable across preparation intervals for the middle target. As time to prepare increased AE to both the short and long target amplitude decreased and came within - 3° of the control group responses. Although significantly less percent variance in end point was explained by the ‘plan’ for this subset of the stroke group, their aiming accuracy was clearly enhanced by the capability to generate compensatory adjustments (see below). However, the optimal selection of default amplitude as part of the plan for the high- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 adjustment group was certainly one factor that contributed to an overall lower AE (7.4°± .12°) compared with that of the high-adjustment group (11.1 °± .25°). Had it not been for the contribution of the low-adjustment sub-group, the difference between the control group and the stroke group in overall mean AE would have been greater. Indirect evidence for the neuroanatomical substrate related to the premovement planning deficits in our stroke subjects comes from analysis of single cell recordings in the brain of behaving primates. Preparation-related neurons have been observed in primary motor cortex (M1), premotor cortex (PM) and supplementary motor area (SMA) (Alexander & Crutcher, 1990; Riehle & Requin, 1993; Riehle et al., 1994; Weinrich & Wise, 1982). More conclusive however, is the mounting body of non-invasive imaging studies in humans that identify bilateral preparatory activity in SM areas as tasks of sufficient complexity are performed (Kawashima et al., 1998; Rao et al., 1997). Using measurement of regional cerebral blood flow, Jahanshahi et al., (1995) found bilateral activation of SMA, PM and the ipsilateral dorsolateral prefrontal cortex in self-initiated versus externally triggered movements. They suggested this activation pattern in self-initiated unilateral hand movements reflected the additional requirement of 'what to do' decision making. Bilateral movement-related cortical potentials in SMA and the primary hand sensorimotor areas started earlier and were larger in a complex finger sequence compared with simple finger movements (Kitamura et al., 1993). The greater and earlier activation presumably represented the requisite planning of the more complex task. All six of the subjects with stroke had lesions affecting one or more of these areas implicated in the preparation or planning of goal-oriented movements (Table 1). Response Updating The relationship between compensatory adjustments and accuracy has clearly been established here. The model analysis was originally developed to determine whether compensatory adjustments add to preprogrammed specification of force impulses to achieve more accurately targeted responses. Previous work has shown that compensatory adjustments significantly contributed to the explanation of final force achieved (Gordon & Ghez, 1987b). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 However, the direct link between implementing adjustments and actual error reduction or increased accuracy had not been established. The high-adjustment stroke subjects generated an optimum plan by selecting a midrange default amplitude. However, considerably less of their end position was explained by the plan than for either the control or the low-adjustment stroke group. The ability to generate compensatory adjustments by the three subjects within the high- adjustment stroke group is clearly what accounted for their superior performance. Along a continuum of most to least accurate, the control subjects appear to rely heavily on a well- constructed plan with the additional ability to generate compensatory adjustments for even greater accuracy. Less reliance on an optimal plan developed by the high-adjustment stroke group necessitates greater use of compensatory adjustments for accuracy. The low-adjustment stroke group relied heavily on a faulty plan from which little modification was made. In addition to the fact that the low-adjustment stroke group was minimally capable of generating compensatory adjustments, they appear to differ from the other ‘groups' with respect to adjustment type. The TRP allows for examination of trajectory adjustments within the various preparation intervals thus enabling a classification of adjustment types. Identifying the type of adjustments that are being implemented provides insight into the control processes governing the behavior. In general, adjustments to the trajectory that occur during the response could be either feedback or feedforward. Feedback means that the adjustment depends on sensing an error (either through internal or external monitoring) and producing a correction. Feedforward means that the adjustment depends on getting updated information about the target. As such, there are two possible sources of error for which an adjustment would be necessary. The first is incomplete programming of the initial trajectory. An adjustment or correction is necessary because even though the individual knows what the target is, an error in the initial specification has been made. The second source of error is incomplete specification of the target. When a subject has reduced information about the target, a correction will be for updated target information. The first error type referred to as a programming adjustment, is by definition then, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 feedback in nature. Theoretically this type of adjustment will be seen in the predictable condition as well as in the longest preparation interval (300-400 ms) of the unpredictable condition. In these conditions, the subject should (in theory) have complete information about the target before response onset. In the shortest preparation intervals, feedforward adjustments should be minimal since there is too little time for updated information to affect the response. Therefore, in these three conditions (predictable; shortest and longest preparation interval), the effect of feedforward adjustments should be minimal and whatever adjustments are generated are most likely to arise primarily from feedback. The second error type referred to as a targeting error adjustment should be most evident in the middle preparation intervals (100-199 and 200-299) in which the subject has partial information. While the initial part of the response by a subject may reflect a 'guess,' halfway through a response as short as 2 0 0 ms in duration, the subject has had 200 to 400 ms to process information about the target, make use of that information and make an adjustment to the latter part of the response. There are two lines of evidence that would suggest that the high-adjustment stroke group and controls are generating feedforward adjustments while those in the low-adjustment group are generating feedback adjustments. For one, there was a significant effect of preparation interval on the percent variance due to compensatory adjustments for the control and high-adjustment group. Figure 8 c demonstrates that for those groups the pattern of change in adjustments took the form of an inverted 'U' with greater percent variance due to adjustments in the two middle preparation intervals. As stated earlier, these are the preparation intervals in which the subjects have partial target information and as such would utilize feedforward compensatory adjustments to modify their responses. Second, implementation of compensatory adjustments is evident in the EMG activation patterns. The second agonist burst (triceps) appears to be prolonged when the control subject moves to the long target in the middle interval (Figure 11 • right panel). For the high-adjustment stroke subject (S9), there is both earlier and higher amplitude antagonist activation to the long target in the same interval (Figure 13). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 In Figure 8 c, the low-adjustment stroke group demonstrates a pattern of change in percent variance in adjustments which reflects progressive use of feedback as time to prepare increases. Additionally, there is no difference in the EMG activation patterns for the low- adjustment stroke subject regardless of interval or target (Figure 12). Target amplitude specification of wrong direction responses was one of the task complexity factors related to updating that was addressed in the present study. The truncated or aborted extent specification of wrong direction responses demonstrated by the present study control and stroke groups was remarkably similar to what was seen in the study by Velicki et al., (2000). Therefore, neither of these arm-movement studies replicated convergence of wrong direction responses on the appropriate target amplitude observed in the isometric force pulses of the original TRP studies (Favilla et al., 1989). Movement time is approximately 3 times longer for the arm movement compared with the force pulse. Therefore, a likely explanation for the difference in wrong direction response specification is that subjects (both with and without stroke) performing an arm movement have ample time to identify movement in the wrong direction and essentially abort the response. The role of SM areas in response updating can account for the compensatory adjustment deficits observed in the low-adjustment group. A substantial role for ipsilateral primary sensorimotor cortex in formulating compensatory changes of task parameters during movement execution has been identified (Stephan et al. 1995). Schluter et al. (1999), studied the signal- set- and movement related activity in the human premotor and motor cortex by temporarily disrupting activation in those areas using transcranial magnetic stimulation (TMS). TMS pulses over M1 were most disruptive when they were delivered around the time of response execution. Similarly, ipsilateral M1 stimulation with TMS during performance of complex finger movements disrupted the timing of the finger sequence. It was suggested that the timing errors were due to the involvement of M1 in the execution of motor programs (Chen et al., 1997). Regional cerebral blood flow measures of ipsilateral M1-S1 confirmed an important role of these areas in the execution of complex sequential finger movements (Shibasaki et al., 1993). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 The increased complexity of the 6 -target task resulted in distinctly different strategies among the three groups (i.e., control and two sub-groups of the stroke group). The behavior of the low-adjustment stroke group is comparable to the stroke group of Velicki et al., (2000) with respect to premovement planning and compensatory adjustment capability (Fisher et al., 2000). Neither the lesion nor clinical data provided insight with respect to the strategic differences within the stroke group. However, the lesion data presented in Table 1 are gross measures of both location and volume and as such, lesion-specific differences within the stroke group cannot be ruled out. Earlier it was stated that only 2 stroke subjects (S9, S10) were able to synchronize movement initiation within a s 90 ms timing window. These were both high-adjustment subjects. The 3r a member of the high-adjustment group (S11) achieved the narrowest timing window of the remaining stroke subjects at 125 ms. Timing windows for S7, S8 , and S12 (i.e., low-adjustment stroke group) ranged from 130-140 ms. What enabled the high-adjustment stroke subjects to synchronize movement initiation with greater precision than the low-adjustment subjects is not clear. What is certain however, is that this ability to manage a smaller window required the high- adjustment subjects to initiate movement with an incomplete plan and utilize updated target information to generate feedforward compensatory adjustments. The divergence within the stroke group may be a function of the 'intentions’ or 'goals' of the different stroke subjects. Adam (1992) found that the objective of a movement or the performer’s intent had important consequences for the kinematic properties of the movement. Using a Fitts' type reciprocal aiming paradigm, subjects were identified as having an accuracy, speed-plus-accuracy, or speed bias. Despite uniform instructions to 'move as accurately and quickly as possible' subjects determined their own accuracy limits and movement speed and could readily be differentiated into one of the three ‘bias' groups. Subjects biased towards accuracy systematically moved slower by reducing peak velocity and spending more time in acceleration and deceleration than the speed-bias subjects (Adam, 1992). Additionally, Gordon and Ghez (1987a) observed substantially greater rise times when subjects performed targeted Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 force pulses within an accurate condition than when they performed in the fast condition. While one or another bias was not specifically assessed for the participants of this study, the high- adjustment group may have had an accuracy bias. Both total acceleration time and total deceleration time were - 40 ms greater for the high-adjustment stroke compared with the low- adjustment stroke subjects. Additionally, post-hoc assessment of velocity rise time for each of the three targets revealed longer rise times for the three subjects in the high-adjustment compared with the low-adjustment group (Table 4). Table 4. Velocity Rise Times for Stroke Subjects by Target Stroke Sub-Group Subject# Short (me) Medium (me) Long (me) 9 f119 ± 2.2 130 ±2.4 142 ± 3.0 High-Adjustment 10 142 ± 2.0 156 ±1.9 168 ±1.9 1 1 144 ± 2.6 148 ± 2.3 151 ±2.5 7 90 ± 2.1 97 ± 2.1 105 ±2.3 Low-Adjustment 8 8 6 ±1.4 94 ±1.8 100 ±1.9 12 117± 2.0 124 ± 2.0 134 ±2.4 t values are means (±SEM) for correct direction responses averaged across S-R interval Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 SUMMARY The importance of ipsilateral SM areas in response planning and updating was revealed here by decreased accuracy of aiming movements for individuals with stroke performing with the arm ipsilateral to the cerebral hemisphere lesion. This diminished accuracy was reflected by both greater extent errors, (primarily to the long-target amplitude), and a greater frequency of direction errors compared to that of the control group. In contrast to our previous work using a 4-target task (Fisher et al., 2000), the group difference in accuracy was not solely accounted for by stroke-related deficits in generating compensatory adjustments and updating the response. Rather deficits in both planning and updating were evident in the stroke group using a 6 -target task. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 CHAPTER 4 Experiment 3: THE ROLE OF THE CEREBELLUM IN THE PLANNING OF AIMING RESPONSE PARAMETERS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 Abstract The purpose of this experiment was to investigate the contribution of the lateral cerebellar hemisphere in the planning and updating of aiming response parameters. While the intermediate part of the spinocerebellum has been implicated in feedforward corrections (i.e., updating), the lateral cerebro-cerebellar hemisphere is thought to have a primary role in movement planning. The performance of five subjects with unilateral cerebellar (CB) lesions secondary to stroke were compared to that of matched control subjects participating in a timed- response movement paradigm. Subjects rapidly moved a horizontal manipulandum with flexion or extension movements to a short (15°), medium (30°), or long (45°) target presented in either a fixed (predictable condition) or random sequence (unpredictable condition). The CB subjects used the limb that was contralateral to the side of the lesion. Time to prepare the response was manipulated by varying the time of target presentation relative to an auditory cue for movement initiation (preparation interval between 0-400 ms). Velocity was derived from displacement data; kinematic analysis and multiple regression were used to determine the effect of premovement planning and trajectory updating on end-point accuracy. The CB group was significantly less accurate than the control group in both predictable and unpredictable conditions. Four key results implicate deficits in premovement planning as the source of response inaccuracy for the subjects with CB lesions: 1) Compared with the control group, the CB group demonstrated only minimal benefit from increased preparation (i.e., planning) time for target amplitude specification; 2) Even with ample time to plan the direction of the response (^300 ms), the frequency of wrong direction responses was -19% for the CB group compared with - 2% for the control group; 3) the percent variance in final position explained by premovement planning was significantly less (83 ± 1% predictable, 6 0 1 2% unpredictable) for the CB group compared with controls (90 i 0.8% predictable, 73 ± 3% unpredictable) in both conditions; and 4) little evidence of a motor plan was found in the EMG activation patterns of the CB subjects. For control subjects only, a triphasic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 muscle activation pattern was observed in both predictable and long-preparation-interval unpredictable condition responses. These data support a role for the lateral cerebellum in the planning of rapid goal-directed aiming actions, even for the contralateral limb. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 INTRODUCTION The purpose of this experiment was to investigate the contribution of the lateral cerebellar hemisphere in the planning and updating of aiming response parameters. It has long been considered that the cerebellum is involved in the control of movement (Middleton & Strick, 1997). However, the specific motor control processes supported by the cerebellum still remains unclear (Jueptner & Weiller, 1998). The cerebellum has been implicated in both the specification (i.e., planning) and updating of accurate movements (Hallett et al., 1991). A model proposed by Allen and Tsukahara (1974), identifies a role for the cerebellum in both generating the initial plan and updating the response. However, these distinct motor control processes (i.e., planning and updating) appear to operate from functionally distinct regions within the cerebellum. The lateral cerebellum processes information originating from motor, premotor (PM), and supplementary motor (SMA) regions as well as primary somatosensory cortex (Schmahmann, 1997). This processing is critical for planning movement and preparing the motor systems to act. Eventually, the processed information forms the commands for movement issued by the lateral cerebellum and the basal ganglia to multiple cortical areas, including PM, SM A, primary motor and prefrontal cortex (Middleton & Strick, 1997; Sakai et al., 1998; Sakai et al., 1996; Tarkka et al., 1993). These motor areas execute movement and inform the spinocerebellum of the ongoing commands. In turn, the spinocerebellum then corrects for errors that have occurred or compensates for impending errors in the commands for movement. Thus, a form of specialization within the cerebellum has been proposed whereby the dentate nucleus might be more involved in movement preparation and the interpositus nucleus linked more to movement execution (Bonnefoi-Kyriacou, etal., 1995). It would seem that greater ‘clarity’ as to the role of the cerebellum in the control of movement is achieved by identifying these functionally distinct areas of the cerebellum. However, there exists conflicting reports as to the primary function within these functional regions of the cerebellum. A number of investigations have identified that the lateral cerebellum Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 (neocerebellum) contributes to premovement planning (Wessel et al., 1994,1996). Conversely, other studies implicate the iateral cerebellum as having a role in generating feedforward modifications and updating the response (Sakai et al., 1998) (Jueptner, Flench, Weiller, Mueller, & Diener 1996; Jueptner, Jenkins, Brooks, Frackowiak, & Passingham, 1996; Jueptner & Weiller, 1998) A number of investigations with primates have identified a role for the cerebellum in the planning of goal-oriented actions (Sakai et al., 1998). Single cell recordings were made from the ‘cerebellar thalamus' while monkeys performed volitional wrist movements. Specifically, these recordings were from the ‘motor’ thalamic nuclei that receive cerebellar afferents projecting from dentate nucleus. In response to perturbations applied just prior to and during movement, modification of the short latency response of cerebellar thalamic neurons suggested that predictive information was being relayed to the motor cortex via the cerebellum. Thus, the dentate nucleus contributes to accurate performance of the ‘intended’ response by providing predictive information to the motor cortex (Home & Butler, 1995). Similarly, the activity of single cells in the cerebellar and motor cortex of awake monkeys was recorded in a series of studies involving both whole-arm reaching movements and application of force -pulse perturbations to hand-held objects (Espinoza & Smith, 1990; Picard & Smith, 1992; Smith et al., 1993). The cerebellar activation appears to reflect purkinje cells recorded from both the lateral and intermediate part of the hemisphere. In general, these studies revealed the contribution of the cerebellum in specifying parameters of the upcoming response such as distance and force. Similar to what has been seen in motor cortex, preparatory neural coding of direction was observed in cerebellar cortical neurons as the animals performed whole- arm reaching. The activity of single cerebellar cortical cells was broadly tuned around a “preferred direction” (Smith et al., 1993). In grasping, lifting and holding an object with a precision grip, the discharge rate of cerebellar cortical neurons reflected the object’s friction, weight or texture. These effects were not only observed during lifts, but also prior to the application of grip and load forces suggesting the establishment of preparatory motor control Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 strategies by the cerebellum (Dugas & Smith, 1992; Espinoza & Smith, 1990). Finally, when the monkeys were required to maintain grasp of an object that was subjected to predictable and unpredictable load force increases, grip force automatically increased with latencies as short as 50-100 ms. Both the motor cortex and the cerebellum seemed to be involved in this control. Specifically, during more predictable loading conditions, a preparatory increase in the 'background' grip force was observed along with changes in cerebellar activity, giving further support to a role for the cerebellum in preparatory parameter control (Dugas et at., 1992; Picard & Smith, 1992; Smith etal., 1993). Studies with human subjects that have demonstrated the contribution of the cerebellum to planning have measured both movement-related cortical potentials (MRCPs) (Kitamura, Shibasaki, Terashi, & Tashima, 1999; Tarkka et al., 1993; Wessel et al., 1994) and regional cerebral blood flow (Krams, Rushworth, Deiber, Frackowiak, & Passingham, 1998; Verlegeret al., 1999). Both the contingent negative variation (CNV) and the Bereitschaftpotential (BP), measures of cortical preparatory activity (MRCPs), were compared in healthy control subjects and in individuals with cerebellar damage. For sequential and goal-directed finger and arm movements (Wessel et al., 1994) (Tarkka et al., 1993) as well a complex coordination tasks (Kitamura et al., 1999), the cerebellar contribution to the preparatory state of the cerebral cortex was evidenced by markedly reduced MRCP amplitudes close to movement onset. (Tarkka et al., 1993; Wessel et al., 1994) Additionally, regional cerebral blood flow measures of healthy adult subjects preparing and executing hand movements demonstrated set-related activity bilaterally in the cerebro- and spinocerebellum. Set activity is thought to reflect motor planning and programming (Krams et al., 1998). The studies that support a role for the cerebellum in generating compensatory adjustments and updating responses, suggest that the neocerebellum is involved in monitoring and optimizing movements using sensory feedback (Grill, Hallett, Marcus, & McShane, 1994, Grill, Hallett, & McShane, 1997; Home & Butler, 1995). As such, errors in movement from cerebellar damage are the result of impaired processing of sensory information. In a series of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 studies using positron emission tomography (PET), Jueptner et al., concluded that the lateral cerebellum is only minimally related to movement planning. They identified a role for bilateral cerebellar hemisphere and vermis in sensory information processing. Specifically, increased activation in those areas were seen when healthy subjects were required to estimate time differences between a test and standard interval (Jueptner et al., 1995); compare the velocity of paired moving stimuli ‘drawn’ on the dorsum of the right hand (Jueptner et al., 1996); re-trace previously generated lines versus generate new lines (Jueptner et al., 1996) and let their arms to be passively moved (Jueptner et al., 1997). These results suggested that the neocerebellum optimizes movement by monitoring sensory information related to movement outcome. Similar conclusions were drawn by Grill et al., (1994,1997) after testing the ability of patients with cerebellar degeneration to perceive differences in kinaesthetic stimuli. The performance of individuals with cerebellar degeneration was significantly worse than that of normal subjects on tasks which tested duration and velocity perception. The investigators concluded that the cerebellum uses or facilitates the use of sensory signals during on-line adjustments in motor output (Grill et al., 1994,1997). Task differences may account in part for the contradictions discussed above. Those studies that have identified a role for the lateral cerebellar hemisphere in premovement planning utilize goal-directed motor tasks requiring precision. Conversely, in those studies that have determined a role for the cerebellum in feedforward updating the tasks used appear to have high- perceptual, low- movement accuracy demands. Additionally, with few exceptions, the individuals with cerebellar damage that have been utilized in these studies have widespread and bilateral cerebellar involvement from a variety of pathological disorders including: cortical cerebellar atrophy, olivopontocerebellar atrophy, multiple sclerosis, spinocerebellar degeneration, and dysynergia cerebellaris myoclonica. Kitamura et al., (1999) represents the one study in which MRCPs preceding voluntary movement were recorded in 5 cases with localized cerebellar lesions due to stroke or tumor. The MRCPs were absent or markedly reduced in amplitude in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 3 cases with dentate nucleus involvement, whereas the MRCPs were present in the 2 cases with no evidence of dentate lesion. The timed-response paradigm (TRP) used in the present investigation allows for examination of the time-course over which target parameters are specified during the preparation interval prior to the initiation of movement (i.e., the parameter specification process). By forcing subjects to move when they have only partial or no target information, the response largely reflects what has been planned in advance. These responses can be compared to responses generated when target information is provided early enough for feedforward adjustments to influence the outcome. As such, it will be possible to determine the distinct contribution of planning and updating to the accuracy of final position achieved. In addition, the percent variance in final position explained by the separate contributions of preplanning and response updating will be determined with a modified statistical model (Gordon & Ghez, 1987b). By partitioning response trajectories into the planned and adjustment components, the model analysis provides insight into the role of the cerebellum in either or both of these motor control processes. In summary, the purpose of the present investigation was to determine the role of the cerebellar hemisphere in planning and updating goal-directed aiming movements. The cerebellum overall has been implicated in both the specification of the initial plan for the action as well as updating the response. However, numerous reports identify a role for the lateral cerebellar hemisphere in the planned trajectory (primary path) of the component model. As such it is hypothesized that subjects with cerebellar stroke will be less accurate than healthy control subjects and that response planning deficits will account for the increased error observed. Planning deficits will be evidenced by a greater magnitude of amplitude (extent) errors and a greater frequency of direction errors than healthy control subjects regardless of the time available to generate the plan for movement. Additionally, there will be significantly less variance in final position due to premovement planning for the cerebellar group compared to the control group under both predictable and unpredictable conditions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 METHOD Subjects Five neurologically healthy control (C) subjects recruited from the Greater Los Angeles area (age range = 23-73 years) and five subjects with unilateral cerebellar (CB) hemisphere lesions (26 - 75 years), participated in this study. Table 1 summarizes the characteristics of each subject as well as lesion information for each CB subject. Unilateral cerebellar lesion was confirmed from MRI or CT scan reports. The site and size of the brain lesion was determined from the actual radiologic image. A board-certified neurologist outlined the lesioned areas from the scans. Lesion location was established primarily to confirm unilateral cerebellar hemisphere involvement. The lesion site was mapped onto one of 6 horizontal axial sections (Figure 1A) taken from the rostral to the caudal cerebellum (Amarenco, 1991). Matching lesion site to the appropriate section displayed in Figure 1A was accomplished by identifying common neuroanatomic landmarks (e.g., fourth ventricle; cortical poles) between the actual MRI or CT scan and the six illustrated levels. The greatest extent of each lesion is displayed in Figure 1B. As in Experiment 2, the Scion Image © program was used to measure cerebellar and lesion area (see Experiment 2 for details of Scion Image algorithm). Total cerebellar volume and lesion volume were determined by multiplying the measured area times thickness of the MRI or CT slices. By summing across ail slices, total cerebellum volume as well as total lesion volume were determined and the percent lesion volume calculated (Table 1 ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Subject 6 C w Subject 7 Subjects Subject 9 Subject 10 Figure 1. A) MRI or CT horizontal axial sections from the rostral to the caudal cerebellum. The sagittal view indicates the section's level. The horizontal levels progress from a-f (rostral to caudal cerebellum). Taken from: Amarenco, P. (1991). The spectrum of cerebellar infarctions. Neurology, 41:973-979. B) Lesion location for each cerebellar-group subject from magnetic resonance imaging (MRI) or computer tomography (CT) images. Each section shows the lesion- site of greatest extent using one of 6 horizontal axial sections taken from the rostral to the caudal cerebellum. The letter to the left of each lesion represents the horizontal section of Amarenco, 1991 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 Each subject in the cerebellar group demonstrated some motor impairment that affected their functional mobility. All of the CB subjects were able to ambulate independently at a household or limited community level. Details of specific impairment and functional limitation measures are described below (see procedures) with assessment results provided in Table 2. Instrumentation The apparatus, task and training have ail been previously described in detail (see method section, Experiment 2). As such, we now describe these aspects of the methods more briefly. Data were acquired using an instrumented manipulandum affixed to a wooden support and elevated to shoulder height. A computer monitor displaying a graphic representation of the target locations was positioned in front of the subject at eye level and at a distance of - 40 cm. All subjects could comfortably read the messages displayed on the monitor. Angular displacement of the lever was sampled at 2500 Hz. Each trial was 2000 ms in duration and consisted of A-D sampling of the lever potentiometer beginning 100 ms before the movement cue (see task for definition of movement cue). Custom ASYST software (Nakai, 1997) was used for the control of data acquisition, data analysis and storage. Electromyographic data (EMG) were acquired simultaneously with the kinematic data. Surface EMGs of biceps and triceps muscles were also sampled at 2500 Hz. Signal amplification and filtering parameters were identical to those previously employed (see methods section, Experiment 2). Task The task involved moving the lever from the home position to one of six target locations that were differentiated by two direction and three amplitude parameters (see Figure 2, Experiment 2). Subjects flexed or extended their elbow to match a target positioned at either 15s , 30s, or 45s from home. Subjects initiated movement towards one of the six targets in synchrony with an auditory cue. Specifically, the onset of lever movement was to be timed with the last in a series of 4 tones. The tones each had a 50 ms duration, and were spaced by an inter-tone interval of 500 ms. The movement cue was presented at variable times between the 3 "* and 4t h tones. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1. Subject and Lesion Characteristics Sub ID Sub- Match Sex Age <yr») Hand- Dominance Hand Used Etiology Lesion Side Lesion Duration (month) %Lesion Volumet Lesion Location Control 1 6 M 73 R L - - - - 2 7 M 6 6 R R - - - - 3 8 F 48 R L - - - - 4 9 F 30 R R - - - - 5 1 0 M 23 R R - - - - CB 6 M 75 R R hemorrhage L 6 2 Intracerebellar hematoma left cerebellar hemisphere secondary to a hemorrhagic infarct; intermediate hemisphere appears to be site of lesion 7 M 62 R R infarct L 7 1 0 Left cerebellar infarct in the posterior inferior cerebellar arterial distribution; intermediate and lateral hemisphere involvement Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 8 M 51 R R infarct L 9 M 35 L L hemorrhage R 10 M 26 R R hemorrhage L f% Lesion Volume = total cerebellar lesion volume + total cerebellar volume x 100 Table 1. Subject and Lesion Characteristics (continued) Left cerebellar infarction; intermediate and lateral hemisphere involvement Right cerebellar A-V malformation; intermediate and lateral hemisphere involvement Left cerebellar hemorrhage secondary to A-V malformation; intermediate hemisphere involvement Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 2. Impairment and Functional Limitation Measures of Cerebellar Subjects Impairment Functional Limitation Disability Subject # tcB assess ‘Grip - C “ Grip -1 "Box & Block - C ""Box & Block -1 ♦f im Motor ~F IM Cognitive 'SF-36 6 4 62 65 52 52 91 35 28 7 8 99 81 74 26 87 35 2 2 8 8 1 2 0 76 65 48 87 33 24 9 17 71 63 51 2 0 83 35 26 1 0 3 71 59 74 67 91 35 30 tClinical rating scale for cerebellar function (range 0 -4 6 points; increased severity with increased number) ‘Grip - C = grip strength of contralateral hand (contralateral to side of brain lesion) “ Grip - 1 = grip strength of ipsilateral hand (ipsilateral to side of brain lesion) "Box & Block - C = dexterity measure for limb contralateral to brain lesion - # of blocks moved in 1 minute ""Box & Block - 1 = dexterity measure for limb ipsilateral to brain lesion - # of blocks moved in 1 minute *FIM Motor = the motor sub-scale of the Functional Independence Measure (total possible = 91) **FIM Cognitive = the cognitive sub-scale of the Functional Independence Measure (total possible = 35) 'SF-36 = physical functioning subscale of the SF-36 is a disability outcome measure (0-100 point scale) 134 Procedures All subjects read, signed and received a copy of the Institutional Review Board approved informed consent form. To determine the homogeneity between groups (CB, control) for visual, auditory, cognitive and motor function status, all subjects were assessed with the following measures: (1) Visual acuity (Rosenbaum Pocket Vision Screener), (2) peripheral vision of each eye (visual field), (3) hearing (4) attentional capabilities, (5) grip strength, and (6 ) manual dexterity (Box and Block test - Mathiowetz, et al., 1985). For the subjects with Cerebellar lesions, function was quantified on a clinical rating scale developed by Massaquoi and Hallett (Wessel et al., 1995). Additionally, the mobility sub-scale of the Functional Independence Measure (Hamilton et al., 1987) and the physical functioning sub-scale of the SF-36 were used (Ware & Sherboume, 1992) All subjects in the CB group performed the task with the arm contralateral to the CB lesion. For all CB subjects, this was their dominant arm. Two control subjects had previously been matched to two individuals with unilateral SM area strokes that had used their non dominant arm to perform the task (see Experiment 2). As such, those two control subjects (S1 and S3) used their non-dominant arm. The remaining 3 control subjects performed the task with the dominant arm (Table 1). Subjects in both groups that used the right arm to perform the task sat to the left of the lever and grasped the lever handle with their right hand. Subjects in both groups performing the task with their left arm used the opposite configuration. Seat height was adjusted for each subject to position the shoulder at approximately the same height as the lever. This assured that lever movement was accomplished by elbow flexion and extension with little if any contribution of shoulder internal and external rotation. Subjects were oriented to the lever movements, the target display and three practice tasks (see Experiment 2 for details of the practice tasks). The practice tasks provided exposure to the basic requirements of the test task by having the subjects: 1 ) accurately locate and move to the six target locations without any time constraints; 2 ) initiate movements within a ± 60 ms Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 timing window around the 4th tone and a constant movement cue presentation (i.e., preparation time) of 200 ms before the 4t h tone; and 3) initiate rapid movements within the timing window using variable preparation times between the 3r d and 4t h tones. All subjects became familiar with moving the lever, locating the targets and contending with the timing demands of the TRP in s 30 minutes and progressed to the testing period. During the testing period, the movement cue was randomly presented between the third and the fourth tone within one of four preparation time bins (1 = 300-400 ms, 2 = 200-299 ms, 3 = 100-199 ms, and 4 = 0-99 ms). For each day of data collection, electromyograms were recorded from the biceps muscle and the lateral head of the triceps using surface electrodes. Skin preparation and normalization procedures were identical to those described in Experiment 2. Data acquisition. As in Experiment 2, preparation time was controlled by the stimulus-tone (S-T) interval (the time between movement cue presentation and the onset of the 4th tone). The stimulus- response (S-R) interval reflected the preparation time actually used by the subject. The timing error was the difference between the imposed S-T interval and the actual S-R interval (see Figure 3a, Experiment 2). Each subject was required to complete 8 blocks (48 trials/block) of a predictable condition (targets presented in a fixed order) and 15 blocks of the unpredictable condition (targets presented randomly). Acceptable block criteria included; 1) a non-significant timing correlation (p s .23) of preparation time and timing error indicating that responses were being timed with the 4th tone and not systematically varied with respect to preparation time and, 2) a significant correlation between Absolute error (AE) and ST interval indicating that responses were more accurate when target information was provided early and less accurate when target information was given later with respect to the 4m tone. Regardless of timing correlation, once the timing window was less than 100 ms, all completed blocks were accepted for analysis. Timing windows of £ 95 ms (including the tone) were achieved by 5 control subjects (range across control subjects; 60-95 ms) and 2 cerebellar Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 subjects (range across cerebellar subjects: 75-155 ms). In addition, for the cerebellar subjects, 67% of the accepted unpredictable-condition blocks had significant accuracy correlations of r > .20 compared with 99% for the control subjects. To complete data collection, cerebellar subjects returned to the laboratory between 5-10 days. Subjects in the control group returned to the laboratory between 5-7 days. All subjects performed between 3 and 9 blocks/day. To produce a full complement of acceptable test trials (i.e., 1104 trials with specific S-T interval, target characteristics within the predictable and unpredictable conditions), each subject performed between 24 and 33 blocks. Data Analysis Data Management and Screening The 24 trial types (6 targets x 4 ST intervals) within predictable and unpredictable conditions were combined for analysis. Displacement and EMG data processing was identical to that described in Experiment 2. 'Anticipation' error trials (i.e., negative S-R interval) resulting from movement initiation prior to target cue were eliminated from analysis. These represented 4.7% of all trials for the CB group and 1.9% of all trials for the control group. Additionally, the kinematic algorithm detailed in Fisher et al., (2000) was applied to each individual trial. As such, all single peaked velocity waveforms were considered for analysis. Double response trials were rejected or salvaged based on the criteria outlined in Fisher et al., (2000). For the cerebellar group, 0.3% of both predictable and unpredictable trials were eliminated from analysis based on the kinematic criteria. Only 0.2% of both predictable and unpredictable trials were eliminated for the control group. After the screening procedure, the full complement of predictable condition trials consisted of 334-380 trials/subject (total n = 1817) for the cerebellar group and 374-381 trials/subject for the control group (total n = 1868). In the unpredictable condition, the number of analyzed trials for the cerebellar group was 332-527 trials/subject (total n = 2113) and 471-586 trials/subject (total n = 4246) for the control group. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 Dependent measures and statistical analyses As in Experiment 2, the algorithm to determine movement onset and offset was a change in velocity from zero in the appropriate direction of movement. The position and velocity profiles for each trial within the above described event period were used to calculate the following variables (see Figure 4 Experiment 2): 1) final position achieved [Displacement = offy - ony (a)], 2) average velocity from the onset of movement (b) until peak velocity, (c) 3) movement time = off, - on,, and (d), 4) velocity rise time = c, - bt. Additionally, error was also computed as the difference between final position and the target. Parameter Specification Analysis The dependent measures and statistical procedures marking the differences between the CB and control groups in the time-course of parameter specification were identical to those described in Experiment 2:1) Constant error (CE); 2) Absolute error (AE) and 3) Direction error. Component Model Analysis The components of the statistical model that were used to determine the variance in final position due to premovement planning and compensatory adjustments are detailed in Fisher et al., (2000). As in Experiment 1 and 2, the dependent variables used in this analysis were the regression coefficients representing variance explained by preplanning (/■% ,), and compensatory adjustments due to target ( ^ y (2 -^y() (see Experiment 1 for detailed description of the statistical model). Additionally, the difference between the CB and control groups in variance of final position explained by preplanning and compensatory adjustments was determined by a 2 group (CB, control) x 2 response directions (correct, wrong) x 4 S-R intervals (4=300-400,3=200-299, 2=100-199,1=0-99) mixed model Analysis of Variance (ANOVA). For all statistical tests, significance was set at p < .05. Post-hoc orthogonal contrasts with Bonferroni correction were performed to determine the locus of any significant main and interaction effects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 Individual subject analyses Ensemble averages of individual subject agonist-antagonist EMG burst patterns were inspected to explore different strategies used in task performance. Over (approximately) the 30 like trials across blocks (i.e., same S-R Interval/Target/Response Direction/condition) biceps and triceps EMG signal averages were computed separately for each muscle (see Experiment 2 for details of EMG averaging parameters. RESULTS Predictable condition responses are described first for all subjects and then between group comparisons. This is followed by a description of the unpredictable condition responses. Group differences that highlight the contribution of the cerebellar hemisphere in response parameter planning and updating are presented for both correct and wrong direction responses. Comparison of means between same-amplitude flexion and extension targets revealed that AE was similar for targets of the same extent P > .05. Therefore, for all analyses, responses to each of the 3 similar-extent targets were combined. Target effects were addressed with respect to the short (15°), medium (30°) and long targets (45°). Predictable Condition Responses All Subjects Across the three target amplitudes ail subjects achieved their highest degree of accuracy in the predictable condition. Additionally, all subjects preplanned their responses before the movement cue. As such, response accuracy, was not dependent on preparation time (Figure 2A). The capability under predictable conditions to plan the response in advance was confirmed by the model analysis. All subjects demonstrated the greatest percentage of variance due to premovement planning in the predictable condition compared to the unpredictable condition (see below) suggesting they used target information in advance to plan their response. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 Group Analyses The CB group was less accurate than the control group across all targets (short, medium and long) [AE M (SEM) = 4.7° (.09) for the CB group; 2.9° (.05) for the control group, P < .0001]. The CB group was least accurate to the long target with a similar degree of accuracy to the short and middle amplitude targets. The control group was most accurate to the middle target with a similar degree of error to the short and long amplitude targets (Figure 2B). This resulted in a significant target by group interaction (P < .0001). In addition, for each target amplitude, the CB group was significantly less accurate than the control group (P < .0001). There was a significant difference in premovement planning capability between the two groups. For the CB group 83% ±1.1% (mean ± SEM) of the variance in final position was explained by premovement planning compared with 90% ± .8 % for the control group (P < .0001) (Figure 2C -filled stack). An additional 8.2% ± .8 % of the variance in end position was explained by adjustments compared with 5.5% ± 1% for the control group (Figure 2C - open stack). The percent variance in final position explained by adjustments was significantly greater for the CB group compared with controls (P=.013). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 tO(M99 200-290 S*R Irrtaraol (m ) Figure 2. A) Bar graph of group mean AE (±SEM) within each S-R interval for predictable condition. Data are averaged across 5 subjects/group and 6 target amplitudes. Control group = striped bars; cerebellar group = dark bars. B) Symbols represent group means for AE (±SEM) to each target amplitude for predictable condition responses. The cerebellar group is represented by the triangle and the control group by the squares. Data are averaged across S-R interval 5 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0001). C) Stack bar plot of percent variance in displacement for predictable trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The cerebellar group is to the left, the control group is to the right Data are averaged across 5/group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 Unpredictable Condition Responses Response Planning: Correct Direction Amplitude Specification Across ail targets and preparation intervals, the cerebellar group was significantly less accurate [M (SEM) = 10.0° (.12)] than the control group [M (SEM) = 7.2° (.11)], P < .0001]. As the S-R interval increased, both groups were able to use movement cue information to improve response accuracy. However, the time course for extent specification was significantly different for the two groups (P < .0001). It can be seen in Figure 3 that the cerebellar group was less accurate than the controls across each S-R interval. Post-hoc comparisons revealed that indeed, the accuracy differences between groups were significant within each S-R intervals (P < .0001 for S-R intervals 100-400ms; P < .04 for the shortest interval). Neither group benefitted from increased preparation time 2300 ms. However, the control group decreased error by 4-5° over successive S-R intervals between 0-300 ms, compared with 1-2° decreases for the cerebellar group. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 y/ A Control 0-99 100-199 200-299 300-400 S-R Interval (ms) Figure 3. Bar graph of group mean AE (iSEM ) within each S-R interval for unpredictable -correct direction responses. Data are averaged across 5 subjects/group and 6 target amplitudes. Control group = striped bars; cerebellar group = dark bars. For the cerebellar group, extent specification largely involved improved performance to the long target amplitude (Figure 4A). There was a significant decrease in AE for the cerebellar group across all four preparation intervals (post-hoc P values between intervals ranged from .0001 to .03). Absolute error to the short target was significantly greater in only the shortest preparation interval relative to the remaining intervals (P < .009); while no change in AE was seen to the middle target across S-R intervals. For the control group, both long and middle amplitude targets were specified between 0 and 299 ms with no further improvement in accuracy beyond 299 ms (Figure 4B). Absolute error to the short target remained relatively stable across preparation intervals for the control group. These accuracy differences between groups, S-R interval and target amplitude resulted in a significant 3-way interaction (P < .0001). Only in the shortest preparation interval was AE similar between groups for the long and middle targets (see Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 figure 4A & B). For every other S-R interval-target amplitude pairing, the cerebellar group was significantly less accurate than controls (post-hoc P values between target/interval pairs ranged from .0001 to .05). Individual Subject Planning: Default Amplitude Selection Absolute error as a function of target amplitude only, reveals that both the cerebellar and the control groups were equally accurate to the short and middle targets (Figure 5). This is largely due to the fact that there was not a consistent default amplitude selection for subjects within each group. For three of the cerebellar subjects the default amplitude was close to the middle target (~ 25°). Whereas, the remaining two cerebellar subjects selected the short target amplitude as the default. Four of the control subjects selected a default amplitude between the short and middle targets. The remaining control subject selected the short target as the default. Figure 6 illustrates individual subject default selection for two control subjects and two cerebellar subjects. Many of the short target responses for subject 7 (cerebellar • A) and subject 5 (control • B) have a constant error (CE) of approximately zero degrees in the earliest S-R intervals. This suggests that these subjects specified the short target amplitude as part of the premovement plan (i.e., the default amplitude). For both subjects 8 (C) and 1 (D) in Figure 6 , the default amplitude is closer to the medium target. Error for the medium target in the earliest preparation period hovers around zero degrees for both of these subjects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Cerebellar I L U < c « £ m 0-99 100-199 200299 300400 B Control 0-99 100-199 200299 300400 S-R Intend (ra) Figure 4. A) Bar graph of mean AE (±SEM) for unpredictable-correct direction responses within each S-R interval. Data are averaged across 5 cerebellar subjects for each target amplitude (short, medium, and long). For the cerebellar group, extent specification largely involved improved performance to the long target amplitude There was a significant decrease in AE for the cerebellar group across all four preparation intervals (post-hoc P values between intervals ranged from .0001 to .03). B) Bar graph of mean AE (±SEM) for unpredictable-correct direction responses within each S-R interval. Data are averaged across 5 control subjects for each target amplitude (short, medium, and long). For the control group, both long and middle amplitude targets were specified between 0 and 299 ms with no further improvement in accuracy beyond 299 ms. Absolute error to the short target remained relatively stable across preparation intervals for the control group. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Control 16 i 12 - at 2 L U < § a > 8 4 - “ A - Cerebellar short medium Target Extent long Figure 5. Symbols represent group means for AE (±SEM) to each target amplitude for unpredictable-correct direction responses. The cerebellar group is represented by the triangle and the control group by the squares. Data are averaged across S-R interval and 5 subjects/group. Lines connecting symbols across the 3 target amplitudes demonstrate the significant target by group interaction (P < .0 0 0 1 ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 I o t . * —I— 1 0 0 ~r~ 3 0 0 “ I— M O - r - 400 “ I 800 Oontnl Sdjat 8 S R latrul [ a n ) 1 0 0 aw 300 4 1 0 S-Rlntov4tm) O tw btU r ■ Ooitoi S r f ja x i & R latral [a^ 3-Rlnle(V^(ms) Figure 6 . Scatter plot of constant error (CE) across S-R intervals for a full complement of correct direction, unpredictable condition responses. Regression lines are run through each target amplitude. A thicker horizontal line at 0s represents the target. For cerebellar subject 7 (A), and control subject 5 (B), the short (15s ) target amplitude was the default For cerebellar subject 8 (C) and control subject 1 (D), the default position was the medium target (30°) amplitude. Note regression line for medium target for S8 is superimposed on 0°. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 Model Analysis: Premovement Planning A significantly greater percent variance in final position due to premovement planning was demonstrated by the control compared with the cerebellar group. Seventy-three percent ± 2 .8 % of the variance in final position was explained by premovement planning for the control group compared with 60% ± 2.3% for the CB group (P = .0001) (Figure 7 - filled stack). Thus, the accuracy differences between groups may reflect a more accurate 'plan' for the control compared with the CB subjects. Plan I I Compensatory Adjustments 90 i Control Figure 7. Stack bar plot of percent variance in displacement for unpredictable-correct direction trials by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The cerebellar group is to the left, the control group is to the right. Data are averaged across 5 subjects/group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 Individual Subject Planning: EMG Activation Patterns Corroborative evidence for the group differences in premovement planning were found in the agonist-antagonist EMG patterns of individual subjects. Biceps and triceps EMG were inspected for 3 out of 5 control and CB subjects. Signal artifact in the EMG data for 2 subjects in each group prevented analysis. The velocity, biceps, and triceps ensemble average profiles for one subject within each group are shown in Figure 8 and 9. Scales and units for each profile are identical to those seen in Figure 10 of Experiment 2. Planning is most evident in the following instances: 1 ) the longest preparation interval as ample time is available to preplan the response; 2 ) the shortest preparation interval given that the response largely reflects what has been planned in advance and little time is available for response modification; 3) in predictable conditions in which the subject has target information in advance. Additionally, premovement planning would be evident in responses to the subjects default target amplitude. Therefore, EMG activation patterns between a subjects default and the long target amplitude were inspected for evidence of planning. For each target amplitude, comparisons were made between the longest and shortest preparation intervals of the unpredictable condition as well as predictable condition responses. A representative subject from the control group (S2, Figure 8 ) was the most accurate subject demonstrating the lowest AE across S-R intervals in the correct direction and the 3r d most accurate for predictable condition responses (AE = 5.3s ± .19 unpredictable; 2.9° i .13 predictable). Eighty-nine percent of the variance in end position was explained by a combination of planning (81% ± 2.9%) and adjusting (8.4% 1 1.8%) across preparation intervals in the correct direction. This increased to 92% ± 1.2 for planning and 3.4% ± .84 for adjusting in the predictable condition. Across the unpredictable and predictable condition responses represented in Figure 8 and for both the default (30°) and 45s target amplitude, a triphasic EMG burst pattern is evident. Error is low to the default regardless of preparation time. Averaged absolute error in the shortest preparation interval to the long target amplitude indicates that the subject actually moved to the default (45s -13° = 32s ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 Control S2 30' Extension Target S2 45' Extension Target Oh m — O ftM f Biceps 300-400 ms v e lo c ity 0-99 ms Predictable 300-400 ms 0-99 ms Predictable A A » 300-400 ms a " ! 0-99 ms Triceps Figure 8 . Ensemble averages of movement velocity, biceps, and triceps muscles for a representative control subject (S2). Left panel are responses to the 30° extension target (default); right panel are responses to the 45° extension target. The first, fourth, and seventh row of signals for both target amplitudes are averaged velocity, biceps and triceps profiles for the longest S-R interval (300-400 ms) in the unpredictable condition. The second, fifth, and eighth row of signals are averaged velocity, biceps and triceps profiles for the shortest S-R interval (0-99 ms) in the unpredictable condition. The third, sixth and ninth row of signals are averaged velocity, biceps and triceps profiles from predictable condition responses. The number of averaged responses for the 30° target was 29,19, and 12 from first to third row. For the 45° target, response number was 30,17, and 11 respectively. Offset of movement, defined as return to zero velocity is represented as the range of offsets for the three preparation intervals. Note averaged AE for the three preparation intervals for both target amplitudes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 B Cerebellar S9 15' Extension Target S S 4S Extension Target Onset Offset Biceps 300-400 ms velocity 0-99 ms 300-400 ms 0-99 ms y Triceps 300-400 ms 0-99 ms Predictable Figure 9. Ensemble averages of movement velocity, biceps, and triceps for a representative cerebellar subject (S9). Left panel are responses to the 15s extension target (default); right panel are responses to the 45° extension target. The row designation is as described in A, The first two signals for both target amplitudes are responses for the longest and shortest S-R intervals in the unpredictable condition. The 3r d signal are averaged predictable condition responses. Arrows in the left panel, signal-rows 6 and 9 (predictable condition biceps and triceps) identify that the EMG activation pattern appears to be triphasic. The number of averaged responses for the 15° target was 26,19, and 16 from top down. For the 45° target, response number was 25,20, and 16, respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 Responses for the representative cerebellar subject are seen in Figure 9. Subject 9 had the highest AE across preparation intervals and targets in both the unpredictable (11.9° ± .45) and predictable conditions (5.6°± .24). This subjects' percent variance in final position due to premovement planning was the 3r d highest among the cerebellar subjects. A total of 64% of the variance in final position was explained by a combination of premovement planning (58% 1 3%) and compensatory adjustments (6 % ± 1.9%) for the unpredictable condition. This increased to 90% (85% ± 2% planning; 5% ± 1% adjusting) in the predictable condition. Little evidence of a motor plan (i.e., triphasic EMG burst pattern) is seen in the EMG activation pattern of this representative cerebellar subject. Across conditions, targets and preparation intervals, there appears to be a prolonged agonist burst (triceps). With the exception of the predictable condition responses to the default target (15°), a late onset antagonist (burst) can be seen. Only in the predictable condition to the default amplitude does the pattern resemble a triphasic burst pattern (see arrows • Figure 9 - first panel) with an earlier antagonist onset and a distinguishable second agonist burst. Response Updating: Correct Direction The proportion of variance in displacement explained by the addition of target amplitude (i.e., adjustments) increased for the control and cerebellar groups as preparation interval increased (Figure 10). There was a significant main effect of preparation interval on the percentage of variance in final position due to compensatory adjustments (p < .004). Thus, when the target was presented progressively earlier with respect to the 4 * tone, both groups of subjects could utilize target information to adjust their responses. Post hoc analysis revealed that the proportion of variance due to compensatory adjustments in the shortest preparation interval (0 -9 9 ms) was significantly less than the two longest preparation intervals (i.e., 200-299 ms, P < .028 and 300-400 ms P < .004). There were no differences in the percent variance in final position due to adjustments between the cerebellar and control groups (Figure 7 - open stack). For the cerebellar group the percent variance in final position due to adjustments was M (SEM) = 7% (1.0%) compared with M Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 (SEM) = 9% (1.4%) for the control group. As such, the accuracy differences between groups does not appear to result from a greater capability to update the response for the control compared with the cerebellar subjects. c © E 1 Q . M b 8 - 8 § I c ? 0.99 100-199 200-299 S-R Interval 300-400 Figure 10. Bar graph of correct-direction, unpredictable-condition responses representing percent variance in displacement due to compensatory adjustments over S-R interval. Data were averaged across 10 subjects (5 control and 5 cerebellar subjects). The shortest preparation interval (0-99 ms) was significantly different than the two longest preparation intervals (i.e., 200- 299 ms, and 300-400 ms), P =.001 and .009, respectively. Response Planning: Wrong Direction Extent Specification of Wrong Direction Responses All Subjects. Wrong direction responses did not converge on the correct target extent as preparation time increased for either the CB or the control groups. Across preparation intervals, AE for both groups to the virtual short target amplitude was 7.4s ± .38 (mean ± SEM); 9.2° ± .41 for responses to the middle target and 21.3° ± .51 for responses to long targets. For all subjects Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 mean AE remained relatively stable across preparation intervals (P = .70). Inspection of CE for wrong direction responses (as in Figure 7 for correct direction) revealed whether error was an overshoot or undershoot of the target. Given that error to the short amplitude target was approximately a 7° overshoot of the 15° target; a 9° undershoot of the 30° or middle amplitude target and a 21° undershoot of the 45s or long target amplitude, default amplitude selection for wrong direction responses was in between the short and middle amplitude target ( - 2 2 °) for both groups (Figure 11 A). Direction Specification Figure 11B shows the percentage of direction errors over preparation interval for each group. When preparation time was minimal (< 100 ms), the percentage of wrong direction responses was approximately 50% for both groups. When preparation time was greater than 300 ms, the frequency of wrong direction responses for the CB group was -19% of all responses within that S-R interval compared with - 2% for the control group. As such, the time course for direction specification was significantly different for the two groups (x2 = 78.40, p < .00001). The control group showed a 87% reduction in the frequency of direction errors between the 2n d longest (200-299) and longest (300-400) S-R intervals compared with 27% for the CB group. The reduction in frequency of direction errors between the two middle S-R intervals (100-199 and 200-299) was 69% for the control and 47% for the CB group. Only between the 100-199 ms interval and the shortest preparation interval (0-99) was the decrease in direction errors similar for the two groups (7% for the CB group compared with 12% for the control group). Model Analysis: Premovement Planning Percent variance in final position due to premovement planning for wrong direction responses did not parallel what was seen in both predictable and correct direction-unpredictable responses. There was no difference between the two groups in the percent variance due to planning [M (SEM) = 71 % (4.7%) for the control group; M (SEM) = 64% (2.7%) for the cerebellar group, P = .16) (Figure 11C - filled stack). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 Response Updating: Wrong direction Similar to what was seen in correct direction responses, there was no difference between groups in the percent variance due to adjustments. For the CB group, an additional 1.8 % ± .35% variance in final position was explained by compensatory adjustments (P = .11). An additional 2.9% ± .72% variance in final position was explained by compensatory adjustments for the control group (Figure 11C - open stack). 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. 