Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Attention demands across extended practice of a bimanual coordination task Bredin, Shannon Stephanie Deanne 2004

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_2004-931013.pdf [ 22.01MB ]
Metadata
JSON: 1.0077415.json
JSON-LD: 1.0077415+ld.json
RDF/XML (Pretty): 1.0077415.xml
RDF/JSON: 1.0077415+rdf.json
Turtle: 1.0077415+rdf-turtle.txt
N-Triples: 1.0077415+rdf-ntriples.txt
Original Record: 1.0077415 +original-record.json
Full Text
1.0077415.txt
Citation
1.0077415.ris

Full Text

Attention Demands across Extended Practice of a Bimanual Coordination Task by SHANNON STEPHANIE DEANNE BREDIN B.P.E. with distinction, The University of Alberta, 1996 B.Ed, with distinction, The University of Alberta, 1996 M.Sc, The University of Alberta, 1998 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES School of Human Kinetics We accept this thesis confirming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA June 22, 2004 © Shannon Stephanie Deanne Bredin, 2004  ABSTRACT Five experiments were conducted, the overall purpose of which was to examine the effects of practice on attention demands of a new bimanual coordination task. In Experiments 1-3 participants received 1, 000 trials to learn a 90° out-of-phase task. The secondary task from which attention demands were derived was probe reaction time (probe-RT). Although cognitive demands decreased with practice, results showed that performance of the 90° pattern continued to demand attention, even after extended practice. Similar results were found in Experiment 4 when examining the attentional costs of performing naturally occurring coordination tendencies (i.e., in-phase, anti-phase). These findings indicate that a minimal level of cognitive control is required for the execution of newly acquired bimanual coordination tasks, as well as for the performance of intrinsic coordination biases. These experiments also examined the influence of continuous on-line visual feedback on learning and attention demands. Individuals received concurrent visual feedback for 65% of each practice trial (Experiment 1) or for only 35% of each trial (Experiment 2). Results showed that participants were highly successful at producing the required task when visual feedback was available, but were less able to inhibit the influence of pre-practice biases whenever visual feedback was unavailable to guide performance. Experiment 3 examined the influence of manual guidance on performance of the 90° pattern under limited conditions of visual feedback. When participants' limbs were physically moved through the required movement via servo torque motors individuals were better able to break away from pre-existing tendencies. Finally, Experiment 5 revealed that probe-RT was slower when performing in the absence of visual feedback, in comparison to its presence. Attentional requirements also increased whenever the dominant source of feedback was changed within a trial, although adding visual feedback to the perceptual display resulted in only a temporary increase in probe-RT. In addition to examining the attention demands of bimanual coordination, these investigations highlight the importance of exploring instructional strategies that reduce the negative effects of pre-existing behavioural tendencies on the learning of new complex motor tasks.  TABLE OF CONTENTS Page Abstract  ii  Table of Contents  iii  List of Tables  x  List of Figures  xi  Acknowledgements  xix  Overview of Document  xxii  Statement of Ethics  xxiii  CHAPTER 1 Experiment 1 Attention Demands and the Effects of Continuous On-line Feedback during Extended Practice of a New Bimanual Coordination Task Introduction Skilled Motor Behaviour Assessing Attention Demands Bimanual Coordination Overview of Experiment Method Participants Primary Task and Apparatus Secondary Task and Apparatus Procedure Acquisition Immediate No-Feedback Transfer Tests Delayed Retention and No-Feedback Transfer Tests Data Analyses Primary Task Performance: 90° Relative Phase (RP) Secondary Task: Probe-RT Results and Discussion Question 1: Did Performance of the Required Pattern Improve with Practice? Changes in Accuracy and Stability across Acquisition Changes in Cycle Frequency across Acquisition Individual Performance Data Question 2: Did Performance Deteriorate with the Withdrawal of Concurrent Visual Feedback?  1 1  1 2 3 4 6 8 8 9 9 14 14 15 16 16 16 18 18 18 18 22 22 26  iv  Vision versus No-Vision during Acquisition 26 Pattern Accuracy and Stability under Conditions of Transfer 35 Question 3: Did Extended Practice on a Bimanual Coordination Task Reduce and Eventually Eliminate Dual-task Interference?...39 Probe-RT during Acquisition 39 Probe-RT during Transfer 46 General Discussion 46 Chapter Footnotes 49 CHAPTER 11 Experiment 2 Effects of Reducing Continuous On-line Visual Feedback on Attention Demands and Extended Learning of a New Bimanual Coordination Task  53 53  Introduction 53 Method 57 Participants 57 Apparatus and Tasks 57 Procedure 58 Data Analyses 58 Results 59 Question 1: Did Performance of the Required Pattern Improve with Practice? 59 Changes in Accuracy and Stability across Acquisition 59 Changes in Cycle Frequency across Acquisition 60 Performance of the 90° out-of-phase task improved with practice 60 Question 2: Did Performance Differences Appear between Segments? 63 Changes in Pattern Accuracy and Stability across Acquisition 63 Performance in Relation to the Criterion Pattern 63 Individual Performance Data 66 Cycle Frequency 68 Pattern Accuracy and Stability across Experiments 71 Performance of the 90° pattern generally improves in the presence of concurrent visual feedback 73 Question 3: Did a Performance Advantage Emerge in No-Feedback Transfer Tests for Learners Receiving Limited Relative Motion Feedback during Acquisition? 73 Question 4: Did Extended Practice on a Bimanual Coordination Task Reduce and Eventually Eliminate Dual-task Interference?...74 Probe-RT during Acquisition 74 Probe-RT during Transfer 77  Dual-task interference was reduced, but not eliminated following extended practice of the 90° pattern Discussion Chapter Footnotes CHAPTER III Experiment 3 Effects of Manual Guidance on Attention Demands and Extended Learning of a New Bimanual Coordination Task  77 77 82 86 86  Introduction 86 Manual Guidance 86 Overview of Experiment 88 Method 90 Participants 90 Apparatus 90 Procedure 90 Results 91 Question 1: Did Performance of the Required Pattern Improve with Practice? 92 Changes in Accuracy and Stability across Acquisition 92 Changes in Cycle Frequency across Acquisition 95 Individual Performance Data 95 Performance of the 90° pattern improved with practice 98 Question 2: Did Performance Differences Appear between Segments? 98 Changes in Pattern Accuracy and Stability across Acquisition 98 Cycle Frequency 101 Pattern Accuracy and Stability across Experiments 101 Performance of the 90° pattern is superior in the presence of concurrent visual feedback 104 Question 3: Did a Performance Advantage Emerge in No-Feedback Transfer Tests when Compared to Experiment 1? 104 Question 4: Did Extended Practice on a Bimanual Coordination Task Reduce and Eventually Eliminate Dual-task Interference?.106 Probe-RT during Acquisition 106 Probe-RT during Transfer 108 Performance of a 90° out-of-phase task remains Attention demanding following extended practice 108 Discussion 109 Chapter Footnotes 112  vi  CHAPTER IV 113 Experiment 4 113 Attention Demands and the Performance of Intrinsic and Learned Bimanual Coordination Patterns. Introduction Attention Demands of Preferred Coordination Tendencies Method Participants Primary Task and Apparatus Secondary Task and Apparatus Procedure Data Analyses Performance of Bimanual Coordination Patterns Central Costs Results Performance of Bimanual Coordination Patterns Pattern Accuracy and Stability Cycle Frequency Probe-RT Discussion Chapter Footnotes CHAPTER V Experiment 5 Removal of Continuous On-Line Visual Feedback Degrades Performance and increases the Attentional Demands of a Bimanual Coordination Pattern Introduction Limitations in Experimental Design: The Segment Problem' Method Participants Apparatus Primary and Secondary Tasks Procedure Session 1 Baseline Trials: Probe-RT Baseline Trials: Intrinsic Coordination Patterns Practice Trials Session 2 Practice Trials Session 3 Test Trials Baseline Trials: Intrinsic Coordination Patterns Baseline Trials: Probe-RT Data Analyses  113 113 116 116 117 117 117 121 121 121 122 122 .122 124 126 129 134 135 135  135 135 139 139 140 140 140 141 141 141 142 142 142 142 .142 143 143 145  Results Baseline RT Intrinsic Coordination Patterns Practice Trials Test Trials Performance of the 90° out-of-phase Pattern Probe-RT Discussion Chapter Footnotes CHAPTER VI Attention Demands across Extended Practice of a Bimanual Coordination Task: General Discussion  145 145 145 146 149 149 152 154 158 160  Attention Demands: Theoretical Perspectives 161 Logan's (1988) Instance Theory of Automaticity 164 Anderson's (1982) Cognitive Framework of Skill Acquisition.... 164 Is Bypassing Attention a Good Thing? 166 Error Detection and Correction 167 Automatic Processing Arrests Development of Skilled Performance 168 Attention Demands, Augmented Feedback and Bimanual Coordination 169 REFERENCES APPENDIX A Review of Literature Understanding Learning Phenomena: A Foundation  172 189 189  Section One The Nature of Skilled Behaviour: An Introduction Perspectives of Expert Performance: Innate Talent of Practice? Acquiring Task Proficiency Investigating Practice  190 190  Section Two Skill Acquisition from an Information-Processing Perspective Representational-based Accounts of Skill Acquisition Classic Theories of Skill Acquisition Schmidt's (1975, 1980) Schema Theory Phases of Skill Acquisition Adams' (1971) Verbal-motor and Motor phases of Learning  196 196 199 199 200 201  191 193 195  203  Fitts' Three-Phase Theory of Skill Acquisition 204 Summary 207 Investigating the Role of Attention in Motor Performance....208 The Study of Divided Attention 209 Capacity-Sharing (or Resource) Models 210 Assessing Attention Demands 212 Probe Reaction Time Technique (Probe-RT) 215 Summary 218 The Relation between Attention and Automaticity 218 Automaticity as a Learned Phenomena 222 Limitations of Current Theory and Empirical Work....225 Summary 228 Section Three Skill Acquisition from a Dynamic Pattern Perspective Fundamental Concepts and Terminology Dynamic Modeling of Interlimb Rhythmic Coordination The HKB Theoretical Model (1985) of Bimanual Coordination Summary Coordination Dynamics and Attention Intentional Modulation of Intrinsic Dynamics Effects of Attention on Coordinated Behaviour Summary Motor Learning and Dynamic Pattern Theory Mapping Coordination Dynamics 90° RP: The Only To-be-learned Coordination Pattern? Optimizing the Learning Process Summary  229 229 231  Section Four An Integrative Experimental Approach to Understanding Learning Phenomena  257  APPENDIX B Experiment 6 Using the Scanning Procedure for Mapping Dynamical Landscapes: Methodological Implications Introduction General Scanning Methodology: Experiments 1-3 Apparatus Presentation of Relative Phasing Patterns (Mode of Scanning)  234 235 239 240 241 242 246 247 250 251 253 256  257 261 261  261 264 264 264  ix  Experiment 1 Procedure Data Analyses Results and Discussion Experiments 2 and 3.... Procedure.. Results and Discussion Evaluation of Scanning Methods for Mapping Dynamical Landscapes Method Participants Apparatus Procedure Orientation Scanning Task Acquisition Task Dependent Measures and Analyses Results Discussion  266 266 266 267 271 271 271 283 284 284 284 285 285 286 286 288 289 293  X LIST OF TABLES Page Table 1.1.  Mean cycle frequency during the vision and no-vision segment for each participant collapsed across block  32  Mean cycle frequency during the no-vision and vision segment for each participant collapsed across block  70  Mean cycle frequency during the no-vision and vision segment for each participant collapsed across block  102  Table 5.1.  Summary of Feedback Conditions  144  Table B1.  Individual Participant Data from 0° to 180° RP for Discrete Preand Post-Scans of the Coordination Dynamics, Experiments 2 and 3  282  Table 2.1. Table 3.1.  LIST OF FIGURES Page Figure 1.1. Figure 1.2. Figure 1.3.  Figure 1.4. Figure 1.5.  Figure 1.6.  Figure 1.7.  Figure 1.8.  Figure 1.9.  Apparatus and Experimental Set-Up  10  Probe positions of auditory tone displayed via a Lissajous Figure  13  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) across 9 days of acquisition (Error bars = SEM)  20  Percent Improvement from Day 2 for RMSE of goal relative phase (top panel) and SD of RP (bottom panel)  21  Performance during the vision segment for Participants 2 (top panel) and 5 (bottom panel) on selected acquisition trials, Day 1 (x-axis = right forearm displacement; y-axis = left forearm displacement)  23  Performance during the vision segment for Participant 6 on selected acquisition trials, Day 1 (x-axis = right forearm displacement; y-axis = left forearm displacement)  25  Performance during the vision segment for Participant 1 on selected acquisition trials, Day 1 (x-axis = right forearm displacement; y-axis = left forearm displacement)  27  Performance during the vision segment for Participant 5 on selected acquisition trials across days (x-axis = right forearm displacement; y-axis = left forearm displacement)  28  RMSE of goal relative phase for the vision (left panel) and the no-vision segment (right panel) across 9 days of acquisition (Error bars = SEM)  30  Figure 1.10. Standard deviation of relative phase for the vision (left panel) and the no-vision segment (right panel) across 9 days of acquisition (Error bars = SEM).....  31  Figure 1.11. Individual trial data across acquisition displayed as Lissajous figures for Participant 2. The solid line represents the vision segment, and the dotted line represents the no-vision segment (x-axis = right forearm displacement; y-axis = left forearm displacement)  34  Figure 1.12. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for condition (acquisition, transfer) across days (Error bars = SEM)  37  Figure 1.13. Individual trial data displayed as Lissajous figures for Participant 5. The left panel illustrates select acquisition trials whereby the solid line represents the vision segment, and the dotted line represents the no-vision segment. The right panel illustrates select no-feedback transfer trials (x-axis = right forearm displacement; y-axis = left forearm displacement)  38  Figure 1.14. Individual trial data displayed as Lissajous figures for Participant 2. The left panel illustrates select acquisition trials whereby the solid line represents the vision segment, and the dotted line represents the no-vision segment. The right panel illustrates select no-feedback transfer trials (x-axis = right forearm displacement; y-axis = left forearm displacement) 40 Figure 1.15.  Curvilinear (polynomial) relationship between RMSE and trial number under no-probe and probe conditions for the vision segment of a representative participant  42  Figure 1.16. Relationship between RMSE and trial number under no-probe and probe conditions for the no-vision trial segment of a representative participant  43  Figure 1.17. Mean probe-RT for acquisition (top panel) and transfer (bottom panel) plotted as a function of day and compared to baseline RT (M = 402 ms) (Error bars = SEM)  45  Figure 2.1. Figure 2.2.  Figure 2.3.  Figure 2.4.  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) across 9 days of acquisition (Error bars = SEM)  61  Percent Improvement from Day 2 for RMSE of goal relative phase (top panel) and SD of RP (bottom panel)  62  RMSE of goal relative phase for the no-vision (left panel) and the vision segment (right panel) across 9 days of acquisition (Error bars = SEM)  64  Standard deviation of relative phase for the no-vision (left panel) and the vision segment (right panel) across 9 days of acquisition. (Error bars = SEM)  65  Figure 2.5.  Figure 2.6.  Figure 2.7.  Figure 2.8.  Figure 2.9.  Mean RP for Participant 4 plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment (top panel). Performance is also presented in the form of Lissajous figures for select trials (bottom panel). The solid line represents the no-vision segment, and the dotted line represents the vision segment (x-axis = right forearm displacement; y-axis = left forearm displacement)  67  Mean RP for Participant 3 plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment (top panel). Performance is also presented in the form of Lissajous figures for select trials (bottom panel). The solid line represents the no-vision segment, and the dotted line represents the vision segment (x-axis = right forearm displacement; y-axis = left forearm displacement)  69  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) plotted separately for each trial segment and compared across experiments. The left panel represents the 8.5 s segment and the right panel represents the 3 s segment (Error bars = SEM)  72  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for immediate and delayed no-feedback transfer tests across days, Experiment 1 vs. Experiment 2 (Error bars = SEM)  75  Mean probe-RT for acquisition (top panel) and transfer (bottom panel) plotted as a function of day and compared to baseline RT (M = 423 ms) (Error bars = SEM)  .76  Figure 2.10. Mean RP for Participant 2 (top panel) and Participant 5 (bottom panel) plotted in blocks of 20 trials across 9 days of acquisition for the no-vision segment and the vision segment. Performance within a bandwidth of ±10 of the criterion pattern did not occur until late stages of learning  .83  Figure 2.11. Mean RP for Participant 1 (top panel) and Participant 6 (bottom panel) plotted in blocks of 20 trials across 9 days of acquisition for the no-vision segment and the vision segment. Relatively accurate performance of the 90° pattern emerged early in learning  84  Figure 3.1.  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) across 9 days of acquisition (Error bars = SEM)  93  xiv Figure 3.2. Figure 3.3. Figure 3.4.  Figure 3.5.  Figure 3.6.  Percent Improvement from Day 2 for RMSE of goal relative phase (top panel) and SD of RP (bottom panel)  94  Schematic description of the 'box strategy' employed by participants to learn the 90° pattern  96  Individual trial data from Day 1 displayed as Lissajous figures for Participant 6. The solid line represents the no-vision segment, and the dotted line represents the vision segment (x-axis = right forearm displacement; y-axis = left forearm displacement)  97  RMSE of goal relative phase for the no-vision (left panel) and the vision segment (right panel) across 9 days of acquisition (Error bars = SEM)  99  SD of relative phase for the no-vision (left panel) and the vision segment (right panel) across 9 days of acquisition (Error bars = SEM)  100  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) plotted separately for each trial segment and compared across experiments. The right panel represents the 8.5 s segment and the left panel represents the 3 s segment (Error bars = SEM)  103  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for immediate and delayed no-feedback transfer tests across days and experiment (Error bars = SEM)  105  Mean probe-RT for acquisition (top panel) and transfer (bottom panel) plotted as a function of day and compared to baseline RT (M = 411 ms) (Error bars = SEM)  107  Figure 4.1.  Illustration of in-phase and anti-phase coordination  118  Figure 4.2.  RMSE of goal relative phase for pattern (0°, 180°, and 90°) by trial segment (Error bars = SEM)  123  Standard deviation of relative phase (SD of RP) for pattern (0°, 180°, and 90°) by trial segment (Error bars = SEM)  125  Self-paced cycle frequency for pattern (0°, 180°, and 90°) by block (Error bars = SEM)  127  Figure 3.7.  Figure 3.8.  Figure 3.9.  Figure 4.3.  Figure 4.4.  XV Figure 4.5. Mean probe-RT for 0°, 180°, and 90° of relative phase compared to baseline RT. An asterisk indicates statistical significance when baseline RT was compared with the corresponding bimanual coordination pattern (Error bars = SEM)  128  Figure 5.1.  Breakdown of the 13 s practice trial in Experiments 1 through 4. The top panel represents the trial design used in Experiment 1 and for the Vision group in Experiment 4. The bottom panel illustrates the trial design used in Experiments 2 and 3, as well as for the No-Vision group in Experiment 4 136  Figure 5.2.  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for preferred coordination modes (0° and 180°) pre- and post-acquisition (Error bars = SEM)  147  Mean probe-RT for 0° and 180° of relative phase compared to baseline RT (M. = 392 ms). An asterisk indicates statistically significant difference when baseline RT was compared with the corresponding measure of probe-RT (Error bars = SEM)  148  RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for condition presented by trial segment. An asterisk indicates that a statistically significant difference was found when comparing segments within a respective condition (Error bars = SEM)  150  Individual trial data displayed as Lissajous figures for Participant 3. Panel A through D represents the V-V, V-NV, NV-V, and NV-NV conditions, respectively. The solid line always represents a vision segment, while the dotted line always represents a no-vision segment (x-axis = right forearm displacement, y-axis = left forearm displacement)  151  Mean probe-RT plotted as a function of segment and probe position for each condition compared with baseline RT (M = 391 ms)  153  Individual trial data for Participants 1 (top panel) and 9 (bottom panel). Mean RP is plotted across 100 trials for Day 2 of practice  158  Figure 5.3.  Figure 5.4.  Figure 5.5.  Figure 5.6.  Figure 5.7.  xvi  Figure 5.8.  Figure A1. Figure A2. Figure A3.  Figure A4.  Figure A5.  Mean probe-RT compared with baseline RT (M = 391 ms) for the V-NV and NV-V conditions (top panel) and the V-V and NV-NV conditions (bottom panel) plotted as a function of segment and probe position (Error bars = SEM)  159  Whiting's (1975) human performance model as depicted in Marteniuk(1976)  198  Theoretical assumptions underlying secondary-task methodology (adapted from Abernethy, 1988, 1993)  214  The Secondary-task paradigm. Theoretical assumptions underlying the achievement of automaticity using the probe-RT technique  227  In A, the potential V for b = 0, demonstrating a monostable system with a single basin of attraction generated between -180° and +180° at 0° relative phase. In B, the potential V for a = 0, illustrating a bistable system where two basins of attraction are generated at ±180° and 0° relative phase (adapted from Haken et al., 1985)  237  A series of potential fields generated by changing the ratio b/a, which represents the control parameter of cycling frequency. The upper left panel signifies the potential at the slowest cycling frequency, while he lower right panel corresponds to the potential at the fastest cycling frequency. As cycling frequency increases, the basin of attraction at 180° begins to destabilize until a critical frequency is reached, and the black ball falls into the basin of attraction at 0° (i.e., a phase transition occurs from the potential well at 180° to the more stable potential well at 0°) (adapted from Haken et al., 1985)  238  Figure A6.  In A, two minimal potentials are represented at 0° and 180°, which represent the intrinsic dynamics of a system prior to learning episodes. In B, the modified intrinsic landscape is displayed after learning episodes specifying the acquisition of a required relative phase of 90°. The resulting minimal potential at 90° is shifted away from the required relative phase, which indicates greater error in performance. In addition, the resulting potential well is wider signifying greater variability in performance in comparison to the potential wells at 0° and 180° (adapted from Zanone & Kelso, 1993) 249  Figure A7.  Relative phase plots representing 0° (top panel), 180° (middle panel), and 90° (bottom panel) of relative phase. Typical  Figure B1. Figure B2.  Figure B3.  Figure B4.  Figure B5.  Figure B6.  Figure B7.  Figure B8.  Figure B9.  criterion templates for each pattern are presented in the right panel, while examples of actual participant performance under conditions of continuous visual feedback are illustrated in the left panel  254  Two boxes, displayed in the center of the computer screen 5 cm apart, served as visual metronomes  265  Illustration of the onset of each flash for the bimanual coordination patterns of in-phase (right panel) and anti-phase (left panel) for one complete movement cycle  265  Required relative phase and observed relative phase as a function of time for continuous pre-practice (top panel) and post-practice (bottom panel) scanning runs for Participant 6, Experiment 1  268  Required relative phase and observed relative phase for discrete pre- and post-practice scanning runs for Participant 4 (Experiment 2)  273  Mean RP for Participant 4 (Experiment 2) plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment  274  Required relative phase and observed relative phase for discrete pre- and post-practice scanning runs for Participant 3 (Experiment 2)  275  Mean RP for Participant 3 (Experiment 2) plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment  276  Required relative phase and observed relative phase for discrete pre- and post-practice scanning runs for Participant 1 (Experiment 2)  278  Mean RP for Participant 1 (Experiment 2) plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment  279  Figure B10. Required relative phase and observed relative phase for discrete pre- and post-practice scanning runs for Participant 2 (Experiment 3)  280  xviii  Figure B11. Mean RP for Participant 2 (Experiment 3) plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment  281  Figure B12. A visual presentation of the Lissajous templates. There are 12 required relatives phase patterns (0° to 360° in steps of 30°)  287  Figure B13. RMSE of goal relative phase for the four experimental groups, as a function of time (early vs. late acquisition)  290  Figure B14. RMSE of goal relative phase for Lissajous learning groups, as a function of time (acquisition vs. scanning)  291  Figure B16. RMSE of goal relative phase for Flashing square learning groups, as a function of time (acquisition vs. scanning)  292  xix  ACKNOWLEDGEMENTS This thesis would not be complete without expressing my personal gratitude to the following individuals.  I would first like to thank Dr. Romeo Chua, my Ph.D. supervisor, who has shown me the upmost of respect and support throughout my years as a graduate student both at the University of Alberta and at the University of British Columbia. Coming to Vancouver has provided me a world of new experiences and for that I say thank you. I would also like to extend my appreciation to Dr. Ian M. Franks and Dr. Daniel J. Weeks for providing guidance and feedback throughout this series of investigations. Special thanks also goes to the remaining members of my examining committee: Dr. Lawrence Ward (Chair), Dr. J. Timothy Inglis, Dr. James T. Enns, and Dr. Janet L. Starkes for reading this document and providing me thought provoking questions and comments. I would also like to give sincere thanks to Paul Nagelkerke for his technical support. In addition, I must also acknowledge my motor behaviour colleagues in Human Kinetics. Our discussions have given rise to a plethora of ideas we have only just started to investigate. I would especially like to acknowledge: Mr. Tony Carlsen, Dr. Mike Garry, Dr. Nicola Hodges, Mrs. Jennifer Ibbotson, Mr. Dana Maslovat, Dr. Tim McGarry, and Ms. Sara Sturrock. And of course, I extend my appreciation to all of the participants who gave freely of their time to engage in these investigations. I would also like to acknowledge Dr. Sandy K.E. Romanow, my friend and mentor, who has shared this journey with me from the very beginning. We will never know where the road would have led me if I had not taken your undergraduate class so many years ago, but I do know that my life has been enriched because of it. Thank you for sharing your knowledge, your creativity, and most of, thank you for your friendship. Finally, I would like to thank those dearest to my heart - although this is the most difficult to write because there are not enough words to truly express my appreciation. First, I would like to thank my parents, Dale and Vivian Bredin, as well as my brother, Fletcher. As I have said before, my family has always provided me  XX  with unlimited opportunities and put up with every inconvenience so that I may pursue my goals. Thank you for being my foundation. On that note, I would like to acknowledge my loving husband, Dr. Darren E.R. Warburton. You are not only my best friend but also my greatest colleague. Thank you for standing by my side and lending me the strength I needed to overcome this obstacle. Your love and support truly inspires me. And finally, I would like to share this thesis with Mr. Cecil Bredin, my grandpa, and Mrs. Marjory Rennick, my husband's grandmother, who both passed away this year. We thank you for all that you have given us. Although, I miss my grandpa more with each passing day, I know it was him who guided me through the long endless nights of writing when I thought I could write no more - you will always remain one of the wisest and wittiest men I will ever have the pleasure of knowing.  Cecil Douglas Bredin December 31 , 1920 - November 10 , 2003 st  th  XXII  OVERVIEW OF DOCUMENT The following dissertation consists of five experiments, presented in Chapters 1 through 5, respectively. In these experiments, attention demands and bimanual coordination were examined, as well as the effects of manipulating availability of continuous on-line visual feedback and the influence of manual guidance for learning new coordination skills. Chapter 6 provides an overview of the findings reported in this thesis focusing discussion on theoretical issues surrounding attention demands and skilled motor performance. Finally, two appendices are included. Appendix A is an extensive review of literature, which sets the groundwork for topics discussed in the present line of investigations. Some readers may elect to read the review of literature first. Appendix B presents a sixth study conducted with fellow colleagues that was directly relevant to the aforementioned studies. This final experiment examined methodological procedures used to assess learning of a bimanual coordination pattern.  STATEMENT OF ETHICS All experiments conducted in this dissertation were carried out according to the ethical guidelines set by the University Behavioural Science Screening Committee at the University of British of Columbia for research involving human participants.  