20D-299 c* M Figure 11. A) Bar graph of mean AE (iSEM ) to the virtual target for wrong direction responses within each S-R interval. Data are averaged across all 10 subjects for each target amplitude (short, medium, and long). Wrong direction responses did not converge on the correct target extent as preparation time increased. Across preparation intervals, wrong direction responses were positioned between the short and middle target amplitudes (-22°). B) Frequency of direction errors as a percent of the total responses within each S-R interval for the stroke and control groups in the unpredictable condition. The time course for direction specification was significantly different for the two groups (x2 = 78.40, p < .00001). C) Stack bar plot of percent variance in displacement for unpredictable, wrong direction responses by group. Premovement planning = black, lower bars; Compensatory adjustments = white, upper bars. The cerebellar group is to the left, and the control group is to the right. Data are averaged across 5 subjects for the cerebellar group and 5 subjects for the control group. Additionally, data are averaged across 6 targets and 4 S-R intervals. SEM error bars are shown for each component of the stack bar plot. 156 DISCUSSION Deficits in premovement planning as the source of response inaccuracy for the cerebellar group has been demonstrated by every index of premovement planning employed in this study. First, specification of both movement direction and amplitude, the key parameters of the 'plan,' was impaired for the cerebellar group. Amplitude errors were significantly greater for the cerebellar group compared with controls in the predictable condition. In the unpredictable condition, the cerebellar group produced more direction errors and were less accurate in attaining target amplitude than the control subjects. As specification time increased to 400 ms, the frequency of direction errors attentuated by 95% for the control group compared with only 60% for the cerebellar group. Additionally, the magnitude of reduction in amplitude errors was significantly less for the cerebellar group compared with controls. Absolute error for the control subjects decreased by - 8 ° as preparation time increased in the unpredictable condition, compared with -4 s for the cerebellar group. Second, in both predictable and unpredictable conditions, the percent variance in final position explained by premovement planning was significantly less for the cerebellar subjects compared with controls. Third, a EMG triphasic burst pattern has been identified as an electromyographic index of premovement planning (Hallett, Bhagwan, & Young,1975; Wadman et al., 1979). As such, in all predictable condition responses as well as unpredictable responses in the longest and shortest preparation intervals (particularly to the default amplitude), a pre-planned response would likely be represented as a triphasic EMG burst pattern. While this pattern of activation was consistently seen in the control subject responses, little evidence of a motor plan in the EMG activation patterns was found for cerebellar subjects. Additionally, the 3 cerebellar subjects in which biceps and triceps EMG were inspected demonstrated abnormalities in the EMG activation pattern that have been reported elsewhere and are thought to reflect impaired programming. Specifically these abnormalities are a prolonged agonist and a delayed antagonist burst (Hallett et al., 1991; Hore & Vilis, 1984). Hore and Villis (1984) showed that with practice, monkeys developed a predictive ability to resist perturbations applied to their arm as they attempted to move to a target. The predictive or Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 preprogrammed component consisted of an early antagonist response. The function of this response was considered one of stabilizing the return movement by generating braking in the antagonist muscle. The evidence that this early antagonist response was preprogrammed was that its onset was synchronized to the onset of the initial perturbation of the agonist and that its presence was dependent on the correct expectation of the forthcoming perturbation. The dependance of the early antagonist response on the integrity of the cerebellum was demonstrated with reversible cooling of the dentate and interposed nuclei. The main effect of the cerebellar lesion was a delay in the onset of the antagonist response (Hore et al., 1984) Finally, damage to the cerebellum has long been known to cause prolonged response latencies or reaction times (RT). This deficit is thought to be due to the loss of cerebellar facilitation of motor cortical activity leading up to a response (van Donkelaar & Lee, 1994). In a study by Bonnefoi-Kyriacou and colleagues (1995), patients with cerebellar dysfunction and age- matched controls performed arm pointing movements. In one condition, the task was a simple RT movement directed toward a spatially defined target. The other two conditions involved choice tasks in which the amplitude and direction of movement were varied. The reaction times and movement times were significantly longer for the cerebellar patients than for the controls. Thus, delay in initiation and execution of the movement was found. Again, it was suggested that this might be the result of loss of excitatory feedback from the cerebellum to the motor cortex (Bonnefoi-Kyriacou et al., 1995). In the TRP, reaction time is a manipulated rather than measured variable and as such, prolonged response latencies in our cerebellar subjects can only be inferred. The only way that 3 out of 5 of the cerebellar subjects could meet the acceptable timing criterion established in this study was by an expanded timing window around the 4* tone. Not unlike a prolongation of reaction time, initiation of movement in concert with the auditory cue was delayed in those subjects. Some planning ability was demonstrated by the cerebellar group in that both AE decreased and the percent variance in final position explained by premovement planning increased as time to prepare the response increased. Additionally, in a predictable condition, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 when target amplitude was Known in advance of the auditory cue to move, enhanced planning ability for the cerebellar group was evidenced by greater accuracy and greater explanation of final position due to premovement planning than that attained in the unpredictable condition. However, not only was the planning ability significantly less than that of controls under both conditions, but the planning that was demonstrated afforded only minimal gains in accuracy (Figure 3). The results from this study corroborate the numerous investigations conducted in both non-human primates and human subjects that have identified a role for the cerebro-cerebellum in premovement planning. The influence of cerebellar involvement on the preparatory state of the cerebral cortex for voluntary movements, was determined by measuring the movement-related cortical potentials (MRCP's) in patients with cerebellar dysfunction and healthy control subjects (Kitamura et al., 1999; Tarkka et al., 1993; Verieger et al., 1999; Wessel et al., 1994,1996). Movement-related cortical potentials can be recorded on the scalp over the cerebral cortex and analyzed with electroencephalographic (EEG) averaging. Three of these potentials, the Bereitschaftspotential (BP), the negative slope (NS) and the contingent negative variation (CNV) are recorded as early as 1 second, 400 ms and 200 ms respectively, before the onset of movement and as such, are likely to represent the general preparation for voluntary movement. The amplitude of the MRCP’s that were recorded just prior to sequential and goal-directed finger and arm movements were significantly reduced in cerebellar patients compared with the control subjects. This suggests inadequate cerebellar activation of the motor cortex. Thus, a strong input from the cerebellum close to the movement onset seems to be crucial for the generation of a normal motor potential. Additionally it was found that the distribution of MRCP’s was more diffuse and bilateral in the cerebellar subjects compared with controls. It was suggested that this finding may be related to a more extended cortical activation as a sort of compensation for the greater difficulty in executing the movement (Tarkka et al., 1993; Wessel et al., 1994). The diffuse activation of the primary motor cortex in the patients was thought to be due to impaired cerebellar input to the primary motor cortex. Specifically, neurons from the dentate area of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 cerebellum discharge in relation to movement intent before the onset of movement, and thus provide important input for the motor cortex to initiate the intended movement. This concept was substantiated in a study by Kitamura et al., (1999) in which, MRCPs were again measured in individuals with cerebellar lesions. Unlike any study that has compared the performances of cerebellar subjects with healthy controls using a variety of experimental methods, this is the only one in which the 'subjects’ lesions’ were localized and unilateral. Most often the study subjects represent a wide range of diagnoses and present with diffuse, bilateral cerebellar disease. Movement related cortical potentials were absent or markedly reduced in amplitude for those subjects in the Kitamura et al., study (1999) with dentate nucleus lesions. For those subjects in which there was no evidence of dentate involvement, BP and NS were present. The behavioral manifestation of the decreased MRCP'S referenced above, is possibly the delay in the initiation of movement commonly observed with lesions of the cerebrocerebellum (Jahanshahi, Brown, & Marsden, 1993). Two mechanisms have been proposed to account for the delay in the initiation of movement. First, the dentate nuclei might provide background facilitation to either cortical or subcortical neurons so that, after dentate lesions, commands to initiate movement could bring the motor neurons to fire only after an increased period of summation. Second, the dentate nucleus might participate in or in fact, convey the commands initiating movement (Ghez et al., 1991). A study by Brooks and Thach (1981), provided some insight as to which of these two alternatives might explain delays in movement initiation. The patterns of activity of neurons in the motor cortex of monkeys were recorded while the animals moved the contralateral arm in response to a visual cue. These patterns were then compared before and after reversible inactivation of the dentate nucleus. Inactivation was achieved by cooling the dentate through an inserted probe. Both the discharge of motor cortex neurons associated with the movement and the onset of the movement itself were delayed. If the dentate was merely providing background excitation, the change in activity of neurons in the motor cortex should have occurred at the normal time, but more time should have elapsed before the onset of movement Instead, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 results suggest that the dentate nucleus provides important commands for initiating movement in the motor cortex. Measures of regional cerebral blood flow (rCBF) using Positron Emission Tomography (PET) have demonstrated ‘set’ related activity in the cerebellar hemispheres and midline as healthy subjects either prepared, both prepared and executed, or immediately executed copied hand movements (Krams et al., 1998). Set activity, recorded after the presentation of a 'pre-cue' and before the presentation of a ‘go’ cue has been argued to reflect motor planning and programming. Additionally, decreased rCBF in PM was observed in cerebellar subjects compared with controls performing a self-paced sequential finger opposition task. It has long been considered that activity in premotor neurons is modulated with 'set' or movement preparation (Riehle et al., 1994). As the premotor cortex is one of the primary output channels of the dentate nucleus (Middleton & Strick, 1997), it was suggested that the decreased PM activation was most likely a consequence of decreased cerebellar input via the thalamus (Wessel etal., 1995). In contrast to the numerous reports identifying a role for the cerebellum in premovement planning, Jueptner and colleagues postulate that the neocerebellar hemisphere has little involvement in movement planning and instead is involved in monitoring the outcome of movements by detecting and correcting errors. Accordingly, the cerebellum processes sensory information to minimize errors and optimize movements. These conclusions are drawn largely from studies that demonstrate sensory functions in the absence of a motor task. Using PET, healthy subjects showed increased cerebellar activation when they were required to estimate differences of time intervals (Jueptner et al., 1995, Jueptner et al., 1996). It may be that the differences between those studies supporting a planning function and those supporting an updating function for the cerebellum are related to differences in the kinds of tasks employed. Goodale and Milner (1992) have demonstrated that the neural substrates of visual perception are distinct from those underlying the visual control of actions (Goodale & Milner, 1992). The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 cerebellar processing related to the largely perceptual tasks of Jueptner et al., (1995,1996) may be altogether different than that necessary for visually guided actions. The cerebellar subjects in the present study showed no deficit in response updating. In unpredictable condition responses there were no differences between groups in the percent variance due to compensatory adjustments. In predictable condition responses, in fact, the cerebellar group was significantly greater than controls in the percent variance in final position due to response updating. In addition, the lack of amplitude scaling of wrong direction responses for both groups suggests that the cerebellar subjects were as capable as controls in detecting a direction error and essentially aborting the response. Allen and Tsukahara (1974) suggested that interposed neurons projecting from the intermediate hemisphere use feedforward mechanisms to update and regulate movements. However, the cerebellar subjects in this study with both lateral and intermediate hemisphere damage did not demonstrate deficits in performing compensatory adjustments and updating responses by the measure used to index that process. The intermediate hemisphere is only part of the functional subdivision of the cerebellum called the spinocerebellum. It is the spinocerebellum that has been identified as having a role in the feedforward regulation of movement. The cerebellar subjects of this study may have been capable of performing compensatory adjustments as the result of spared regions of the spinocerebellum. In addition, it has been demonstrated that while interposed neurons are activated later than those in the dentate, their activation is usually before voluntary movement (Home & Butler, 1995). Therefore, the intermediate along with the lateral hemisphere may have a role in movement planning. Finally, neuronal activity correlated with movement execution and response updating is largely recorded from primary motor cortex (M1) (Yoshino et al., 1998). Corticocortical input to primary motor cortex arises from primary sensory cortex as well as the primary motor cortex of the opposite hemisphere. The major subcortical input to the primary motor cortex comes from basal ganglia and cerebellum. Thus, a lesion affecting the cerebellar influence on primary motor cortex would not eliminate response updating functions in M1 . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 Summary The importance of contralateral cerebellar areas in response planning has been demonstrated by decreased accuracy for individuals with cerebellar damage performing with the arm contralateral to the cerebellar hemisphere lesion. This decreased accuracy took the form of both greater extent errors, as well as a greater frequency of direction errors than what was observed in the control group. Deficits in premovement planning as the source of response inaccuracy for the cerebellar group has been demonstrated by every index of premovement planning employed in this study. First, specification of both movement direction and amplitude, the key parameters of the 'plan,' was impaired for the cerebellar group. Second, in both predictable and unpredictable conditions, the percent variance in final position explained by premovement planning was significantly less for the cerebellar subjects compared with controls. Third, little evidence of a motor plan (i.e., triphasic EMG burst pattern) in the EMG activation patterns was found for cerebellar subjects. Abnormalities in the EMG activation pattern that are thought to reflect impaired programming were seen in the responses of the cerebellar subjects. These consisted of a prolonged agonist and delayed antagonist burst. (Hore et al., 1984). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 CHAPTER 5: CONCLUSIONS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 The purpose of this investigation was to provide evidence for a functional neural distinction between premovement planning and feedforward compensatory adjustments. It has long been considered that these two motor control processes underlie the performance of accurate movements. As such, the distinction between premovement planning and ongoing adjustments or modifications has been considered crucial to an understanding of the neural control of goal-directed movement (Gordon & Ghez, 1987b). The mechanisms governing the compensatory adjustments are different than those associated with the early trajectory, and thus, would need to operate in parallel with the preplanned commands implementing the initial response. It has been suggested that rapid responses without manifest discontinuities are controlled by these parallel neural processes that concurrently provide for trajectory planning and error correction as the response unfolds. However, the functional neural substrates associated with these parallel 'paths' have yet to be determined. The effort made here was to determine the neural systems associated with premovement planning and compensatory adjustments by studying individuals who have damage to a particular component of the neural substrates thought to be involved with these processes. However, it is not possible to relate any deficits that are uncovered to the specific location of the lesion and some distinct function of that area versus an alteration in the transmission of afferent information from the damaged area to other important cortical and subcortical areas involved in the control of the action. Likewise the deficits that are revealed may reflect disruption in translating efferent information coming into the damaged region from those same distributed areas. Indeed, numerous efferent and afferent projections exist between several regions of the cerebral cortex and the cerebellar hemisphere. Middleton and Strick (1997) have identified that regions of the motor cortex (M1), premotor cortex (PM), and prefrontal cortex (PFC) are each the target, via the thalamus, of projections from distinct regions of the dentate nucleus. Thus, a lesion affecting the dorsal portion of the dentate nucleus may result in behavioral deficits distinct from a lesion affecting the ventral portion of the dentate. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 Nonetheless, it has been demonstrated that individuals with unilateral sensorimotor area (SM) lesions using their ipsilaterai arm and individuals with unilateral cerebellar hemisphere lesions (CB) using their contralateral arm are less accurate in the performance of a goal-directed aiming action than healthy, age and handedness -matched control subjects. The source of the response inaccuracy for both individuals with SM area and CB strokes, has been identified. Now it is possible to further characterize the nature of the inaccurate responses by comparing the responses of these two groups. Overall, the results from the first and third experiments support the hypotheses with respect to the differential roles of SM areas and cerebellum. The results suggest a significant role for the cerebellum in premovement planning, while the SM area appears to have a role in the execution of ongoing adjustments to the movement trajectory. As such, it was demonstrated that individuals with unilateral cerebellar damage were capable of implementing compensatory adjustments but demonstrated deficits in the ability to generate an adequate plan. In both predictable and unpredictable conditions, CB subjects demonstrated significantly less variance in final position due to premovement planning with seemingly no deficit in response updating compared to healthy controls. Conversely, individuals with unilateral SM area damage showed deficits in the ability to generate compensatory increases in response parameters (i.e., perform feedforward adjustments) while the ability to preplan the response appeared to be spared. Final position was more strongly predicted by the preplanned component for subjects with ipsilaterai sensorimotor area damage than non-lesioned controls in both predictable and unpredictable conditions. However, compared to control subjects, those with stroke were less proficient in updating the default response, as indexed by significantly less percent variance explained by compensatory adjustments in both predictable and unpredictable conditions. One obvious difficulty in the above comparison is that the task of Experiment 1 was sufficiently less complex than that of Experiment 3. When individuals with SM area lesions performed the same task (Experiment 2) as that performed by the individuals with CB lesions (Experiment 3), it was revealed that both planning and updating deficits appeared to account for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 the accuracy differences between the stroke and control groups. However, the nature of the planning deficits for the stroke subjects of Experiment 2 and the CB subjects of Experiment 3 appear to be different. For one, the CB group was significantly less accurate than controls in a predictable condition in which the response can presumably be planned in advance. Second, the CB group did not benefit from increased planning time in the unpredictable condition to the same extent as the stroke subjects of Experiment 2. While less accurate than controls, the stroke group of Experiment 2 similarly improved in response accuracy with increased preparation time. Compared with the control group, the CB group demonstrated only minimal benefit from increased preparation (i.e., planning) time for target amplitude specification. Finally, for the 3 CB subjects from which EMG data could be analyzed, little evidence of a motor plan was found. None of these subjects demonstrated a triphasic muscle activation pattern as would be expected in conditions and time frames in which the response could be fully planned in advance. For all SM area stroke subjects whose EMG data were inspected, triphasic muscle activation patterns were seen in both predictable and long-preparation-interval unpredictable condition responses. The planning deficits demonstrated by the CB subjects largely correspond to a proposal made by Thach (1997), that the cerebellum functions in a 'context-response linkage' manner. The concept is that as a movement is practice, the cerebellum leams to link within itself the context in which the movement is made to the lower level movement generators (Thach, 1997). When the linkage is complete, the occurrence of the context (represented by a certain input to the cerebellum) will trigger (through the cerebellum) the appropriate motor response. The fact that the cerebellar output extends to all planning areas of association cortex as well as to what has been characterized as the 'ultimate' frontal planning area, the prefrontal cortex, suggests that the cerebellum may be involved in context-response linkage and in response combination even at these higher levels. This implies that again through practice, an experiential context would automatically evoke a certain mental action plan. The specific cerebellar contribution would be one of the context linkage and the shaping of the response through trial and error learning. Reminiscent of the planning deficits seen in the stroke subjects of Experiment 2, Thach further Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 asserts that "the prefrontai and premotor areas could still plan without the help of the cerebellum, but not so automatically, rapidly, stereotypically, so precisely linked to context, or so free of error. Nor would their activities improve optimally with mental practice” (Thach, 1997). Recall that the stroke subjects in Experiment 2 (with an intact cerebellar input) did not demonstrate planning deficits in the predictable condition. Additionally, a planning deficit observed for the low- adjustment stroke group was one in which the subjects produced stereotypic, default responses regardless of preparation time or condition. Bastian and Thach (1995) revealed that planning deficits would differ with a lesion directly affecting the cerebellum compared with a lesion disrupting the cerebellar projection to the thalamus. They compared goal-directed reaching and pinching performance in healthy controls, in individuals with lateral cerebellar lesions and in individuals with discrete lesions of the ventrolateral thalamus (VL). A precision pinch task involved retrieving a coin out of a slot using the index finger and thumb. An accurate reach task was one in which the subject was to reach out and precisely touch a 1-cm spot on a ball that was suspended at shoulder height and at a distance of 90% of full arm length away from the subject. In fast reach subjects were to move as fast as they could, touch any part of the ball, and stop at the ball. Lesions of VL resulted in impaired pinching movements, but normal reaching movements. The control and thalamic group did not differ on any measure of reaching movements in either the fast or accurate condition. In contrast, the subjects with lateral cerebellar lesions involving the dentate nucleus demonstrated profound pinching and reaching impairments with significantly greater end-point error in both reaching tasks than either the control or VL group. The study thus provides evidence that the disruption of the cerebellar- thalamo-cortical pathway at the level of the thalamus does not produce similar behavioral deficits to those seen after dentate damage. Questions with respect to planning versus updating functions of the cerebellum was not a part of this study. However, planning deficits can be inferred by the behaviors underlying the inaccurate performance of the cerebellar group. For one, the control and VL groups demonstrated straight path movements to reach the target with minimal corrective movements. The cerebellar group required multiple corrective movements Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 evidenced in the wrist velocity profiles, to compensate for poor specification of both amplitude and direction. Secondly, while the control and VL groups moved the elbow and shoulder together smoothly and continuously throughout the movement, the cerebellar group ‘decomposed’ the movement into first shoulder then elbow movement. One interpretation of the observed elbow- shoulder incoordination is that the cerebellum normally provides the plan for simultaneous execution of elbow and shoulder movement (Bastian & Thach, 1995). On first inspection, it would seem that the subjects with direct cerebellar involvement (Experiment 3) have a much greater deficit in planning than the subjects of Experiment 2 that have disruption of cerebellar-thalamo-cortical projections or planning areas of association cortex (see Table 1 - Experiment 2 • lesion information). The cerebellar subjects were significantly less accurate than controls in predictable condition responses. Horak and Diener (1994) found that cerebellar subjects could not scale the magnitude of their postural responses to expected (predictable) amplitudes like healthy subjects. The cerebellar subjects always used the same response magnitude whether or not the displacement amplitude was predictable or randomized, suggesting that they were not capable of planning the appropriate response magnitude in advance. However, the severity of a deficit is determined by the functional consequence that results. While the CB group was significantly less accurate than controls in the predictable condition, AE was very similar for the CB group and both subgroups of the SM area stroke group (Figure 1A). Additionally, 83% of the variance in final position due to planning was seen in the CB group compared with 80% for the SM area stroke group. While planning deficits were identified for the CB group and both subgroups of the SM area stroke group, across the three experiments, it appears that the most profound deficits are the result of inability to generate compensatory adjustments and update the response. Figure 1B demonstrates AE for correct direction-unpredictable condition responses. The least accurate group was the low-adjustment stroke group of Experiment 2. The high-adjustment stroke sub group was able to compensate for deficient planning, by generating feedforward adjustments and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 updating. This group whose greater error than controls was significant but not meaningful, was more accurate than either the cerebellar group or the low-adjustment subgroup. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 Predictable 12 9 - CB Low Adjust High-Adjust 6 - 3 - U n pi* < S c tab I* C I 9 - 0 - Low Adjust K S High-Adust Figure 1. A) Bar graph of group mean AE (±SEM) for predictable condition responses. Data are averaged across group, S-R interval and 6 target amplitudes. CB group (Experiment 3) = striped bar • averaged across 5 subjects; Low-Adjustment SM area stroke group (Experiment 2) = black bar - averaged across 3 subjects; High-Adjustment SM area stroke group (Experiment 2) = black bar • averaged across 3 subjects. B) Bar graph of group mean AE (±SEM) for unpredictable condition - correct direction responses. Data are averaged across group, S-R interval and 6 target amplitudes. CB group (Experiment 3) = striped bar - averaged across 5 subjects; Low- Adjustment SM area stroke group (Experiment 2) = black bar - averaged across 3 subjects; High- Adjustment SM area stroke group (Experiment 2) = black bar - averaged across 3 subjects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 How does this hypothesis with respect to response updating explain the error for the CB group? The fact that there was no difference between the CB group and the control group in percent variance in final position explained by compensatory adjustments (Experiment 3) would suggest that the CB group did not have a response updating deficit. In fact, the CB group demonstrated significantly greater percent variance due to compensatory adjustments than controls in the predictable condition. Thus, when the CB group is compared with the control group (Experiment 3), no updating deficit for the CB group is evident. However, additional insight with respect to response inaccuracy for the CB group is gained by comparing across Experiments 2 and 3. That is, the compensation for deficient planning seen in the high- adjustment stroke group was not evident in the CB group. For the CB group and the high- adjustment stroke group, the percent variance in final position due to planning was similar (CB = 60%; High-Adjust = 56%) and significantly less than their respective control groups. However, when the contribution of compensatory adjustments was determined total percent variance in final position increased by 18% for the high-adjustment stroke group compared with only 7% for the CB group. Therefore, an inability to compensate for deficient planning by generating feedforward adjustments may have limited the performance of the CB subjects. All five CB subjects had lesions affecting both the lateral and intermediate cerebellar hemispheres. As such, deficits related to feedforward processing, a function attributed to the spinocerebellum could have been the result of interposed nuclei involvement. Nonetheless, the cerebellar group demonstrated profound planning deficits when compared with an age-matched control group. While is it interesting to speculate about potential deficits in the cerebellar group when compared to the high-adjustment SM group, comparisons are limited by two specific differences between the CB and SM groups. First, average age of the CB group was 49.8 years compared with 61.3 years for the SM area group. Age-related changes in motor control have been well established (Pohl, Winstein, & Fisher, 1996) and as such comparisons between groups that differ in age must be considered cautiously. Second, all of the CB subjects performed the task with their dominant arm. The three high-adjustment SM subjects Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 performed the task with their non-dominant arm. Hand dominance affects on motor control are also well established (Winstein & Pohl, 1995). The purpose of the three experiments presented was to determine the functional neural substrates for response planning and response updating. It was hypothesized that the cerebellar hemispheres had a preferential role in premovement planning while the SM areas (M1 in particular) had a role in generating compensatory adjustments. In Figure 2 the neural substrates related to planning and updating are depicted. This scheme is an adaptation of a model originally developed by Allen and Tsukahara (1974). It has been updated with current information regarding cerebellar output pathways as well as current anatomical terminology (Middleton & Strick, 1997; Schmahmann & Pandya, 1997). While the lesions that have been placed along various pathways in Figure 2 are identified by the groups presently studied, they are not intended to specifically 'tag' the area of brain damage in these subjects. The lesions are representative of areas in the central nervous system that when damaged might produce similar ‘behaviors' to what was observed for each of the three groups presented. Certainly, the location of the brain lesions in Figure 2 reflect the lesions of the subjects in the three experiments. Each subject in Experiment 2 had internal capsule involvement (Figure 2: high-adjustment stroke group lesions). Each CB subject had lesions involving Doth dentate and interposed nuclei (Figure 2: cerebellar group lesions). The planning pathway seen in Figure 2 (solid-thick line arrows) involves the association cortex, including premotor, supplementary motor areas, as well as prefrontal cortex and posterior parietal. Each of these areas is known to process both efferent and afferent information from the neocerebellum via numerous thalamic nuclei. The adjustment pathway (broken-line arrows) includes the spinocerebellum and the output of interposed nuclei to M1 via the thalamus, as well as M1. The behavior of the high-adjustment stroke group suggests that their lesions involved the planning pathway but spared the adjustment pathway. Deficits in both planning and updating observed by the low-adjustment sub-group suggests damage along both pathways depicted in Figure 2 as both the solid and broken lines from cerebellar thalamus to motor cortex. Deficits in both planning and updating observed in the CB group need not be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 related to any brain damage outside of the cerebellum itself when the lesion involves both dentate and interposed output nuclei. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 Figure 2. Scheme (adapted from Allen & Tsukahara, 1974) showing proposed roles of several brain structures in premovement planning and response updating. It is proposed that basal ganglia (not addressed in this study) and the neocerebellum are involved with association cortex in the planning, programming of volitional movement. Solid- thick black lines and arrows represent efferent and afferent connections of the planning pathway: 1) Association cortex (Assn Cx) known to have a role in planning includes: Supplementary motor area (SMA); Premotor (PM); Prefrontal cortex (PF) and Posterior parietal; 2) Cerebrocerebellum includes the lateral cerebellar hemisphere with the dentate as the primary output nucleus to the cerebellar thalamus. The cerebellar thalamus refers to the histologically identified nuclei which receive cerebellar afferents and send efferent projections to motor cortex and back to association cortex. Listed for the purposes of this study are the ‘motor’ thalamic nuclei of the cerebellar thalamus: VPLo = the pars oralis of the ventral posterolateral nucleus; VLc = ventral lateral, pars caudalis; VIps = ventral lateral, pars postrema; X = nucleus X. For the purposes of this study the adjustment pathway (depicted by broken lines and arrows) includes: 1) Motor cortex - specifically primary motor cortex (M1); 2) the spinocerebellum which includes the intermediate cerebellar hemisphere with the interposed nuclei as the primary output nucleus to the cerebellar thalamus with efferent projections to motor cortex. The 3 different symbols placed along various lines represent sites that with damage, could potentially produce behaviors identified for each of the 3 groups: Low-adjustment SM area stroke group; High-adjustment SM area stroke group and Cerebellar group. The symbols do not represent specific lesion locations of the subjects in the 3 experiments presented. 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. PLAN, PROGRAM EXECUTE IDEA y /- } SM A PM PF Posterior Parietal „ w 1 VPLo Cerebellar I VLc Thalamus f VLps X Assn CX f t 1 i A Cerebro- cerebellum Dentate r Basal Ganglia Spmoct nMlum Cerebellar Thalamus IT ^ Interposed Somatostnsory Planning Pathway - — — — -► Adjustment Pathway X Low-Adjuetment Stroke Group HJgh-Adjuttment Stroke Group Cerebellar Group 0 1 176 Bibliography Adam, J. (1992). The effects of objectives and constraints on motor control strategy in reciprocal aiming movements. Journal of Motor Behavior. 2 4 .173-185. Allen, G. I. & Tsukahara, N. (1974). Cerebro-cerebellar communication systems. Physiological Reviews. 54, 957-1006. Alexander, G.E. & Crutcher, M.D. (1990a) Preparation for movement: neural representations of intended direction in three motor areas of the monkey. Journal of Neurophvsioloav. 64, 133-150. Alexander, G.E. & Crutcher, M.D. (1990b) Neural representations of the target (goal) or visually guided arm movements in three motor areas of the monkey. Journal of Neurophvsioloov. 64,164-178. Amarenco, P. (1991). The spectrum of cerebellar infarctions. Neurology. 41.973-979. Ashe, J., Taira, M., Smymis, N., Pellizzer, G., Georgakopoulos, T., Lurito, J.T. & Georgopoulos, A.P. (1993). Motor cortical activity preceding a memorized movement trajectory with an orthogonal bend. Experimental Brain Research. 95.118-130. Bahill, A.T., Clark, M.R. & Stark, L. (1975). The main sequence, a tool for studying eye movements. Math Bioscience. 24.191-204. Bastian, A.J. & Thach, W.T. (1995). Cerebellar outflow lesions: a comparison of movement deficits resulting from lesions at the levels of the cerebellum and thalamus. Annals of Neurology. 38. 881-892. Blakemore, S.J., Goodbody, S.J. & Wolpert, D.M. (1998). Predicting the consequences of our own actions: the role of sensorimotor context estimation. Journal of Neuroscience. 18. 7511-7518. Bock.O. (1994). Scaling of joint torque during planar arm movements. Experimental Brain Research. 101. 346-352. Bonnefoi-Kyriacou, B., Trouche, E., Legalist, E. & Viallet, F. (1995). Planning and execution of pointing movements in cerebellar patients. Movement Disorders. 1 0 .171-178. Brooks, V.B. (1974). Introductory lecture to session III some examples of programmed limb movements. Brain Research. 7 1 .299-308. Brooks, V.B. (1979). Motor programs revisited. In R.E.Talbott. & D.R.Humphrey (Eds.), Posture and movement, (pp. 13-48). New York: Raven Press. Brooks, V.B. & Thach, W.T. (1981). Cerebellar control of posture and movement. In V.B.Brooks (E d.),. Handbook of Dhvsioloov. Section 1: The Nervous System. Vol. II. Motor Control. Part 2. (pp. 877-946). Bethesda: American Physiological Society. Brown, J.S. & Slater-Hammel, A.T. (1949). Discrete movements in the horizontal plane as a function of their length and direction. Journal of Experimental Psychology. 3 9 .84-95. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 Bullock, D. & Grossberg, S. (1988). Neural dynamics of planned arm movements: emergent invariants and speed-accuracy properties during trajectory formation. Psychological Review. 95,49-90. Caminiti, R„ Johnson, P.B., Galli, C., Ferrauba, S., Bumod, Y. & Urbano, A. (1991). Making arm movements within different parts of space: the premotor and motor cortical representation of a coordinate system for reaching at visual targets. Journal of Neurophvsioloov. 11,1182-1197. Chen, R., Gerloff, C., Hallett, M. & Cohen, L.G. (1997). Involvement of the ipsilaterai motor cortex in finger movements of different complexities. Ann Neurol. 4 1 .247-254. Cooke, J.D. & Diggles, V.A. (1984). Rapid error correction during human arm movements: evidence for central monitoring. Journal of Motor Behavior. 16. 348-363. Corcos, D.M., Gottlieb, G.L. & Agarwal, G.C. (1989). Organizing principles for single joint movements II. A speed-sensitive strategy. Journal of Neurophvsioloav. 6 2 .358-368. Crammond, D.J. & Kalaska, J.F. (1989). Neuronal activity in primate parietal cortex area 5 varies with intended movement direction during instructed-deiay period. Experimental Brain Research. 7 6 .458-462. Crammond, D.J., Kalaska, J.F. (1996). Differential relation of discharge in primary motor cortex and premotor cortex to movements versus actively maintained postures during a reaching task. Experimental Brain Research. 108.45-61. Culham, J.C., Brandt, S.A., Cavanagh, P., Kanwisher, N.G., Dale, A.M. & Tootell, R.B. (1998). Cortical fMRI activation produced by attentive tracking of moving targets. Journal of Neurophvsioloav. 80. 2657-2670. Damasio, H. (1995) Human Brain Anatomy in Computerized Imaaes. New York: Oxford University Press Damasio, H. & Damasio, A.R. (1989). Lesion analyses in neuropsychology. Oxford University Press, New York. 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. Deiber, M.P., Ibanez, V., Sadato, N. & Hallett, M. (1996). Cerebral structures participating in motor preparation in humans: a positron emission tomography study. Journal of Neurophvsioloav. 75. 233-247. Desmedt, J.E. & Godaux, E. (1978). Ballistic skilled movements: load compensation and patterning of the motor commands. In J.E.Desmedt (Ed.), Progress in clinical neurophvsioloov. cerebral motor control in man: Iona loop mechanisms, (pp. 21-55). Basel: Karger. Di Lazarro, V., Restuccia, D., Molinari, M., Leggio, M.G., Nardonne, R., Fogli, D. & Tonali.P. (1994). Excitability of the motor cortex to magnetic stimulation in patients with cerebellar lesions. Journal of Neurology. Neurosuraerv and Psychiatry. 5 7 .108-110. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 Dugas, C. & Smith, A.M. (1992). Responses of cerebellar purklnje cells to slip of a hand-held object. Journal of Neuroohvsioloov. 67. 483-495. Espinoza, E. & Smith, A.M. (1990). Purkinje cell simple spike activity during grasping and lifting objects of different textures and weights. Journal of Neuroohvsioloov. 64. 698-714. Evarts, E.V. (1971). Central control of movement. V. Feedback and corollary discharge: a merging of the concepts. Neurosci Res Proa Bull. _9, 86-112. Faviila, M., Hening, W. & Ghez, C. (1989). Trajectory control in targeted force impulses VI. Independent specification of response amplitude and direction. Experimental Brain Research. 75. 280-294. Faviila, M., Gordon, J., Hening, W. & Ghez, C. (1990). Trajectory control in targeted force impulses VII. Independent setting of amplitude and direction in response preparation. Experimental Brain Research. 79. 530-538. Faviila, M. & De Cecco.E. (1996). Parallel direction and extent specification of planar reaching arm movements in humans. Neuropsvcholooia. 34.609-613. Finger, S. (1994). The brain in antiquity. In: Finger S (Ed), Origins of Neuroscience. Oxford University Press, New York, pp 3. Fisher, B.E., Winstein, C.J. & Velicki, M.R. (2000). Deficits in compensatory trajectory adjustments after unilateral sensorimotor stroke. Experimental Brain Research. http://dx.doi.ora/10.1007/s002219900316 Fortier, P.A., Kalaska, J.F. & Smith, A.M. (1989). Cerebellar neuronal activity related to whole- arm reaching movements in the monkey. Journal of Neurophvsioloov. 6 2 .198-211. Freund, H.-J. & Budingen.H.J. (1978). The relationship between speed and amplitude of the fastest voluntary contractions of human arm muscles. Experimental Brain Research. 31. 1- 12. Fu, Q.-G., Flament, D., Coitz, J.D. & Ebner, T.J. (1995). Temporal encoding of movement kinematics in the discharge of primate primary motor and premotor neurons. Journal of Neurophvsioloov. 7 3 .836-853. Fu, Q.-G., Suarez, J.l. & Ebner, T.J. (1993). Neuronal specification of direction and distance during reaching movements in the superior precentral premotor area and primary motor cortex of monkeys. Journal of Neurophvsioloov. 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. Georgopoulos, A.P. (1994a). Population activity in the control of movement. International Review of Neurobioloov. 3 7 .103-123. Georgopoulos, A.P. (1994b). Behavioral neurophysiology of the motor cortex. Journal of Laboratory and Clinical Medicine. 124. 766-774. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 Georgopoulos, A.P. (1994c). New concepts in generation of movement. Neuron. 13. 257*268. Georgopoulos, A.P. (1995). Current issues in directional motor control. Trends in Neuroscience. 18, 506-510. Georgopoulos, A.P., Kalaska, J.F., Caminiti, R. & Massey, J.T. (1982). On the relations between the direction of two-dimensional arm movements and cell discharge in primate motor cortex. Journal of Neurophvsioioov. 2,1527-1537. Georgopoulos, A.P., Caminiti, R., Kalaska, J.F. & Massey, J.T. (1983a). Spatial coding of movement: a hypothesis concerning the coding of movement direction by motor cortical populations. Experimental Brain Research Supplementum. 7, 327-336. Georgopoulos, A.P., Kettner, R.E. & Schwartz, A.B. (1988). Primate motor cortex and free arm movements to visual targets in three-dimensional space. II. Coding of the direction of movement by a neuronal population. Journal of Neuroscience. 8 , 2928-2937. Georgopoulos, A.P., Crutcher, M.D. & Schwartz, A.B. (1989). Cognitive spatial-motor processes 3. Motor cortical prediction of movement direction during an instructed delay period. Experimental Brain Research. 7 5 .183-194. Georgopoulos, A.P., Ashe, J., Smymis, N. & Taira, M. (1992). The motor cortex and the coding of force. Science. 2 5 6 .1692-1695. Gerioff, C., Cohen, L..G., Floeter, M.K., Chen, R., Corwell, B. & Hallett, M. (1998). Inhibitory influence of the ipsilaterai motor cortex on responses to stimulation of the human cortex and pyramidal tract Journal of Physiology. 510.249-259. Ghez, C. (1979). Contributions of central programs to rapid limb movement in the cat. In H.Asunama & V.J.Wilson (Eds.), Integration in the nervous system, (pp. 305-319). Tokyo: Igaku-Shoin Ltd. Ghez, C. & Vicario, D. (1978) The control of rapid limb movement in the cat II. Scaling of isometric force adjustments. Experimental Brain Research. 3 3 .191-202. Ghez.C., Hening, W. & Faviila,M. (1989). Gradual specification of response amplitude in human tracking performance. Brain. Behavior and Evolution. 33.69-74. Ghez, C., Hening, W. & Gordon, J. (1991). Organization of voluntary movement. Current Opinion in Neurobioloav. 1, 664-671. Ghez, C., Faviila, M., Ghilardi, M.F., Gordon, J., Bermejo, R. & Pullman, S. (1997). Discrete and continuous planning of hand movements and isometric force trajectories. Experimental Brain Research. 115.217-233. Ghez, C. & Thach, W.T. (2000). The Cerebellum. In: E.R.Kandel, J.H.Schwartz, & T.M.Jessel (Eds.), Principles of Neural Science, (pp. 833-852). New York: Elsevier Science Publishing Co. Goodale, M.A. & Milner, A.D. (1992). Separate visual pathways for perception and action. Trends in Neuroscience. 1 5.20-25. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 Gordon, A.M., Westling, G., Cole, K.J. & Johansson, R.S. (1993). Memory representations underlying motor commands used during manipulation of common and novel objects. Journal of Neurophvsioloov. 6 9 .1789-1796. Gordon, A.M., Lee, J.H., Flament, D., Ugurbil, K. & Ebner, T.J. (1998). Functional magnetic resonance imaging of motor, sensory, and posterior parietal cortical areas during performance of sequential typing movements. Experimental Brain Research. 121.153- 166. Gordon, J. & Ghez, C. (1984). EMG patterns in antagonist muscles during isometric contraction in man: relations to response dynamics. Experimental Brain Research. 5 5 .167-171. Gordon, J. & Ghez, C. (1987a). Trajectory control in targeted force impulses. II. pulse height control. Experimental Brain Research. 6 7 .241-252. Gordon, J. & Ghez, C. (1987b) Trajectory control in targeted force impulses. III. compensatory adjustments for initial errors. Experimental Brain Research. 6 7 .253-269. Gottlieb, G.L., Corcos, O.M., & Agarwal, G.C. (1989a). Organizing principles for single-joint movements I. A speed-insensitive strategy. Journal of Neurophvsioloov. 62. 342-357. Gottlieb, G.L., Corcos, D.M. & Agarwal, G.C. (1989b). Organizing principles for single-joint movements II. A speed-sensitive strategy. Journal of Neurophvsioloov. 62. 358-368. Gottlieb, G.L., Corcos, D.M., Agarwal, G.C. & Latash, M.L. (1990). Organizing principles for single joint movements III. Speed-insensitive strategy as a default. Journal of Neurophvsioloav. 63, 625-636. Gottlieb, G.L., Chen, C-H & Corcos D.M. (1996). Nonlinear control of movement distance at the human elbow. Experimental Brain Research. 112.289-297. Grill, S.E., Hallett, M., Marcus, C. & McShane.L. (1994). Disturbances of kinaesthesia in patients with cerebellar disorders. Brain. 117.1433-1447. Grill, S.E., Hallett, M. & McShane, L.M. (1997). Timing of onset of afferent responses and of use of kinesthetic information for control of movement in normal and cerebellar-impaired subjects. Experimental Brain Research. 113. 33-47. Haaland, K.Y., Harrington, D.L. & Yeo, R. (1987). The effects of task complexity on motor performance in left and right CVA patients. Neuropsvcholooia. 2 5 .783-794. Hager-Ross, C. (1995). Sensorimotor mechanisms in reactive control of grasp stability. To grip and not to slip', (pp 1-34). Sweden: Umea University Medical Dissertations. Hallett, M., Bhagwan, S.T. & Young, R.R. (1975). EMG analysis of stereotyped voluntary movements in man. Journal of Neurology. Neurosuroerv. and Psychiatry. 3 8 .1154-1162 Hallett, M., Berardelli, A., Matheson, J., Rothwell, J. & Marsden, C.D. (1991). Physiological analysis of simple rapid movements in patients with cerebellar deficits. Journal of Neurology. Neurosuroerv and Psychiatry. 5 3 .1163-1169. Hallett, M., Shahani, B.T., & Young, R.R. (1975). EMG analysis of stereotyped voluntary movements in man. Journal of Neurology. Neurosuroerv and Psychiatry. 38.1154- 1162. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 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. Hening, W., Vicario, D. & Ghez, C. (1988). Trajectory control in targeted force impulses. IV. influences of choice, prior experience and urgency. Experimental Brain Research. 71, 103-115. Hening, W., Favilla, M. & Ghez, C. (1988). Trajectory control in targeted force impulses. V. gradual specification of response amplitude. Experimental Brain Research. 7 1 .116-128. Henry, F.M., (1953). Dynamic kinesthetic perception and adjustment. Research Quarterly. 24. 176-187. Henry, F.M., & Rogers, D.E. (1960). Increased response latency for complicated movements and a "memory drum" theory of neuromotor reaction. Research Quarterly. 3 1 .448-458. Hick, W.E. (1952). On the rate of gain of information. The Quarterly Journal of Experimental Psychology. 4,11-26. Higgins, J.R. & Angel, R.W. (1970). Correction of tracking errors without sensory feedback. Journal of Experimental Psychology. 8 4 .412-416. Horak, F.B., Diener, H.C. & Nashner, L.M. (1989). Influence of central set on human postural responses. Journal of Neurophvsioloav. 62. 841-853. Horak, F.B. & , Diener, H.C. (1994). Cerebellar control of postural scaling and central set in stance. Journal of Neurophvsioloav. 