1  CHAPTER I EXPERIMENT 1 Attention Demands and the Effects of Continuous On-Line Feedback during Extended Practice of a Bimanual Coordination Task Introduction In numerous everyday activities individuals are required to produce more than one goal-related task at the same time. Unfortunately, performing two tasks concurrently can prove difficult and the quality of performance is often less than if each task had been performed alone. Explanations for this common occurrence are often based on resource theories of attention, which postulate the existence of an inherent limited pool (or pools) of processing capacity (e.g., Kahneman, 1973; Navon & Gopher, 1979; Wickens, 1980, 1984). According to this perspective, some (or even all) of the limited processing resources are drawn upon or allocated to the performance of a task. Moreover, this limited resource capacity may be partitioned in a flexible and voluntary manner if the simultaneous performance of two or more events is required. When two concurrent tasks use the same resource pool and the joint demands exceed the total capacity of available resources, a decrement incurs in one of, or in both of, the simultaneously performed tasks. Referred to as dual-task interference, any improvement in performance on one task comes at the expense of performance on the other task (e.g., Kahneman, 1973; Norman & Bobrow, 1975) (Refer to Appendix A-The Study of Divided Attention). One captivating position of this theoretical perspective of attention concerns the effects of practice. A number of investigations have shown that the concomitant performance of two or more tasks can be substantially improved upon as a function of practice (e.g., Spelke, Hirst, & Neisser, 1976), and this capability may be a necessary aspect of expert levels of performance (Abernethy, 1993). According to capacity sharing theories, extended practice on a task leads to 'automaticity'. One perspective, referred to as the modal view (see Logan, 1988, 1992), defines automaticity as processing without attention, whereby practiced operations no longer impose demands on a system's resources for task execution. Performance  2  proceeds without the voluntary allocation of attention to various task components (Laberge & Samuels, 1974; Schneider & Shiffrin, 1977) and mental operations function without experiencing interference from, or generating interference with, ongoing processing activities of concurrent tasks (Pashler, 1998). In essence, practice permits the development of the ability to bypass attention (e.g., LaBerge & Samuels, 1974; Shiffrin & Schneider, 1977) with the underlying assumption that more practice on a task leads to more automaticity (Brown & Bennett, 2002)(7?efer to Appendix A - The Relation between Attention and Automaticity). Although the acquisition of automaticity is operationally interpreted as the gradual withdrawal of attention, it is generally accepted that complete elimination of dual-task interference is not necessary to demonstrate automatic processing. Rather, evidence of automaticity is taken whenever there is a significant decrease in dual-task interference following practice of a task (Brown & Bennett, 2002). In recent years, researchers have begun questioning whether elimination of dual-task interference is actually achievable, as suggested in the basic theoretical tenets of a capacity perspective (see Pashler, 1998), even after extended periods of practice (e.g., Ahissar, Laiwand, & Hockstein, 2001; Blischke, 1998). In contrast, even well practiced motor tasks operating at consistently high levels of excellence are postulated to demand attention. This idea is examined in the present experiment, whereby the primary purpose was to investigate the effects of practice on attention demands and dual-task interference across the extended acquisition of a novel motor task. Skilled Motor Behaviour The association between extended practice and the notion of automaticity is a central tenet in traditional theories of motor skill acquisition, which assume that learning progresses in phases that can be differentiated on the basis of the cognitive activity associated with the production of the response (e.g., Adams, 1971; Fitts & Posner, 1967). As learning progresses from an inexperienced to skilled state of behaviour, it is understood that a shift occurs from the conscious and deliberate allocation of attention to the use of automatic processes. Diminishing demands on  3  attention are advantageous to the performer in that processing load is reduced, thereby permitting the individual to focus on higher-order aspects of performance (Refer to Appendix A - Representational-Based Accounts of Skill Acquisition). Despite these theoretical claims, studies examining the acquisition of novel motor tasks rarely include periods of prolonged training. One reason for this lack of empirical foundation is that investigating the acquisition of a complex motor task mandates extended observation before an individual begins to approximate high degrees of task proficiency. Therefore, the amount of practice trials typically employed in learning studies have been limited, constraining investigation to the early acquisition stages of the learning process (see Adams, 1987, for a review). Recognizing the limitations in studying novices, a more recent descriptive approach has been to compare the characteristics of performers at various levels of skill. However, the limitation of protocols using highly skilled performers or novice-expert comparisons is that it only identifies discriminative factors between skilled and less skilled performers, prohibiting inferences as to the cause of these differences. Although many factors have been investigated using this approach - the recall of briefly presented information (e.g., Allard, Graham, & Paarsalu, 1980), visual search patterns (e.g., Savelsbergh, Williams, van der Kamp, & Ward, 2002), cue recognition within the environmental display (e.g., Abernethy & Russell, 1987) - the manner in which skilled and less skilled performers allocate attention for the execution of a task, especially as it unfolds in real-time, has yet to be adequately examined (Beilock, Carr, MacMahon, & Starkes, 2002). Taking these ideas together, the evolution ofcattentional mechanisms across the acquisition of a complex motor task has received little focus, and is not well understood.  Assessing Attention Demands One way to assess attention demands of a given task is to employ a secondary-task paradigm. This approach is based on having participants perform two tasks concurrently - a basic task, which is referred to as the primary task, and a secondary task. The primary task is the task for which attention demands are assessed, while performance on the secondary task is used to infer the attention  4  demands required by the primary task. In the secondary-task paradigm, participants are instructed to maintain a given level of primary task performance during the concurrent performance of the second task. If poor performance is displayed on the secondary activity, it is generally assumed that a large proportion of informationprocessing capacity has been allocated to the primary task to maintain an acceptable level of performance. The more demanding a primary task is the greater proportion of attentional capacity is required for task execution. Therefore, at early stages of learning, secondary task performance should be relatively poor since it is postulated that the primary task demands a large portion of the learner's processing capacity. After an extended period of task-specific practice, secondary task performance should become relatively unimpaired because primary task demands have diminished. Eventually, the secondary task should be performed just as well under dual conditions, as performed alone. The absence of any single-to-dual task decrement in secondary task performance is assumed to indicate that the primary task is performed 'automatically', requiring little or none of the performer's attentional capacity (for reviews of dual-task methodology, see Abernethy, 1988, 1993) (Refer to Appendix A - Assessing Attention Demands). To examine attention demands across the time course of learning, we asked individuals to learn a continuous bimanual coordination movement. The secondary task, from which attention demands of the primary task were derived, was the classic measure of discrete reaction time (RT). In the dual-task condition, participants were instructed that attentional priority should be given to the bimanual coordination task such that the secondary task does not interfere with the primary task even if performance on the RT task decreases. Bimanual Coordination In recent years, there has been a trend towards examining more complex, coordinative skills. An advantage of this approach is that these skills are thought to be more representative of real-world motor learning environments (e.g., Hodges & Lee, 1999). Moreover, recent research has shown that the study of interlimb  5  coordination provides a useful framework with which to investigate the acquisition of new motor skills. In studies of bimanual coordination (i.e., requiring concurrent activities of the left and right limb), inherently stable (or preferred) coordination patterns have been identified: a strong symmetrical in-phase relationship and a somewhat weaker antiphase limb relationship (Kelso, 1984). During in-phase (0° of relative phase), the limbs are coordinated to be in the same phase of the cycle at the same time. That is, in-phase coordination refers to the simultaneous contraction of homologous muscles, which requires the limbs to flex and extend in synchrony. In contrast, the limbs are in the opposite phase of the cycle at the same time during anti-phase performance (180° out-of-phase). The limbs flex and extend in alternation because of the simultaneous activation of nonhomologous muscle groups. Both in-phase and anti-phase coordination are thought to be intrinsic to the system, and can be performed skilfully, without practice (Cohen,1971; Kelso, 1984; Tuller & Kelso, 1989; Yamanishi, Kawato, & Suzuki, 1980). However, with practice, recent research has shown that less intrinsic patterns of coordination can emerge as additional, relatively stable patterns, against the backdrop of pre-existing coordination tendencies (e.g., Fontaine, Lee, & Swinnen, 1997; Lee, Swinnen, & Verschueren, 1995; Swinnen, Lee, Verschueren, Serrien, & Bogaerds, 1997; Wenderoth & Bock, 2001; Zanone & Kelso, 1992; 1997) (Refer to Appendix A - Motor Dynamics: Fundamental Concepts and Terminology, as well as Motor Learning and Dynamic Pattern Theory). One of the most common to-be-learned patterns utilized to-date has been the 90° relative phase (RP) pattern. To perform this pattern, participants must learn to coordinate the limbs with an offset of 90° between the limbs, whereby one limb leads the other by a quarter of a cycle. This pattern has received considerable attention within the literature because it is considered to be an unstable point half-way between the stable attractor states of in- and anti-phase (Zanone & Kelso, 1992). The difficulty with learning a 90° out-of-phase pattern is that there is a tendency for the individual to be drawn to the more stable pre-existing coordination pattern of anti-phase, or (less frequently) the in-phase pattern of coordination. However, as learning progresses, the tendency to perform an intrinsic motor pattern diminishes  6  and is replaced by an increased attraction to the criterion pattern (e.g., Lee et al., 1995; Swinnen et al., 1997). Therefore, practice leads to the evolution and stabilization of the required 90° pattern, which involves a relatively permanent change in the pre-existing coordination dynamics of a system towards an attractor layout that incorporates the new phasing pattern (see Zanone & Kelso, 1992). The provision of relative motion feedback has been shown to be a powerful source of information for the acquisition of the 90° out-of-phase task (Lee et al., 1995; Swinnen et al., 1997). However, it remains unclear whether or not individuals become increasingly dependent on this visual information source as learning progresses. Proteau and colleagues (e.g., Proteau, Marteniuk, Girouard, & Dugas, 1987) have provided evidence, which suggests that afferent information sources, especially vision, become increasingly important for motor performance as a function of practice. This view postulates that a single source of afferent input is quickly identified early in learning as optimal, and progressively dominates other information sources during the time course of learning (Tremblay & Proteau, 1998). A corollary of this perspective suggests that movement learning is relatively specific to the sources of information under which practice occurs. If a dominant source of information is withdrawn at later stages of acquisition, performance deteriorates because the individual no longer has a reliable source of reference with which to support his or her performance (see Proteau, 1992, for a review of early work; see Proteau & Isabelle, 2002 for an example of more recent work). Therefore, a secondary purpose of this investigation was to manipulate the provision of relative motion feedback during performance to determine its relative contribution to the acquisition and transfer of coordination skills. More specifically, we were interested in examining whether a specificity of learning effect would also be evident when learning a 90° out-of-phase task across extended practice.  Overview of Experiment In the present experiment, we examined how practice affected the production and stability of a bimanual coordination task corresponding to a 90° out-of-phase pattern (the primary task). Participants engaged in ten sessions of 100 practice trials  7  per day across a two-week period (i.e., 1, 000 practice trials). To facilitate acquisition of the 90° pattern, continuous on-line visual feedback was provided on a computer monitor. Specifically, displacement of the right limb resulted in horizontal movements on the computer screen, while displacement of the left limb resulted in vertical movements. Augmented feedback regarding current performance was displayed by plotting the real-time displacement of the right-limb on the ordinate against displacement of the left-limb on the abscissa. When the individual correctly moved his or her limbs in a 90° offset, the resulting plot (referred to as a Lissajous figure) appeared as a circle configuration on the monitor (e.g., Lee et al., 1995). Concurrent augmented feedback was available to participants for only the first ten seconds of each 13s trial. Participants received no feedback for the final three seconds of the trial, followed by terminal feedback regarding the just-performed no-vision trial segment. Terminal feedback was also presented in the form of a Lissajous trace. Manipulating the provision of afferent information via occlusion allowed us to investigate the role played by vision across prolonged practice of a 90° bimanual coordination pattern. To assess the individual's capability to perform the criterion task in an environment different from learning, participants were required to reproduce the 90° pattern without the use of on-line feedback immediately following 7 of the 9 acquisition sessions, and after a one-week retention interval. The final session was included to assess the extent to which post-practice changes in pattern accuracy and stability were retained across a period of no practice. To determine the time course of attention demands across extended learning of the 90° task, participants were required to vocally respond to a computergenerated auditory signal on one-third of all practice and transfer trials (the secondary task). On probe trials, the auditory signal was either delivered during the vision segment of the trial or during the no-vision segment. This latter manipulation allowed us to investigate the relationship between attention demands and provision of visual sensory information across extended practice. In the present experiment, it was hypothesized that practice would diminish attention demands of the 90° relative phase task, as measured by decreases in  8  reaction time of the secondary task above that when performed alone. A reduction in dual-task interference was predicted to occur following practice of the primary task, but a complete elimination of interference was not expected to emerge despite the extended and highly specific nature of practice. Early in learning, it was also postulated that participants would demonstrate considerable problems in producing the required 90° pattern. More importantly, at these initial stages, withdrawal of concurrent visual feedback was expected to cause a change (or drift) from the produced pattern towards a more intrinsically stable pattern of coordination (e.g., a drift towards an anti-phase pattern of coordination). At early stages of acquisition, we contend that vision is an important information source for learning the required limb phase relationship, especially considering the self-directed nature of the learning environment. Without vision, performance of this coordination pattern should be difficult to maintain prior to establishment of an attractor state at the criterion pattern. When the individual finally acquires the capability to perform a relative phasing of 90° with the limbs, stability of the produced pattern was expected to be maintained even upon withdrawal of relative motion feedback. However, if participants adopt a control strategy where vision is relied upon and processed to the detriment of other sensory sources during execution of the coordination pattern, its withdrawal is expected to have a debilitating effect on accuracy and stability of the required coordination pattern even at the later stages of learning.  Method Participants Informed consent was received from six self-professed right-handed individuals from a university population (n = 5 female; n = 1 male). The mean age of participants was 28.2 years (SD = 7.7, range = 2 3 - 4 3 years). All individuals were naive to the purpose of the experiment and none had previous experience as a participant in a bimanual coordination investigation. Upon completion of the final session, all participants received remuneration. The experiment was carried out according to the ethical guidelines set by the University Behavioural Science Screening Committee for research involving human participants.  9  Primary Task and Apparatus Participants were seated at a table facing a 14 in. colour monitor (Zenith flatscreen, ZCM-1490; VGA 640 x 480 pixels) with their forearms strapped comfortably into two angular manipulanda (Figure 1.1). The manipulanda were positioned such that the elbow joint was aligned with the axis of rotation and the hands were placed palm down on adjustable metal plates. The middle finger of each hand was placed between two vertical pins and secured by Velcro straps around the forearms and hands of the participant. The goal of the primary task was to learn how to correctly move the arms in the horizontal plane to produce a 90° RP pattern of coordination. When the limbs were correctly moved in 90° RP and with similar amplitudes, a circular pattern was produced on the computer monitor. Leading with the right-limb produced a circle pattern in a counter-clockwise direction, while a leftlead created a circle in a clockwise direction. The required amplitude of the movement for each arm was 40°, which corresponded to a 15 cm movement on the computer monitor. Feedback regarding movement amplitude was presented to participants via augmented computer feedback, as well as two table-top markers identifying the boundaries or maximal "IN" and "OUT" positions for each arm (Figure 1 . 1 - bottom right panel). Angular positions of the manipulanda were recorded using two optical encoders (Dynapar, E2025001303), one attached to the shaft of each manipulandum. Both optical encoders were connected to an interface card (Advantech Quadrature, PCI-833), giving a resolution of 10, 000 counts per revolution. Angular position was sampled at a rate of 1000 Hz. An MS-DOS computer running custom software for data collection and analysis controlled the temporal events within each trial.  Secondary Task and Apparatus While performing the primary task, participants were required to emit a vocal response if at any time an auditory 'beep' occurred. The auditory signal was computer generated at 2000 Hz and was presented through a set of multi-media amplified speakers (EP-691 H) (Figure 1.1 - left panel). The duration of the auditory probe was 100 ms. A microphone (C-315 Labtec Headset/Boom microphone),  10  Figure 1.1. Apparatus and Experimental Set-Up. (Left panel) Participants were asked to produce a 90° RP pattern of coordination by making continuous flexion and extension movements about the elbow joint in the horizontal plane. Producing a 90° RP pattern of coordination resulted in a circle pattern on the computer screen situated directly in front of the participant. (Top right panel) A schematic diagram of the criterion circle pattern represented as a Lissajous figure. (Bottom left panel) Amplitude boundaries were identified by two tabletop markers ("IN and OUT"), representing a required amplitude of 40° for each arm. When correctly moving the limbs in a 90° RP pattern of coordination, one limb will lag the other by a quarter of a cycle in time (e.g., when the participant's left limb is reversing direction at the "IN" marker, the right limb should be located midway between the "IN" and "OUT" boundary markers).  11  12  connected to a vocal response time control unit (Lafayette Instruments, 6602 A) and interfaced with the computer, picked up the individual's vocal response. The auditory tone was delivered randomly at one of four possible positions for all trials randomly assigned as a probe condition. Probe positions were referred to as: Probe A (Pr A), Probe B (Pr B), Probe C (Pr C), and Probe D (Pr D), and were defined according to the degree of horizontal displacement of the right and left limbs (see Figure 1.2). For example, Probe A was delivered when a participant displayed 0° displacement with the right limb (i.e., the limb was mid-way between the IN and OUT boundary markers) and the left limb was within a bandwidth of "5° and "25° around the OUT marker. Conversely, Probe B was delivered when a participant displayed 0° displacement with the left limb and the right limb was within a bandwidth of "5° to "25° around the OUT marker. In contrast to in-phase and anti-phase patterns of coordination, performing a 90° out-of-phase pattern implies that reversals in direction are not made simultaneously with both limbs. Therefore, these four conditions were designed to probe attention requirements at critical reversal points within the movement cycle as each probe type corresponded to a point of extreme flexion ("20°) or extension ( 20°) +  of one of the two limbs. Critical reversal points were examined because peak position may demand greater attention when compared to spatiotemporal limb position during the rest of the movement cycle. Bandwidths were designed to maintain the integrity of the probe conditions while compensating for the expected difficulties in producing the pattern at early phases of learning. The auditory tone was delivered at variable times across probe trials. On a probe trial, the computer 'armed' itself at a randomly designated time (i.e., there was now the potential for the delivery of the auditory tone). Once armed, the respective probe 'triggered' (or sounded) when the appropriate limbs displayed 0° displacement and were within the bandwidth described above. If the limbs did not meet the criteria, an auditory tone was not delivered on that trial.  13  Figure 1.2. Probe positions of auditory tone displayed via a Lissajous Figure. On approximately one-third of trials, an auditory stimulus was presented at one of four possible probe (Pr) positions (identified as Pr A, Pr B, Pr C, and Pr D). When one limb is at the mid-point between the IN and OUT boundary markers (0° displacement), the other limb is within a bandwidth of 5° to 25° around the IN marker or within a bandwidth of "5° to "25° around the OUT marker, dependent on the respective probe. +  +  25° 20° 15° 10° 5° +  +  +  +  0°  "5°  "10° "15° "20° "25°  OUT  S O  V3  (mid)  IN  Right Hand Displacement Boundary Labels  +  14  Procedure Acquisition After participants were given an opportunity to familiarize themselves with the apparatus, the criterion circle was displayed on the monitor. Participants were informed that they could achieve this circular pattern by continuously flexing and extending the arms in a particular relationship between the two amplitude markers. Participants were instructed that they could move at their own pace , but the goal of 1  the task was to produce continuous full movement cycles within each acquisition trial that matched as closely as possible to the criterion circle. The criterion for one movement cycle was explained to consist of the full moving distance between the "IN" and "OUT" markers so that each hand returned to the same starting position. Participants were told that they could move their limbs with either a right- or left-lead (i.e., the screen cursor moves in a counter-clockwise or clockwise direction, respectively), but after the first ten trials on Day 1 they were required to move the limbs in the same lead-lag relationship for the duration of the experiment. Participants were also instructed that they needed to learn the required task so that they could eventually perform 90° RP from memory. No instructions were provided to participants concerning how to produce the required circle pattern. All trials were 13 s in duration. A 1000 Hz auditory tone signalled the start and end of each trial. The duration of each of these tones was 100 ms and 300 ms, respectively. For the first 10 s, concurrent visual feedback was provided in the form of a Lissajous figure. Relative motion feedback stayed on the screen throughout the 10 s such that a history plot emerged as the trial progressed and remained there until the completion of the trial. The latency of the real-time display was limited only by the screen refresh rate (60 Hz). Ten seconds into the trial, on-line feedback disappeared from the screen, and did not reappear until the start of the next trial. At the completion of each trial, participants received terminal relative motion feedback regarding the just-performed no-vision segment of the trial in the form of a Lissajous trace overlaying the criterion (circle) pattern and the orthogonal plot produced during the first 10 s of the trial. Participants received no experimenter-delivered augmented feedback at any time.  15 For the first ten trials of each session, participants performed only the circle pattern. Following these trials, participants received auditory probes on approximately one-third of the remaining trials. If the stimulus for the secondary task occurred during a trial, participants were instructed to respond by saying the word "pop" as quickly as possible. The word "pop" was chosen because the voiced consonant at the beginning of the response helps trigger the microphone. Participants were informed that there was the potential of a probe to occur in the vision or no-vision segment of the trial, but only 1 probe could occur per trial. Instructions to participants also stressed that in dual-task conditions primary task performance was of greater importance than RT to the probe and should be afforded greater priority of attention. Secondary probes occurred at each of the four probe positions 9 times, 7 of which occurred during the vision segment of the trial, while the remaining 2 occurred during the no-vision segment. Probes were presented in a randomised fashion for each participant across acquisition sessions. Prior to, and on the final day of testing, each participant performed ten control trials during which reaction times (RT) to the auditory tone were collected in the absence of the primary task and served as a baseline measure of RT. Baseline RT was collected for later comparison to the probe-RT measures collected during the experimental trials  2  Individuals participated in ten sessions of acquisition across a two-week period. For each of the ten practice sessions, participants performed a total of 100 acquisition trials, subdivided into 10 blocks of 10 trials. Each participant determined the interval between blocks, in order to offset any potential effects of fatigue across the 100 trials. Immediate No-Feedback Transfer Tests  3  At the end of each practice session (with the exception of sessions 3, 6, and 9), participants were asked to reproduce the 90° RP pattern of coordination without the use of continuous on-line visual feedback. To provide participants the opportunity to appropriately position the cursor on the computer screen, vision was briefly provided at the start of the trial (1.5 s), whereupon it was immediately  16  removed. Participants performed 24 no-feedback trials. Secondary probes occurred two times at each of the four probe positions (i.e., on one-third of the total trials administered). Probes were presented in a randomised fashion for each participant across no-feedback tests. Neither terminal nor experimenter-delivered augmented feedback was presented to the participant at any time. Delayed Retention and No-Feedback Transfer Tests The 24 no-feedback trials were repeated following a one-week retention interval of no practice. Following transfer trials, participants performed an additional 24 trials of the 90° out-of-phase task, but were given the same schedule of relative motion feedback as they had previously received while learning the 90° RP pattern. Data Analyses Primary Task Performance: 90° Relative Phase (RP) To measure bimanual coordination performance, measures of relative phase (RP) were calculated. Relative phase is commonly used because it is a quantifiable evaluation of how coordination performance changes during practice. By calculating the spatiotemporal difference between the limbs at any point in time, a kinematic measure was obtained indicating the participant's success in achieving the goal pattern of 90° RP (Tsutsui, Lee, & Hodges, 1998). In the present investigation, participants were required to continuously move their limbs 90° out-of-phase to make circular patterns. Therefore, continuous measures of RP were calculated from the angular displacement data using methods described by Scholz and Kelso (1989) . Specifically, displacement data were filtered 4  at 10 Hz for each limb and plotted as cosine functions. Peak displacements within a cycle were designated 0°, 180°, and 360° values and the intermediate points of 90° and 270° were determined at points of maximum slope displacement as a function of time (i.e., peak velocity). These points were normalized for displacement and velocity within the range of 1 to - 1 . The angle formed when velocity and displacement were plotted on a phase diagram gave a phase angle for each sample point (i.e., 1, 000 samples/second). This was calculated for both arms and the  17  difference between the limbs at any point in time gave a phase angle difference from -360° to +360°. Since participants could move their limbs with a right- or left-lead, RP values were normalized for comparison. That is, RP values obtained using a leftlead were wrapped to a range between 0° and 180°. From the relative phase values, the standard deviation around the mean relative phase (SD of RP) was computed to obtain an estimate of the variability in relative phase. Standard deviation of relative phase is a valuable measure for assessing acquisition of the to-be-learned pattern since it is an index of pattern stability. Variability serves to illustrate the exploration of new coordination patterns and the break from preferred coordination tendencies at the start of practice, as well as determine the stability and quality of the newly acquired pattern of coordination at the end of practice. Root mean square error (RMSE) relative to the 90° RP pattern of coordination was calculated as an overall measure of performance error. Specifically, RMSE was obtained from the square root of the sum of constant error squared and total number of observations. This measure has been commonly used by others (e.g., Hodges & Franks, 2000; Tsutsui et al., 1998) because of its sensitivity to both bias and variability. This is important when considering the problems inherent in averaging relative