7 2 .479-493. Hore, J. & Vilis, T. (1984). A cerebellar-dependent efference copy mechanism for generating appropriate muscle responses to limb perturbations. In J. R. Bloedel (Ed.), Cerebellar Functions, (pp. 24-35). Berlin: Springer-Veriag. Home, M.K. & Butler, E.G. (1995). The role of the cerebello-thalamo-cortical pathway in skilled movement. Progress in Neurobioloav. 4 6 .199-213. Jahanshahi, M., Brown, R.G. & Marsden, C.D. (1993). A comparative study of simple and choice reaction time in Parkinson's, Huntington's and cerebellar disease. Journal of Neurology. Neurosurgery, and Psychiatry. 56.1169-1177. Jahanshahi, M., Jenkins, I.H., Brown, R.G., Marsden, C.D., Passingham, R.E. & Brooks, D.J. (1995). Self-initiated versus externally triggered movements. I. An investigation using measurement of regional cerebral blood flow with PET and movement-related potentials in normal and Parkinson’s disease subjects. Brain. 118.913-933. Johansson, R.S. & Cole, K.J. (1994). Grasp stability during manipulative actions. Can J Phvsiol Pharmacol/Canadian Journal of Physiology and Pharmacology. 7 2 .511-524. Jueptner, M., Rijntjes, M., Weiller, C., Faiss, J.H., Timmann, D., Mueller, S.P., & Diener, H.C. (1995). Localization of a cerebellar timing process using PET. Neurology. 45.1540- 1545. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 Jueptner, M., Flerich, L., Weiller, C., Mueller, S.P. & Diener, H.C. (1996). The human cerebellum and temporal information processing-results from a PET experiment. Neuroreport. 7, 2761-2765. Jueptner, M., Jenkins, I.H., Brooks, D.J., Frackowiak, R.S. & Passingham, R.E. (1996). The sensory guidance of movement: a comparison of the cerebellum and basal ganglia. Experimental Brain Research. 112. 462-474. Jueptner, M., Ottinger, S., Fellows, S.J., Adamschewski, J., Flerich, L., Muller, S.P., Diener, H.C., Thilmann, A.F. & Weiller, C. (1997). The relevance of sensory input for the cerebellar control of movements. Neuroimaae. J , 41-48. Jueptner.M. & Weiller,C. (1998). A review of differences between basal ganglia and cerebellar control of movements as revealed by functional imaging studies. Brain. 121.1437-1449. Kalaska, J.F., Cohen, D.A.D., Hyde, M.L. & Prud'homme, M.A.. (1989). Comparison of movement direction-related versus load direction-related activity in primate motor cortex using a two-dimensional task. Journal of Neuroscience. 9,2080-2102. Kalaska, J.F. & Crammond, D.J. (1992). Cerebral cortical mechanisms of reaching movements. Science. 255.1517-1523. Kawashima, R., Roland, P.E. & O’Sullivan, B.T. (1994). Fields in human motor areas involved in preparation for reaching, actual reaching, and visuomotor learning: a positron emission tomography study. The Journal of Neuroscience. 14. 3462-3474. Kawashima, R., Matsumura, M., Sadato, N., Naito, E., Waki, A., Nakamura, S., Matsunami, K., Fukuda, H. & Vonekura, Y. (1998). Regional cerebral blood flow changes in human brain related to ipsilateral and contralateral complex hand movements - a PET study. European Journal of Neuroscience. 10.2254-2260. Kitamura, J., Shibasaki, H., Takagi, A., Nabeshima, H. & Yamaguchi, A. (1993). Enhanced negative slope of cortical potentials before sequential as compared with simultaneous extensions of two fingers. Electroencephalography and Clinical Neurophvsioloav. 8 6 . 176-182. Kitamura, J., Shibasaki, H., Terashi, A., & Tashima, K. (1999). Cortical potentials preceding voluntary finger movement in patients with focal cerebellar lesion. Clinical Neurophvsioloav. 1 10 .126-32. Klapp.S.T. (1977). Reaction time analysis of programmed control. Exercise Sport Sci Rev. 5, 231-253. Krams, M., Rushworth, M.F., Deiber, M.P., Frackowiak, R.S. & Passingham, R.E. (1998). The preparation, execution and suppression of copied movements in the human brain. Experimental Brain Research. 120.386-398. Kunesch, E., Binkofski, F., Steinmetz, H. & Freund, H.J. (1995). The pattern of motor deficits in relation to the site of stroke lesions. Eur Neurol. 35. 20-26. Kurata, K. (1993). Premotor cortex of monkeys: set- and movement-related activity reflecting amplitude and direction of wrist movements. Journal of Neurophvsioloav. 6 9 .187-199. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 Latash, M.L. & Anson, J.G. (1996). What are ’normal movements” in atypical populations? Behavioral Brain Sciences. 19. 55-106. Macefield, V.G. & Johansson, R.S. (1994). Electrical signs of cortical involvement in the automatic control of grip force. Neuroreport. 5,2229-2232. Martin, J.H. & Ghez.C. (1988). Red nucleus and motor cortex: parallel motor systems for the initiation and control of skilled movement. Behavioural Brain Research. 28.217-223. Martin, J.H., Cooper, S.E. & Ghez.C. (1995). Kinematic analysis of reaching in the cat. Experimental Brain Research. 102. 379-392. 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. 3 9 .386-391 Meyer, D.E., Abrams, R.A., Komblum, S., Wright, C.E. & Smith, J.E.K. (1988). Optimality in human motor performance: ideal control of rapid aimed movements. Psychological Review. 95. 340-370. Middleton, F.A. & Strick, P.L. (1997). Cerebellar output channels. International Review of .Neurobioloav. 41. 61-82. Nakai, R. (1997). The timed response paradigm for interface with an arm manipulandum [Computer program]. Los Angeles, CA: USC Movement Performance and Learning Laboratory Department of Biokinesiology Ojakangas, C.L. & Ebner, T.J. (1991). Scaling of the metrics of visually-guided arm movements during motor learning in primates. Experimental Brain Research. 85,314-323. Pfann, K.D., 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. Picard,N. & Smith, A.M. (1992). Primary motor cortical responses to perturbations of prehension in the monkey. Journal of Neurophvsioloav. 6 8 .1882-1894. Pohl, P.S., Winstein, C.J. & Fisher, B.E. (1996). The locus of age-related movement slowing: sensory processing in continuous goal-directed aiming. Journal of Gerontology: Psychological Sciences. 51B. 94-102. Poulton, E.C. (1981). Human manual control. In: V.B. Brooks (Ed.), Handbook of ohvsioloov. Sec 1: The nervous system II: motor control. (Vol. 2, pp. 1337-1389). Bethesda: American Physiological Society. Rao, S.M., Binder, J.R., Bandettini, P.A., Hammeke, T.A., Yetkin, F.Z., Jesmanowicz, A., Lisk, L.M., Moms, G.L., Mueller, W.M., Estkowski, L.D., Wong, E.C., Haughton, V.M. & Hyde, J.S. (1993). Functional magnetic resonance imaging of complex human movements. Neurology. 4 3 .2311-2318. Richter, W., Anderson, P.M., Georgopoulos, A.P., Kim, S.G. (1997). Sequential activity in human motor areas during a delayed cued finger movement task studied by time-resolved fMRI. Neuroreoort. 8,1257-1261. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 Riehle, A. & Requin, J. (1989). Monkey primary motor and premotor cortex: single-cell activity related to prior information about direction and extent of an intended movement Journal of Neurophvsioloov. §1, 534-549. Riehle, A. & Requin, J. (1993). The predictive value for performance speed of preparatory changes in neuronal activity of the monkey motor and premotor cortex. Behavioural Brain Research. 53,35-49. Riehle, A., Mackay, W.A. & Requin, J. (1994). Are extent and force independent movement parameters? Preparation- and movement-related neuronal activity in the motor cortex. Experimental Brain Research. 99. 56-74. Riehle, A. & Requin, J. (1995). Neuronal correlates of the specification of movement direction and force in four cortical areas of the monkey. Behavioural Brain Research. 7 0 .1-13. Rosenbaum, D.A. (1980). Human movement initiation: specification of arm, direction, and extent. Journal of Experimental Psychology: General. 109.444-474. Rosenbaum, D.A. (1983). The movement precuing technique: assumptions, application, and extensions. In RAMagill (Ed.), Memory and control of action, (pp. 231-274). Amsterdam: North-Holland Publishing Company. Sakai, K., Takino, R., Hikosaka, O., Miyauchi, S., Sasaki, Y., Putz, B. & Fujimaki, N. (1998). Separate cerebellar areas for motor control. Neuroreoort. 9, 2359-2363. Sakai, ST., Inase, M. & Tanji, J. (1996). Comparison of cerebellothalamic and pallidothalamic projections in the monkey (Macaca fuscata): a double anterograde labelling study. Journal of Comparative Neurology. 368.215-228. Sasaki, K., Gemba, H. & Hashimoto, S. (1981). Influences of cerebellar hemispherectomy upon cortical potentials preceding visually initiated hand movements in the monkey. Brain Research. 205.425-430. Schluter, N.D., Rushworth, M.F.S., Mills, K.R. & Passingham, R.E. (1999). Signal-, set-, and movement- related activity in the human premotor cortex. Neuropsvcholooia. 3 7 .233- 243. Schmahmann, J.D. & Pandya, D.N. (1997). The cerebrocerebellar system. International Review of Neurobioloav. 41. 31-60. 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. Smith, A.M., Dugas, C., Fortier, P., Kalaska, J. & Picard, N. (1993). Comparing cerebellar and motor cortical activity in reaching and grasping. Canadian Journal of Neurological Science. 20. 53-61. Stephan, K.M., Fink, G.R., Passingham, R.E., Silbersweig, D., Ceballos-Baumann, A.O., Frith, C.D. & Frackowiak, R.S. (1995). Functional anatomy of the mental representation of upper extremity movements in healthy subjects. Journal of Neurophvsioloov. 7 3 .373- 386. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 Tanji, J. & Evarts, E.V. (1976). Anticipatory activity of motor cortex neurons in relation to direction of an intended movement Journal of Neurophvsioloav. 3 9 .1062-1068. Tanji, J., Okano, K. & Sato, K.C. (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. & Mushiake, H. (1996). Comparison of neuronal activity in the supplementary motor area and primary motor cortex. Brain Research. 3,143-150. Tarkka, I. M„ Massaquoi, S. & Hallett, M. (1993). Movement-related cortical potentials in patients with cerebellar degeneration. Acta Neurolooica Scandinavica. 8 8 .129-135. Thach.W.T. (1997). Context-response linkage. International Review of Neurobioloov. 41. 599- 611. van Donkelaar, P. & Lee, R.G. (1994). Interactions between the eye and hand motor systems: disruptions due to cerebellar dysfunction. Journal of Neurophvsioloov. 72(4), 1674- 1685. Vanman, E (1992) The timed response paradigm for interface with an arm manipulandum [Computer program]. Los Angeles, CA: USC Movement Performance and Learning Laboratory Department of Biokinesiology Velicki, MR., Winstein, CJ. & Pohl, PS. (2000). impaired direction and extent specification of aimed arm movements in humans with stroke-related brain damage. Experimental Brain Research. 130.362-374. Verleger, R., Wascher, E., Wauschkuhn, B., Jaskowski, P., Allouni, B., Trillenberg, P. & Wessel, K. (1999). Consequences of altered cerebellar input for the cortical regulation of motor coordination, as reflected in EEG potentials. Experimental Brain Research. 127.409- 422. Vicario.D.S. & Ghez.C. (1984). The control of rapid limb movement in the cat. IV. Updating of ongoing isometric responses. Experimental Brain Research. 55.134-144. Wadman, W.J., Denier van der Gon, J.J., Geuze, R.H., & Mol, C.R., (1979). Control of fast goal- directed arm movements. Journal of Human Movement Studies. 5, 3-17. Ware, J.E. & Sherboume, CD. (1992). The MOS 36-item short-form health survey (SF-36): Conceptual framework and item selection. Medical Care. 3 0 .473-483. Weinrich, M. & Wise, S.P. (1982). The premotor cortex of the monkey. Journal of Neuroscience. 2,1329-1345 Wessel, K... Verleger, R., Nazarenus, D., Vieregge, P. & Kompf., D. (1994). Movement-related cortical potentials preceding sequential and goal-directed finger and arm movements in patients with cerebellar atrophy. Electroencephalography and Clinical Neurophvsioloov. 92,331-341. Wessel, K., Zeffiro, T„ Lou, J-S., Toro, C. & Hallett, M. (1995). Regional cerebral blood flow during a self-paced sequential finger opposition task in patients with cerebellar degeneration. Brain. 118. 379-393. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 Wessel, K., Tegenthoff, M„ Vorgerd, M„ Otto, V., Nitschke, M.F. & Malin, J.P. (1996). Enhancement of inhibitory mechanisms in the motor cortex of patients with cerebellar degeneration: a study with transcranial magnetic brain stimulation. Electroencephalography and Clinical Neurophvsioloov. 101. 273-280. Winstein, C.J. & Pohl, P.S. (1995). Effects of unilateral brain damage on the control of goal- directed hand movements. Experimental Brain Research. 105.163-174. Winstein, C.J., Grafton, S.T. & Pohl, P.S. (1997) Motor task difficulty and brain activity: an investigation of goal-directed reciprocal aiming using positron emission tomography (PET). Journal of.Neurophysiology 77:1581-1594 Winstein, C.J, Merian, A.S. & Sullivan, K.J. (1999). Motor learning after unilateral brain damage. Neuroosvcholoqia. 37:975-987. Winstein, C.J., Horak, F.B. & Fisher, B.E. (2000). Influence of central set on anticipatory and triggered grip-force adjustments. Experimental Brain Research. 130. 298-308. Winstein, C.J., Onla-or, S., Rose, D.K., Sullivan, K.J. & Pohl, P.S. (submitted). Aiming movement deficits in the ipsilateral upper extremity following stroke. Woodworth,R.S. (1899). The accuracy of voluntary movement. Psychological Monographs. 3, 1-114. Wolpert, D.M., Ghahramani, Z. & Jordan, M.l. (1995). An internal model for sensorimotor integration. Science. 269. 1880-1882. Yoshino, K., Mikami, A. & Kubota, K. (1998). Neuronal activities in the ventral premotor cortex during a visually guided jaw movement in monkeys. Neuroscience Research. 30. 321- 332. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Subject 9 : Subject 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Subject 1 1 : Subject 12: 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Subject 6: Experiment 3 190 (O a o . o Subject 7 Cl a o . C i Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 191 Subject 8 Subject 9 a o . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192 Subject 10 Q a c j o . Q C X 3 0 2 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 of 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 affect 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. 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. ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 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 3018077 _ _ _ < fi) UMI UMI Microform 3018077 Copyright 2001 by Bell & Howell Information and Learning Company. Ail rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Bimanual coordination of goal-directed multi-joint rapid aiming movements following stroke
PDF
Biomechanical and neuromuscular aspects of non-contact ACL injuries: The influence of gender, experience and training
PDF
Factors contributing to patellofemoral joint stress: a comparison of persons with and without patellofemoral pain
PDF
Control and dynamics during horizontal impulse generation
PDF
Biomechanical and neuromuscular aspects of non-contact ACL injuries: the influence of gender experience and trainting
PDF
Human and environmental factors contributing to slip events during walking
PDF
The effects of jump strategy modification on energetics during dance jumps
PDF
Learning self-agency during a contingency paradigm in infants at commnity and elevated risk of autism
PDF
Fluid balance and neuropeptides in the lateral hypothalamic area
PDF
At what age are gait characteristics mature? Evaluation of gait kinematics, kinetics, and intersegmental dynamics in 7 year-old children
PDF
Biochemical evidence for calpain's involvement in long term synaptic plasticity
PDF
The influence of eccentric resistance training on bone mass and biochemical markers in young women
PDF
Isolation and characterization of mouse Zac1, a novel co-regulator for transcriptional activation by nuclear receptors and p53
PDF
Assessment Of Beta-Cell Function In Vivo
PDF
Dopamine related temporal processing errors in prenatally stressed rats
PDF
Cardiovascular fitness, exercise, and metabolic disease risk in overweight Latino youth
PDF
The brain and behavior of motor learning: the what, how and where
PDF
Cost -efficient design of main cohort and calibration studies where one or more exposure variables are measured with error
PDF
Correlates of smoking behaviour among Egyptian adolescents
PDF
A typology of corporate environemental strategy and its driving factors in jultinational corporations
Asset Metadata
Creator
Fisher, Beth Ellen
(author)
Core Title
Functional brain correlates for premovement planning and compensatory adjustments in rapid aimed movement
School
Graduate School
Degree
Doctor of Philosophy
Degree Program
Biokinesiology
Degree Conferral Date
2000-08
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
biology, neuroscience,health sciences, recreation,OAI-PMH Harvest,psychology, experimental
Language
English
Contributor
Digitized by ProQuest
(provenance)
Advisor
Winstein, Carolee (
committee chair
), Azen, Stanley (
committee member
), Baker, Lucinda (
committee member
), Bradley, Nina (
committee member
), McNitt-Gray, Jill (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c16-82915
Unique identifier
UC11338724
Identifier
3018077.pdf (filename),usctheses-c16-82915 (legacy record id)
Legacy Identifier
3018077.pdf
Dmrecord
82915
Document Type
Dissertation
Rights
Fisher, Beth Ellen
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
biology, neuroscience
health sciences, recreation
psychology, experimental