phase values. Early in learning, for example, a participant may switch between 0° and 180° RP resulting in a mean RP value of 90°. Performance would then be incorrectly assessed as accurate with respect to the criterion pattern. However, as a result of the switch between patterns, within-trial variability would be high. This variance is taken into account when using RMSE and the error score is modified accordingly. Although RMSE is made up of both within-trial accuracy and within-trial standard deviation, SD of RP was also analyzed separately (as previously stated). High values of RMSE may be due to one of, or both, variability and accuracy. Therefore, variability is an important variable to examine on its own accord (Hodges, 2001). In addition to RMSE and SD of RP, the frequency of each cycle was quantified. Cycle time was defined as the time interval from peak flexion to peak flexion, while cycle frequency was defined as the inverse of cycle time (Frequency (in Hz) = 1/ cycle time (in seconds)). For every trial, cycle frequency was averaged  18  across each segment to provide the self-selected movement frequency of the criterion relative phase pattern. Secondary Task: Probe-RT Probe-RT (in ms) was measured as the interval between the onset of the auditory stimulus and the moment at which the microphone detected the participants' vocal response. To determine baseline RT, a mean reaction time was determined for pre- and post- acquisition control conditions and submitted to a oneway repeated measures analysis of variance. No significant difference was found between pre-acquisition (M = 392 ms, SD = 56 ms) and post-acquisition (M = 411 ms, SD = 59 ms) baseline RT (p. = .357). Therefore, mean RT for the secondary task control condition was calculated using both pre- and post-acquisition measures (M 5  = 402 ms, SD = 58 ms). Results and Discussion The results are presented according to three main questions, namely: (a) Did performance improve with practice? (b) Did performance deteriorate with the withdrawal of concurrent visual feedback? and (c) Did extended practice on a bimanual coordination task reduce and eventually eliminate dual-task interference? Question 1: Did Performance of the Required Pattern Improve with Practice? Changes in Pattern Accuracy and Stability across Acquisition. Initial analysis of the data examined the effect of practice on the acquisition of a cyclical bimanual coordination task requiring a phase offset of 90°. The 100 trials performed in each acquisition session were divided into 5 blocks of 20 trials. Dependent measures (i.e., RMSE; SD of RP) were submitted to a Day (9) x Segment (vision, no-vision) x Block (5) ANOVA, with repeated measures on all factors. In the current section, results will only be presented for day and block. Presentation and discussion of 90° RP performance with respect to segment will be examined in Question 2, which focuses on the effects of removing concurrent visual feedback on performance of the  19  criterion pattern. As such, individual participant data in this section will only focus on the vision segment of the trial. In terms of the analysis, participants displayed considerable problems in producing the required 90° RP pattern on Day 1. Besides the difficulties inherent in the acquisition of any complex bimanual coordination task, the self-regulatory nature of the experimental procedure was a contributing factor to the difficulties observed (i.e., no explicit instructions were given to participants regarding how to produce the 90° criterion pattern and participants were instructed to move at their own pace). Although participants were instructed to make full movement cycles, some individuals moved so slowly they did not complete one full movement cycle within a trial (see Figure 1.7). Due to the lack of data in the initial block of practice for these participants, Day 1 could not be included in the statistical analysis of the acquisition sessions. However, Day 1 data are presented in Figures 1.5 -1.8 for discussion of participants' individual data. All post-hoc comparisons were performed using the Tukey HSD procedure. The level adopted for achievement of statistical significance was _ < .05. An overall picture of the evolution of task performance is illustrated in Figure 1.3 across all participants. Root-mean-square error (RMSE) is presented in the top panel, while standard deviation of relative phase (SD of RP) is presented in the bottom panel. These data illustrate several features characteristic for learning a new bimanual coordination pattern (see Zanone & Kelso, 1997). As practice proceeded, there was an improvement in performance as reflected by a change in both pattern accuracy (RMSE) and stability (SD of RP). Practice substantially improved performance towards the criterion pattern (i.e., 90° RP), and variability of the pattern decreased. Most importantly, this improvement continued throughout the nine days of acquisition for both measures. This is confirmed by a main effect across days for RMSE, F(8, 40) = 4.87, _ = .0003, and SD of RP, F(8, 40) = 5.68, _ = .0001. Percent improvement scores are illustrated in Figure 1.4 for RMSE (upper panel) and SD of RP (bottom panel). When compared to Day 2, RMSE improved by 8% for Days 3 and 4, respectively . This improvement continued through Day 5 (14%) and Day 6 6  (19%), followed by a plateau. In the final days of acquisition, pattern accuracy improved by 24% and 25% (Days 9 and 10, respectively) in comparison to Day 2  Figure 1.3. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) across 9 days of acquisition (Error bars = SEM).  3  4  5  6  7  8  9  10  7  8  9  10  Day 14 i $  13-  2  3  4  5  6 Day  22  performance. Similar percent changes were evidenced for SD of RP. By Days 4 and 8, pattern stability improved by 8% and 19%, respectively, and reached a maximal level of improvement on Day 10 (28%). Performance of the 90° out-of-phase task showed a general increase in accuracy across blocks for RMSE, although this was not significant (p_ = .089). In comparison, a significant block effect was observed for SD of RP, F(4, 20) = 4.78, _ =.007. Analysis revealed that performance of the 90° RP task was significantly more stable in Blocks 3 (JM = 10.0°, SD = 4.6°) and 5 (M = 9.8°, SD =4.0°) when compared to Block 1 (M = 10.6°, SD = 4.1°). Changes in Cycle Frequency across Acquisition. Participants performed the 90° RP pattern of coordination at a self-selected frequency. To gauge preferred rates of movement for the criterion task after Day 1, cycle frequency was averaged across each segment for every trial. The 100 trials of each acquisition session were then divided into 5 blocks of 20 trials and submitted to a Day (9) x Segment (vision, no-vision) x Block (5) ANOVA, with repeated measures on all factors. Findings revealed no significant main effect for day (p. = .708) or block (p_ =.734). From Day 2 to the end of acquisition, mean preferred rate of cycle frequency was 0.6 Hz (range = 0.4 Hz to 0.9 Hz). Stated differently, participants performed approximately 8 cycles per trial on average. This was maintained across blocks. Only one participant increased cycle frequency from 0.4 Hz to 0.6 Hz with practice (see Question 2, Table 1.1).  Individual Performance Data. Individual performances on the 90° out-ofphase task are illustrated in Figures 1.5 to 1.8 for the vision segment of acquisition. In each of these figures, the data has been presented in the form of Lissajous figures. Some interesting insights into the acquisition process can be obtained through visual inspection of these figures. Attainment of a circular configuration occurred rather quickly for two participants (see Figure 1.5). On the very first practice trial, it is evident that Participant 2 (top panel) is attempting to break free of an initial bias to anti-phase. On trial 10, the first full movement cycle is circular in  23  TJ CD  _CD CO  O  0)  §> .8  CD CO c  o  c CO CL.  E  C CD  o E _ CD  o  "ro  "ca  CN  • CO  zi a.  c  i  E  o  o  XJ  CO > CO CD L _ CO CO CO CD c °  ro J2 o _  a.  £  Q.  E  CO CO CD JZ  o  CD CD ® CD CD CD m TO L_  Si ro  CO  •e X ro i a. >< 'o ±s '•F ro D_  c «  ^  co  o  o  o  5  E  CO  "co  "ca  o5 .52  CD I £f  CD ^  -£  £ Q-st 8 a _  n  c  XJ  +1  C  > o tt  ro ro x: roK  2  .2 £  «E  .52  •rz x  in  CD .S? JZ II c 52  ro  X ^  T  C  -  ro co £  Q  vg _co CD  ro  CL  £  co  9>  3  a  CT  O  CO  13 CO  CD  CO  O  -ft  x 5 o ro .9- co ^ o a,  CD fc CO CO  3  h-  CO CO c CD CO c CD  E "°  "° CD  CO  CO CD  CD CD .  O <D CO  CL CL  ro o  o  CD CO C 0  CN  >» c CO  C L " > ^ CO CD CO c ro c JZ  Q.  -a .52 C  CD CD JZ  t ; -S> o. co c CD  I—  .a o lO  O CM  CO ®  "ro  "ro  c  CO CD JZ CO  CD  Q.  a  CL o xO - O CO ro .. S c +1 o X o ^ o co co ro ZJ c > XyJ « i CD co CD o B ro c O 3 CD CD E c o ro CL +1 E ro CD CO a. CD c O g • = co£ ® a) -o "co 2> ro o J=- 33 x a) co „ CO ro -E  . . XJ CD CD JZ  ±> 5 CD  W  "5 ^ 5.=2r = ? * 1 o .9?  24  nature, but then the produced pattern drifts to a more elliptical form for the remainder of movement cycles. By the end of the first block of 20 practice trials, emergence of the circle pattern became more consistent. Remnants of the early influences of pre-existing coordination patterns were replaced by the criterion circle pattern whereby the occupation of elliptical configurations (or diagonals) were no longer evident when attempting to perform a phase offset of 90° with the limbs (see Swinnen et al., 1997). For participant 5, performance of a relatively consistent pattern near 90° relative phase occurs even more rapidly. Although small in amplitude, the participant coordinated the limbs to produce a pattern that resembled a circular configuration by the end of the first trial. By trial 10, performance of 90° RP rapidly improved, although the participant demonstrated a tendency to spontaneously shift towards anti-phase in the middle of the trial, followed by a return to a phase offset of 90° on the very next movement cycle. This is indicated by the occupation of an elliptical configuration (or diagonal) in the Lissajous trace. Maintenance of a relatively consistent 90° out-of-phase pattern was produced by the end of block 1. For several individuals the acquisition of the criterion pattern did not occur as rapidly when compared to the evolution of task performance for Participants 2 and 5. Participant 6, for example, was unable to maintain a pattern of performance near 90° RP until the middle of Day 1 (see Figure 1.6). Early in acquisition, performance was strongly biased toward an anti-phase pattern. Slowly, the participant was able to break free of the bias. This is indicated by an increase in pattern variability as the individual searched for the required phasing over a period of 50 trials. The end of the third block of trials demonstrated emergence of a pattern larger in amplitude and circular in nature. One factor that may have contributed to the slow rate of acquisition for several participants was that individuals were required to learn the 90° RP pattern at their own pace. Most experiments investigating the acquisition of bimanual coordination have required participants to move at a pre-determined movement frequency such as 1 Hz (e.g., Hodges & Lee, 1999). When permitted to move at a self-selected frequency, the tendency of the performer was to move very slowly. In  25 Figure 1.6. Performance during the vision segment for Participant 6 on selected acquisition trials, Day 1 (x-axis = right forearm displacement; y-axis = left forearm displacement).  26 this approach, participants often attempted to learn what each limb did by moving the limbs in an independent fashion, rather than the nature of the phasing (or new coupling) relationship between the two limbs. In contrast, it appears that requiring an individual to make, for example, one complete movement cycle per second early in learning, serves to increase the variability of the currently performed pattern. This encourages the learner to break away from pre-existing coordination tendencies and more readily explore possible solutions to the movement problem, which in turn, facilitates the rate of acquisition early in practice. Figure 1.7 demonstrates the former strategy employed by Participant 1. This individual did not coordinate the limbs to produce one complete movement cycle until the second block of trials. Although participants displayed different rates of acquisition at early stages of the learning process, all participants were able to produce a pattern of performance near 90° relative phase by the end of practice on the first day of acquisition. In essence, participants acquired a general idea of the 90° out-of-phase pattern on Day 1, and continued to improve the accuracy and the stability of the required pattern across the entire time course of practice. This improvement was facilitated by the provision of continuous on-line feedback, which presented individuals an opportunity to use information regarding the relative motions of the limbs during the performance to correct and optimize their movements. Figure 1.8 demonstrates the evolution of 90° RP performance across extended practice for a representative participant when relative motion feedback was available to guide motor performance. Question 2: Did Performance Deteriorate with the Withdrawal of Concurrent Visual Feedback? Vision versus No-Vision during Acquisition. In acquisition, concurrent on-line  visual feedback was removed ten seconds into the trial, and did not reappear until the start of the next trial. Performance of RMSE and SD of RP are illustrated in Figures 1.9 and 1.10 for the 8.5 s vision segment (left panel) and the 3 s no-vision segment (right panel) collapsed across all participants. Analysis of RMSE (see Question 1) revealed a significant main effect for segment, F(1, 5) = 6.73, _ =.049.  27  Figure 1.7. Performance during the vision segment for Participant 1 on selected acquisition trials, Day 1 (x-axis = right forearm displacement; y-axis = left forearm displacement).  CO CO XJ CO CO  2o  CO CO  .CO  c o 'io  'D  cr  oCO XJ CP  oCP CP CO  IO •4—  em  c "c o CP  icip ispl  cCO CO o _ XJ CO c  Q_ cCO o CP <+o c CP v£ E 0 1_  •4—'  CJ) „ CP II CO CO  c 'xCO o  'co  •> CP  x:  c  0  E CO "l_ o •4—*  rfo rma ce earr spl  np  XJ  c  CP Q_  -C  CO  _  —  II CO  qur -ax  CD  ii.  29 A lower RMSE was observed during the vision segment of the trial as compared to the no-vision segment on each day of acquisition, as revealed by further post-hoc analysis (Figure 1.9). Analysis of SD of RP revealed a significant interaction effect for day and segment, F(8, 40) = 2.94, _ =.011. Stability of the 90° RP pattern increased at a faster rate when provided with relative motion feedback as compared to the novision segment, when concurrent feedback was removed. Post-hoc analysis revealed that performance of the 90° RP task was significantly more stable in the vision segment as compared to the no-vision segment from Day 4 to the final day of acquisition; although, the 90° out-of-phase task reached its greatest stability on Day 10 for both segments (Figure 1.10). Findings also revealed a significant interaction for segment by block for both RMSE, F(4, 20) = 7.08, p. = .001, and SD of RP, F(4, 20) = 4.51, _ = .009. For vision, the 90° out-of-phase pattern was significantly more variable and performed with the greatest error in Block 1 when compared to all other bocks of acquisition, except Block 2. There was no significant difference between blocks for the no-vision segment of the trial. Finally, when comparing cycle frequency between conditions of vision and no-vision, no significant main effect for segment was observed (_ = .193). Mean cycle frequency was 0.6 Hz, which was maintained across segments from Day 2 to the end of acquisition. These findings suggest that performance differences between vision and no-vision were not a result of differences in mean cycle frequency. Table 1.1 displays mean cycle frequency during the vision and no-vision segment for each participant collapsed across blocks. These data suggest that visual feedback has a critical impact on the acquisition and performance of a 90° out-of-phase pattern of coordination. We predicted that withdrawal of concurrent visual feedback early in practice would cause a drift from the produced pattern to a more stable intrinsic pattern of coordination. However, with practice, it was hypothesized that the performer would gain the capability to maintain performance of 90° RP, even upon the withdrawal of concurrent visual feedback for the last 3 s of the trial. This prediction is supported in  >  CO  in  LO  CO  z  0'  0  d  d  >  CO  10  LO  CO  0  0  0  d  >  CO  10  LO  z  0  0  0  d  >  CO  10  LO  0  0  >  10  z  t3  a  co co  §  CO  CD CD CO  •S o o c  CD  •SO •5 CD Q.  CD  • s  CD CD  C  CD  CD  a  in  I  CD CO C .O  .55  S• o c •o c CD  c  co  .0 .55  S  CD CN  CD  c c CD CD  r-  2 CD  I C  CD  d  d  CO  CD  d  d  CD  CO  CD  d  d  d  d  10  LO  CD  CD  0  0  d  d  d  d  >  10  m  LO  CD  00  CD  0  0  d  d  d  d  >  CD  CO  CO  CD  z  O  0  d  d  d  d  >  CD  10  XT  CD  00  CD  O  0  d  d  d  d  >  CD  co  co  CO  z  O  d  d  d  >  CD  00  LO  d  d  0  O  0  d  d  d  h-  LO  10  CO  CO  10  z  O  0  d  d  d  d  >  h-  10  CO  LO  O  O  d  d  d  d  >  CD  LO  LO  co  CD  LO  z  O  0  d  d  d  d  >  CD  LO  O  d  d  d  d  d  >  CD  LO  LO  CO  co  LO  z  O  0  d  d  d  d  >  CD  LO  O  d  d  d  d  >  lO  CO  LO  CO  00  z  O  d  d  d  d  >  ID  CD  CD  CO  O  d  d  d  COf  LO  z  > 0  CD  a.  c o  CD CL  JD JD  CD C  >  >  CD  CO d  CD $ CO  c  CD Q.  'o 't:  CD CL  2  CD > CD CO CD CO 3  CD  .52 is CD  8 U-> CD CO CD  0  O  rf 0  LO  CO  d  0  CN  CO  0  LO d  d  0  d  T  —  LO  CD  CD  15 0  O  1 CD CO  c 3  CD  •o CD  co  o c  CO CD  c  CO CD  y  O  c  5 CD c  0  CT  'i_ •+-*  d  d  CO '3  o  O d  c o  CD :£=  CD  C  d  g 'co CO CD CO  J2  -*—<  co  c  O  CO  CD  pant  CD  O  CO d  CD  E E C>D •8 oE CD CD  CL  CD  >• 2CD  Q  E o  33  only two of the six participants. The rest of the participants routinely demonstrated a phase drift upon withdrawal of concurrent visual feedback even on the final day of the extended acquisition period. When examining anecdotal evidence, these participants disclosed that they were taken aback to see the extent to which they had drifted from the criterion pattern when viewing terminal feedback of the justperformed no-vision trial segment. The participants commented that their performance during the no-vision segment surprised them, especially as practice continued across days and when taking into consideration their efforts to maintain the required pattern for the last 3 s of the trial. For these participants, differences between vision and no-vision performance remained prominent until the end of acquisition. Figure 1.11 illustrates the drift effect across acquisition for Participant 2 in the form of Lissajous figures. During the vision segment, the participant produced a pattern resembling the criterion early in the learning process. This pattern was produced with considerable accuracy and consistency by Day 4, which persisted for the duration of acquisition. However, when visual feedback was withdrawn 10 s into the trial, the produced pattern typically shifted away from the criterion towards a more stable anti-phase bias. This phenomenon continued even after 1, 000 trials of practice. Rather than demonstrate a capability to maintain a phase offset of 90°, Participant 2 produced an elliptical configuration, which actually became more stable and accurate across practice trials. Within the motor learning literature, one issue that has arisen concerns the frequency with which feedback is presented to the individual in the learning environment. Although the regular provision of augmented information has been shown to facilitate performance during acquisition, research has also revealed that performance deteriorates in transfer conditions when feedback is no longer available to guide and correct the required movement (e.g., Salmoni, Schmidt, & Walter, 1984). In the present experiment, most participants performed substantially better in the vision segment as compared to the no-vision segment. Although participants received both conditions within each trial, participants' were exposed to relative motion feedback for a greater portion of time (total vision =10 s; total no-vision =3 s). Therefore, it is possible that participants exhibiting a performance decrement in the  CD  C  l i CO CO  0_ c  c  CO ||  11 11  c  to  0)  0  —  Csi  E  CD  => CD D) CO C CO o CO  CO  _ o . o  > i  Qo:  c  CO T-  o  co  o  l_l <D 8 *CO  o o IO  X> -K  __  I S CL  £  1-  o  CNI  S-  ._ XJ  r §  % "S E  S _ 0  Q.  BC  0£ "co E J C • - cn oj 0— g  -  i-  x  "  o  z  > .CO CO  =5 > ^  0 c 0 c 0 CO 0 0 E 0 LL o C  CO  _  Q.  o  >  E  — CO || CO _ " - c CO  CC  o to  o o  C  6  2  c  CO XJ _ CO CD  CO XJ  >  II  -  P  1 £  5>  CD ^ O C -£  co _  o  CL  ~ 2  o  C  ._2  co Q-  _ 0  CO  o  >  o ro  35 no-vision segment developed a dependency on the feedback information, which they used on-line to monitor performance. As a consequence, these participants had difficulty maintaining performance of the 90° out-of-phase task when visual feedback was removed and performance drifted away from the criterion pattern to more intrinsically stable modes of coordination. This observation also supports the recent proposals of Proteau and colleagues (e.g., Tremblay & Proteau, 1998) who suggest that there is an optimal source of afferent information for successful performance of a task. Early in learning, it is possible that these participants deemed relative motion feedback as the most efficient information source available and processed it to the detriment of other information sources such as proprioception. In contrast, participants who were able to maintain performance under no-vision conditions may have used the relative motion feedback to update and refine the processing of more intrinsic sources of afferent information, rather than place a reliance on vision to guide the on-going movement (see Swinnen et al., 1997). As such, they were capable of maintaining performance in the absence of the relative motion feedback.  Pattern Accuracy and Stability under Conditions of Transfer. Participants  were also asked to reproduce the 90° out-of-phase task without use of on-line visual feedback in immediate and delayed no-feedback transfer tests. On days when transfer tests were conducted comparisons were made between the transfer tests (i.e., immediate transfer = Days 1,2,4, 5, 7, 8, 10; delayed transfer = Day 11) and the last 24 trials of acquisition. The 24 trials performed in each no-feedback transfer test (immediate and delayed) were divided into 4 blocks of 6 trials. The last 24 trials of acquisition on each no-feedback test days were also divided into 4 blocks of 6 trials. For each dependent measure, final blocks of acquisition were then compared to transfer blocks in a Day (1, 2, 4, 5, 7, 8,10,11) x Condition (acquisition, transfer) x Segment (8.5 s, 3 s) x Block (4) ANOVA, with repeated measures on all factors. Analysis of variance revealed a significant three-way interaction between day, segment, and condition for RMSE, F(7, 35) = 2.30, p = .048. Although performance improved with practice for both conditions, changes in performance were greater for the no-feedback transfer condition when compared to acquisition  36  (top panel, Figure 1.12). Post-hoc analysis revealed that the acquisition performance was superior to performance under the transfer condition for all days, except Day 10 (Day 10: M. acquisition = 11.4°, SD = 5.3°; M transfer = 21.6°, SD = 9.3°). This level of pattern accuracy was retained after a one-week retention interval of no practice (Day 11: M acquisition = 13.0°, SD = 6.8°; M transfer = 20.7°, SD = 4.9°). When relative motion feedback was provided during the 8.5 s segment of acquisition, participants performed the 90° out-of-phase pattern with greater accuracy when compared to performance during the 8.5 s segment of no-feedback in the transfer condition. However, when visual feedback was withdrawn in acquisition, performance deteriorated and no significant difference was displayed between the final 3 s of acquisition and transfer performance. Analysis of SD of RP revealed that stability of the 90° RP pattern increased in both conditions (bottom panel, Figure 1.12) as evidenced by a main effect for Day, F(7,35) = 6.04, fi = .0001. Post-hoc analysis indicated that the SD of RP on Day 1 was significantly different than variability of the pattern observed on Days 4 through 11. There was also a main effect for condition, F(1, 5) = 16.89, p =.009. Similar to RMSE, these data illustrate that acquisition performance was superior when compared to performance under the transfer condition. Pattern stability was greatest when participants were provided with concurrent visual feedback during acquisition, but variability increased when vision was removed. Similar levels of variability were demonstrated for the final 3 s of acquisition performance and performance under nofeedback transfer conditions. Although all participants showed decreases in performance under nofeedback transfer conditions, there was a range in the severity and duration of this decrement. Acquisition and transfer performance on the 90° out-of-phase task are compared in Figures 1.13 and 1.14 for two participants in the form of Lissajous figures. Figure 1.13 represents the performance of a participant who demonstrated highly accurate and consistent performance on 90° RP at early stages of the learning process. This individual was also able to maintain the required phase offset when relative motion feedback was removed during the no-vision segment of the acquisition trials (left panel). When the individual was tested at early stages of  37  Figure 1.12. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for condition (acquisition, transfer) across days (Error bars = SEM).  Figure 1.13. Individual trial data displayed as Lissajous figures for Participant 5. The left panel illustrates select acquisition trials whereby the solid line represents the vision segment, and the dotted line represents the no-vision segment. The right panel illustrates select no-feedback transfer trials (x-axis = right forearm displacement, y-axis = left forearm displacement).  Trial #195  -20  -10  0  10  Trial #239  20  Trial #418  Trial #505  Trial #987  Trial #1016  39  learning in the no-feedback transfer condition (right panel), performance of 90° RP deviated towards anti-phase as evidenced by the leftward-leaning elliptical shape of the pattern. However, with practice, the participant was able to overcome this tendency and performance under the no-feedback transfer condition improved in accuracy and stability to resemble the participants' performance in acquisition by the end of the test condition on Day 10. In contrast, several participants were unable to overcome the influence of pre-existing coordination tendencies whenever visual feedback was unavailable to guide performance. This is illustrated in Figure 1.14 for a representative participant. By Day 10, the participant was able to produce the relative phasing of the required pattern with considerable accuracy and stability when concurrent visual feedback was available, but could not overcome the influence of anti-phase with its removal for the last 3 s of the trial (left panel). Moreover, severe performance decrements were observed when the participant was required to perform under transfer conditions. Without relative motion feedback, the individual was unable to perform an accurate and consistent 90° RP pattern. Clearly, these figures demonstrate that removal of vision did not have the same lasting effects on performance for all participants. Question 3: Did Extended Practice on a Bimanual Coordination Task Reduce and Eventually Eliminate Dual-Task Interference? Probe-RT during Acquisition. The primary aim of this investigation was to examine the impact of practice on the attention demands of performing a 90° relative phase pattern of coordination. As such, participants were instructed to allocate attentional priority to performance of the 90° coordination pattern even if performance on the secondary RT task decreased. Maintaining primary task performance in the dual (or loaded) condition is essential if changes in secondary task performance are to be used to assess attentional characteristics of the primary task. In the dual-task condition, deterioration of performance on the 90° out-of-phase task would indicate that participants were trading off attention from the coordination task to the probe task to improve probe-RT performance (see Abernethy, 1988).  40  Figure 1.14. Individual trial data displayed as Lissajous figures for Participant 2 . The left panel illustrates select acquisition trials whereby the solid line represents the vision segment, and the dotted line represents the no-vision segment. The right panel illustrates select no-feedback transfer trials (x-axis = right forearm displacement, y-axis = left forearm displacement).  Trial #200  •40  -30  -20  -10  0  10  Trial #239  20  30  40  -49  -30  Trial #420  -40  -30  -20  -te  0  10  Trial # 950  -20  -10  •  10  N  36 4*  20  3*  Trial #508  20  30  40  -40  -30  -20  -10  0  10  Trial #1020  4*  41  A preliminary analysis was conducted to verify that the performance of the 90° outof- phase task was not affected by introduction of the secondary probe. Figure 1.15 illustrates the relationship between RMSE and trial number under no-probe and probe conditions across ten days of acquisition for the vision segment of the trial in a representative participant. A no-probe trial refers to baseline trials where participants were instructed that no probe could occur (n = 100), and catch trials where participants were instructed that there was a potential for the probe to occur, but no probe was ever introduced (n = 540). A probe trial refers to the loaded condition (n = 360). This figure reveals that there was a curvilinear reduction in RMSE with increasing trial number. This figure also reveals that there was no systematic difference in RMSE between no-probe and probe conditions, indicating that the probe condition did not adversely affect performance. Figure 1.16 illustrates the relationship between RMSE and trial during the 3 s no-vision trial segment. Visual inspection also reveals that there was no systematic difference between probe and no-probe conditions. Similar results were found for all participants in this investigation. The participants' data were further analysed by calculating mean RMSE for both no-probe and probe conditions in each acquisition session, and submitted to a Trial Type (no-probe, probe) x Day (9) repeated measures ANOVA. No significant differences were found for trial type (p. = .904). Mean RMSE was 13.2° regardless of whether participants received or did not receive an auditory probe while performing a bimanual coordination task requiring a phase offset of 90°. On all probe trials, a second analysis was conducted to determine whether performance before a probe was any different than performance following a probe. When the auditory stimulus occurred during the 8.5 s segment of the trial, RMSE was computed for 1 cycle preceding the stimulus (pre-probe) and for 1 cycle immediately following the stimulus (post-probe). These data were then submitted to a Time (pre-probe, post-probe) x Day (9) repeated measures ANOVA. Analysis was constrained to the 8.5 s trial segment because of the significant main effect reported in Question 2 for the segment factor. Post-probe movement cycles commencing in the vision segment, but ending in the no-vision segment were also excluded from  42  Figure 1.15. Curvilinear (polynomial) relationship between RMSE and trial number under no-probe and probe conditions for the vision segment of a representative participant.  V  70  o ° — — — —  i  CD  g> 6 0 — 50 <D CO CO  sz  CL  No-Probe Trials Probe Trials Line of Identity for No-Probe Trials Line of Identity for Probe Trials  o o o  40  CD  >  !« 3 0 £  1  20  LU  10 H  00  or  0 0  200  400  600  Trial Number  800  1000  1200  43  Figure 1.16. Relationship between RMSE and trial number under no-probe and probe conditions for the no-vision trial segment of a representative participant.  100 <D CD CD TJ  CD C/> CO  JZ CL  No-Probe Trials Line of Identity for No-Probe Trials ° Probe Trials — — - Line of Identity for Probe Trials  90 80 70 60  CD  >  50  jco £  40  CO  o  30  o  20  LU OD  10  Ui M—  o  b o  0 0  200  400  600 Trial Number  800  1000  1200  44  the analysis. Results revealed no significant differences in RMSE prior to and following an auditory probe (_ = .490). These findings suggest that participants maintained performance on the 90° out-of-phase task and did not trade-off attention from the coordination task to improve secondary task probe-RT. Following preliminary analysis, the probe data were subjected to a Condition (probe-RT, baseline RT) x Day (9) x Segment (vision, no-vision) x Probe Position (4) analysis of variance (ANOVA), with repeated measures on all factors. The mean reaction time of pre- and post-test measures were used for all cells of the baseline condition in the ANOVA (see data analysis). Probe reaction times elicited during acquisition are illustrated in Figure 1.17 (top panel) for each trial segment. A significant three-way interaction emerged, F(8, 40) = 5.80, _ = .0001, for condition x day x segment. Probe-RT was significantly slower during the no-vision segment (M. = 628 ms, SD = 137 ms) than during the vision segment (M_ = 566 ms, SD = 82 ms). However, this difference decreased with practice. On Day 2, probe-RT during no-vision was 161 ms slower than during vision, but only 13 ms slower by Day 10. Findings also revealed that probe-RT was significantly faster when responding to the auditory probe alone than when responding while performing the 90° out-of phase task. Although Probe-RT decreased with practice, it did not reach baseline levels (M. = 402 ms, SD = 58 ms) even after ten days of acquisition (Day 10: M = 546 ms, SD = 67 ms). There was no significant main effect for probe position (p. = .169). These findings suggest that practice resulted in a general decrease in dual-task interference. However, probe-RT did not reach baseline RT even after 1, 000 acquisition trials. One explanation for the present findings is that the amount of practice was not sufficient to produce an automatic response (even though participants engaged in a substantial number of practice trials of the 90° RP task). Earlier analysis of our data (Question 1) shows that performance of the coordination pattern continued to improve throughout learning (refer back to Figure 1.8 for an example). This is an important finding because as long as learning takes place and motor performance continues to improve, Blischke (1998) predicts the persistence of dual-task interference. Although dual-task interference was reduced in the present  45  Figure 1.17. Mean probe-RT for acquisition (top panel) and transfer (bottom panel) plotted as a function of day and compared to baseline RT ( M = 4 0 2 ms) (Error bars = SEM).  46  investigation, it was not eliminated despite the extended and highly specific nature of practice.  Probe-RT during Transfer.  Probe-RT collected under no-feedback transfer  trials were submitted to a Condition (transfer probe-RT, baseline RT) x Day (1, 2, 4, 5, 7, 8, 10, 11) x Probe Position (4) ANOVA, with repeated measures on all factors. Probe reaction times elicited during transfer are displayed in Figure 1.17 (bottom panel). Results revealed a significant interaction effect for condition x day, F(7, 35) = 9.74, _ = .0000. Probe-RT improved across days (Day 1: M = 746 ms, SD = 188 ms vs. Day 11: M = 510 ms, SD = 56 ms), but never reached baseline RT (M = 402 ms, SD = 58 ms). Post-hoc analysis showed that baseline RT was significantly faster in comparison to probe-RT collected on each day of transfer. There was no main effect for probe position (_ = .921). These data indicate that dual-task interference continued to persist in transfer conditions, even after repeated exposure to performing the 90° out-of-phase task in the absence of continuous on-line visual feedback. General Discussion In the past, the probe-RT technique has been used extensively to investigate attention demands during the execution of simple-laboratory movement tasks (e.g., Ells, 1973; Posner & Keele, 1969) and, more recently, in real-world sporting tasks (e.g., Castiello & Umilta, 1988; Prezuhy & Etnier, 2001; Rose & Christina, 1990). The present line of investigation sought to explore the attention requirements demanded by a bimanual coordination task over 1, 000 practice trials. Although dual-task interference was reduced, we found that bimanual coordination tasks continue to demand some form of attention even after extended periods of taskspecific practice. Our data support recent views of automaticity, which hold that attention is still involved at high levels of performance (e.g., Logan, 1988) and that elimination of dual-task interference is unlikely achievable, even after prolonged durations of training (e.g., Ahissar et al., 2001). That is, dual-task interference may persist as long as learning takes place and motor performance continues to improve  47 (Blischke, 1998). Although improvements in performance are generally greatest at the early stages of learning, it is well documented that further practice continues to yield gains in performance (see Snoddy, 1926, power law of practice). For example, Crossman (1959) reported small performance improvements in the task of cigarrolling by workers who had made over 10 million cigars and possessed seven years of task experience. In the present investigation, performance of the 90° out-of-phase pattern improved throughout the duration of the experiment, although gains exhibited at the end of practice were smaller than the gains displayed at the beginning. Given even more practice trials, it is likely improvements in performance would have continued. Other investigators have recently reported that cognitive involvement is required to maintain stability of bimanual coordination tasks intrinsic to the system. For example, in-phase and anti-phase modes of coordination may be intentionally stabilized in a dual-task setting by giving attentional priority to the coordination task. Although improvement occurs at the behavioural level, the volitional stabilization of the pattern increases the coordinative activity of the central nervous system, leading to a trade-off between relative phase variability and RT (Temprado, Zanone, Monno, & Laurent, 1999) (see Experiment 4 in this dissertation). In related literature, experiments examining the coordination of walking have revealed that regulating gait also incurs a cost at the central level. Normal gait control of walking has been investigated because it is often considered to be an automatic process given the extensive level of practice it receives across the lifespan. Regardless of the relative ease with which we manage bipedal walking, evidence shows that gait control does not operate automatically but requires attention for movement control. For example, dual-task data have shown an increased attentional cost for walking when compared to standing and sitting (Bardy & Laurent, 1991; Lajoie, Teasdale, Bard, & Fleury, 1993), as well as secondary-task interference to gait control when a concurrent cognitive task is introduced (Evbersbach, Dimitrijevic, & Poewe, 1995). As such, it appears the performance of fundamental motor skills and tasks requiring bimanual coordination, as suggested by the present investigation, demand at least a minimal level of cognitive control for successful execution.  48 An important feature of the present investigation was the use of a selfdirected learning approach. Participants were provided no instructions from the experimenter on how to move their limbs to produce the criterion pattern, nor did the experimenter intervene during the learning process to interpret or direct the learner's performance. For example, the experimenter provided no potential strategies to the learner, nor aided the performer in the interpretation of concurrent and terminal feedback. In this manner, the individual was left to his or her own devices to find an optimal solution to the motor problem at hand. This procedure is different than the methods implemented in many learning studies examining the acquisition of bimanual coordination. In these investigations, the experimenter often provides instructions or demonstrations as to how to produce the criterion pattern, or aids in interpretation of any feedback provided the performer throughout the learning process (e.g., Hodges & Franks, 2002; Hodges & Lee, 1999). Although these methods may be important for answering certain questions about the learning process, they may also bias the performer or even constrain the search for optimal learning solutions specifically suited to the individual. For example, experimenterdriven instructions and demonstrations are often very prescriptive in nature, and do not consider the capabilities or movement tendencies each individual brings with him or her to the learning environment; yet, the differences between people in the strength of underlying coordination tendencies might necessitate the use of different instructional techniques for skill acquisition. In the present investigation, participants were left to their own devices to discover how to break away from existing coordination tendencies and incorporate a new phasing pattern into their movement repertoire. In lieu of specific instructions and movement demonstrations on how to produce the required pattern, on-line relative motion feedback and terminal feedback were used by participants to generate a variety of solutions to the problem in a trial and error method of learning. The provision of continuous on-line visual feedback has been shown to be a powerful source of information for the acquisition of new bimanual coordination patterns (Lee et al., 1995). Relative motion feedback facilitates skill acquisition by demonstrating the properties and relationships inherent within a task, which specify  49  how the system can be controlled. The present data suggest that continuous on-line visual feedback had a critical impact on performance. Some performers developed a control strategy that was dependent on their capability to use the continuous on-line visual feedback to guide performance. When visual feedback was removed, these participants exhibited difficulties in overcoming the attraction of preferred coordination modes and were unable to maintain stability of the new relative phasing pattern. This influence was evident even at late stages of learning. In the present experiment, it is possible that the extensive use of on-line visual feedback influenced the individual's capability to use information sources other than vision to make the same gains in performance under conditions of withdrawal. Although the provision of relative motion feedback is a powerful source of information and potentially important learning aid in the acquisition of coordinated movement, these findings identify potential pitfalls when it is provided for extensive periods of practice. Further work examining ways to overcome the influence of preferred coordination tendencies, as well as the consequences of various forms of feedback, is clearly warranted.  Chapter Footnotes 1. According to Abernethy (1993), there are a number of methodological issues that must be taken into consideration if a secondary-task paradigm is to provide valid measures of attentional cost (also see Abernethy, 1988 for a review). The task of interest in the present line of investigation was the acquisition of a 90° RP pattern, which became the primary task. Therefore, our first methodological concern involved selecting an appropriate secondary task (see Ogden, Levine, & Eisner, 1979 for a review). To assess the degree of 'free attentional space' available after the attentional needs of the primary task have been met, care was taken to remove any structural interference between the primary and secondary task. This was accomplished by selecting a secondary task that utilized different sensory and response modalities than the primary task. We selected a probe-RT technique whereby the participant responded to the auditory stimuli by initiating a  50  vocal response. Therefore, the primary and secondary tasks did not use common input or output processes. To further eliminate the potential for structural interference, participants were permitted to move at their own pace. This contrasts typical methodological protocol in that participants are commonly required to move in time with an auditory metronome that beeps at a particular frequency. A second rationale for selecting probe-RT as the secondary task was because it was a discrete measure. When performing a continuous secondary task, there is a potential for errors to accumulate. As such, it becomes difficult to equate attentional fluctuations with specific phases of the primary task. In the present line of investigation, we were interested in attention demands at points of critical limb reversal points, and therefore, required a discrete measure such as probe-RT. A second methodological concern relates to probe frequency. Salmoni, Sullivan, and Starkes (1976) have demonstrated that the shape of a curve relating RT to probe position in a movement is different depending on the probe frequency employed. For a simple positioning movement, a V-shaped curve (i.e., probe-RT elevated at movement initiation and termination) results when using a probe frequency of two-thirds (see Posner & Keele, 1969), while a negatively accelerating curve (i.e., probe-RT elevated at movement initiation) is produced when a probe occurs on every trial (see Ells, 1973). If the experimenter probes too frequently, participants may begin to anticipate the onset of a stimulus. As a consequence, RT to the probe can be considerably enhanced. In addition, the individual may begin to see the primary and secondary tasks as one. This is problematic in that measures of RT no longer give an accurate description of the underlying attention demands of the task of interest. In contrast, if the experimenter probes too infrequently, the individual may be taken by surprise and probe-RT is artificially elevated. Taking the issues presented by Salmoni et al. into consideration, we elected to use a probe frequency of approximately onethird.  51  2. Prior to, and at the end of the final day of testing, participants performed a bimanual coordination scan as discussed in Appendix A - Motor Learning and Dynamic Pattern Theory. To obtain a picture of each participant's landscape before and after the learning episodes, individuals were scanned in a continuous manner. These data are not presented here because methodological issues have arisen concerning the use of this procedure for scanning an individual's attractor landscape, which is further discussed in Appendix B (Using the Scanning Procedure for Mapping Dynamical Landscapes: Methodological Implications). 3. A retention test is a common procedure used in motor learning to assess how much of the skill has been retained following an interval of time after practice has ceased. Typically, researchers employed this type of test by requiring a person to perform in the absence of augmented feedback. Removal of augmented feedback was deemed acceptable because the practiced skill was performed during the test. In recent years, some individuals have argued that altering availability of augmented feedback changes the context in which the learner must perform the skill. Therefore, a test that requires an individual to perform without augmented feedback should be classified as a transfer test (not a retention test) because the learner performs the practiced skill in a context different from the learning context (see Magill, 2004, p. 200). In the present experiment, any test requiring the removal of augmented feedback was referred to as a transfer test. 4. F = tan" [(dX /dt)/X ] - tan [(dX /dt)/Xi], where F = relative phase between the 1  1  R  R  L  limbs, X is the position of the limb within a cycle re-scaled to the interval [-1,1], (dX/dt) refers to normalized instantaneous velocity, and R and L are the right and left limbs (adapted from Scholz & Kelso, 1989, p. 129). 5. Mean baseline RT across participants was 402 ms, which is elevated in comparison to typical voluntary reaction time responses exhibiting latencies as  52  low as 120 to 180 ms. However, vocal probe reaction times are found to be longer than manual probe reaction times. For example, Starkes (1987) showed that simple RT requiring a vocal response was over 300 ms, while simple manual probe RT was less than 250 ms. Furthermore, findings revealed that mean vocal RT was elevated in dual-task conditions when compared to simple vocal RT displaying values ranging from 531 ms to 679 ms for three different probe positions. More recently, Bourdin, Nougier, & Teasdale (1997) used a dual-task paradigm to examine the attentional demands of grasping movements in a rock climbing situation. The secondary task required responding vocally (say "top") as fast as possible to an auditory stimulus. Probe-RT for different levels of reaching and hold complexity ranged from 416 ms to 582 ms. The results of Starkes (1987) and Bourdin et al. (1997) suggest that vocal probe-RT values are comparable in the present experiment. 6. Percent change from Day 2 was calculated using the following formula: -1*(Day x - D a y 2)/Day 2 ) * 100.  53  CHAPTER II EXPERIMENT 2 Effects of Reducing Continuous On-line Visual Feedback on Attention Demands and Extended Learning of a New Bimanual Coordination Task Introduction Studies in interlimb coordination have shown that the majority of individuals demonstrate bi-stable dynamics prior to practice. Bi-stability is described as the tendency to move the limbs in one of two modes of coordination, called in-phase and anti-phase, which are defined by the relative timing of the two limbs. In-phase movements correspond to synchronized flexion and extension of the limbs; while, anti-phase movements are characterized by flexion and extension of the limbs in alternation (Kelso, 1984). In addition to these pre-existing behaviours, new patterns can emerge with relative stability and accuracy as a function of practice. For example, individuals can learn to coordinate their limbs with a phase offset of 90°, whereby one limb leads the other by one-quarter of a movement cycle in time (e.g., Fontaine, Lee, & Swinnen, 1997; Lee, Swinnen, & Verschueren, 1995; Wenderoth & Bock, 2001; Zanone & Kelso, 1992, 1997). For most individuals, the 90° out-ofphase task is not part of their pre-existing movement repertoire. Therefore, substantial practice is needed to obtain accurate and stable levels of performance. The addition of new phasing patterns into an existing movement repertoire changes the dynamic stability of bimanual coordination from a bi-stable regime to a tri- or multi-stable regime (see Fontaine et al., 1997; Smethurst & Carson, 2001; Zanone & Kelso, 1992 for discussion on dynamic stability). Most important to the acquisition of new motor patterns is the finding that preexisting coordination tendencies appear to disrupt the learning process. At early stages of learning, attempts to coordinate the limbs in a 90° offset usually results in performance of the anti-phase pattern (or less frequently) the in-phase pattern (e.g., Fontaine et al., 1997; Lee et al., 1995; Zanone & Kelso, 1997). For some individuals, these biases continue even after considerable practice (see Experiment 1). Therefore, a major difficulty of the learning process is inhibiting the strong tendency  54  to perform more stable, preferred patterns. This includes breaking away from prepractice behaviours, as well as resisting the influence of intrinsic patterns when task conditions change. Determining practice variables that will decrease the effects of pre-practice behaviours on the acquisition of new coordination skills has obvious importance from both a theoretical and practical perspective. To break away from pre-existing coordination biases, individuals may use a number of different information sources. One type of information that has demonstrated its effectiveness for acquiring the 90° out-of-phase task is the provision of continuous on-line visual feedback (e.g., Lee et al., 1995; Swinnen et al., 1997). This mode of augmented feedback can be presented to the learner in 2dimensional form on a computer monitor by orthogonally plotting displacements of the left and right limbs in real-time. When correctly moving the limbs in a phase offset of 90°, the resulting relative motion plot (also referred to as a Lissajous figure) corresponds to a circle. In contrast, in-phase motions produce straight, diagonal lines between the lower left and upper right corners of the monitor, while anti-phase movements produce straight lines extending from the lower right to the upper left corner. To specify the task goal during the learning process, a criterion template is displayed on the monitor, and the displacements of the right and left limbs are superimposed onto the template. In addition to the task goal, the relative motion plot provides learners with non-prescriptive feedback alerting them to the exact nature and extent of their movement. This information may then be used to guide subsequent performance attempts. Relative motion feedback is a valuable source of information because it demonstrates the properties and relationships inherent within the task, which specify how the system can be controlled. One issue that has recently surfaced is whether individuals become increasingly dependent on relative motion feedback to guide coordination movements as learning progresses. This concern arises from empirical evidence, which suggests that augmented feedback provides strong informational support when it is available, but degrades performance when it is removed or task conditions change (see Salmoni, Schmidt, & Walter, 1984). This negative learning effect is reported to be especially prevalent when augmented feedback is provided  55  concurrently with practice of the movement (Schmidt & Wulf, 1997; van der Linden, Caraugh, & Greene, 1993; Verschueren, Swinnen, Dom, & De Weerdt, 1997; Wihstein, Pohl, Cardinale, Green, Scholtz, & Waters, 1996). Rather than directing attention to the critical proprioceptive features of the task, the learner specifically focuses on the augmented feedback. As such, an internal representation of the movement is derived from feedback extrinsic to the system, rather than on important task-intrinsic information. Consequently, performance is disrupted when concurrent feedback is unavailable because the learner is unable to effectively process intrinsic feedback sources for execution of the movement task. In Experiment 1, we examined how practice affected the production and stability of the 90° out-of-phase task. To facilitate acquisition, participants received continuous on-line visual feedback for the first 10 s of each 13 s trial. This feedback was removed for the final 3 s of the trial, followed by terminal feedback displaying the just-performed no-vision segment in the form of a Lissajous trace. Findings revealed that relative motion feedback had a critical impact on performance. In the presence of feedback, individuals were successful in producing the required phasing pattern. However, when visual feedback was removed for the final 3 s of the trial, several participants exhibited a strong bias towards the more preferred motor pattern of anti-phase. Even after a considerable number of practice trials (i.e., 1, 000 trials), the inability to inhibit or overcome the influence of the anti-phase mode of coordination was evident. Effects of pre-existing coordination tendencies were also apparent when examining performance of 90° RP under a no-feedback transfer condition. Performance deteriorated as demonstrated by decreases in both accuracy and stability of the required phase relationship. It is possible that providing relative motion feedback for a significant portion of each and every trial may have prevented some individuals from processing critical proprioceptive features of the task. As a consequence, these individuals were unable to modify the influence of pre-practice behaviours and produce the required movement successfully when visual feedback was no longer available to guide performance. Therefore, the first objective of the present investigation was to further examine the role of relative motion feedback for learning a 90° RP bimanual coordination task. More specifically, we employed the  56  same experimental design as described in Experiment 1 with one exception. Instead of receiving relative motion feedback for 10 s of the trial, participants received concurrent visual feedback for only initial calibration to the computer monitor (1.5 s) and the final (3 s) segment of the trial. This means that relative motion feedback was removed for the middle (8.5 s) segment . The purpose of providing relative motion 1  feedback for only a small trial portion was to encourage attention to critical sensory sources intrinsic to the task, thereby facilitating performance of the 90° out-of-phase task when feedback became unavailable to guide movement execution. Limiting the availability of visual feedback, in comparison to Experiment 1, was expected to influence the learning process in several ways. First, it was predicted that participants would demonstrate a slower rate of acquisition. After initial calibration to the computer monitor, individuals were unable to check the accuracy or consistency of their movement trace against the criterion until the end of the trial. Most important, however, was that once relative motion feedback was available in real-time, the participant was only provided 3 s to make on-line corrections. Given that participants received no specific instructions regarding how to produce the required phasing pattern, information gleaned from provision of concurrent (and terminal) feedback were essential for "getting an idea of the movement" (see Gentile, 1972). Based on this rationale, it was also postulated that performance during the vision segment should be superior in comparison to the novision segment early in learning. However, differences between the two segments were expected to decrease as a function of practice. With respect to performance under conditions of transfer, we postulated that a performance advantage would emerge in immediate and delayed no-feedback transfer conditions, as compared to the results of Experiment 1. This prediction was based on evidence from the investigation of unimanual discrete tasks, which suggest that learning is specific to the sources of sensory feedback available during practice. Referred to as the specificity of learning hypothesis (e.g., Proteau, Marteniuk, Girouard, & Dugas, 1987), this perspective proposes that sensory information made available to the performer during learning is integrated into a sensorimotor memory representation of the task. As the individual increases the  57  amount of practice with the sources of information, a dependency transpires. A consequence of this reliance is that performance may deteriorate when the individual is required to perform in a context where the sensory information is withdrawn. In contrast to our former investigation, the conditions of practice more closely matched the conditions of transfer. Therefore, when compared to the results of Experiment 1, participants were expected to perform the 90° pattern with less error and greater stability when continuous on-line visual feedback was withdrawn as an information source to guide the movement. The second objective of the study was to continue our investigation into the attention demands of bimanual coordination. To assess the attentional costs of performing a 90° out-of-phase task we employed the same classic measure of discrete reaction time (RT) as described in Experiment 1. Considering the difficulties associated with learning a bimanual coordination task under conditions of limited instructions and feedback, it was hypothesized that practice would diminish attention demands of the 90° pattern, but elimination of dual-task interference was not expected in the present investigation. Method Participants Informed consent was received from six self-professed right-handed individuals from a university population (n = 4 females; n = 2 males). The mean age of participants was 26.3 years (SD = 2.7, range = 23-31 years). All individuals were naive to the purpose of the experiment and none had previous experience as a participant in a bimanual coordination investigation. Upon completion of the final session, all participants received remuneration. The experiment was carried out according to the ethical guidelines set by the University Behavioural Science Screening Committee for research involving human participants. Apparatus and Tasks The apparatus, as well as the primary and secondary tasks were the same as described in Experiment 1. The goal of the primary task was to learn to correctly  58  move the forearms in the horizontal plane to produce a 90° RP pattern of coordination. While performing the primary task, participants were required to emit a vocal response as quickly as possible if at any time a computer-generated auditory 'beep' occurred. The auditory tone was delivered at one of four possible probe positions: Probe A (Pr A), Probe B (Pr B), Probe C (Pr C), and Probe D (Pr D), which were also defined in the methods of Experiment 1. Procedure All procedures for the acquisition, immediate transfer, and delayed transfer phases of the present investigation were identical to Experiment 1 except for the amount of relative motion feedback provided during each acquisition session. Participants received visual feedback for the first 1.5 s of each trial, whereupon vision was removed and participants received no relative motion feedback for the next 8.5 s. At the 10 s mark, visual feedback was restored and remained on the screen for the final 3 s of the trial. At the completion of each trial, participants received terminal feedback of the just-performed no-vision (8.5 s) trial segment, which was superimposed over the criterion template and the orthogonal plot produced during the vision (3 s) segment of the trial. Each segment was plotted in different colours in the form of a Lissajous figure. Presenting vision at the beginning of the trial provided an opportunity for participants to calibrate the cursor to the computer screen and begin the 90° out-of-phase pattern. The experimenter provided no instructions regarding how to produce the required pattern prior to, or during the investigation. Data Analyses As outlined in Experiment 1, spatiotemporal differences between the limbs were determined using continuous measures of relative phase (RP). Performance of the 90° out-of-phase task was further quantified by calculating: root mean square error (RMSE) relative to the criterion goal, standard deviation around the mean relative phase (SD of RP), and cycle frequency. The secondary task probe technique (probe-RT) was used to determine attention demands of the 90° relative  59  phase task. All post-hoc comparisons were performed using the Tukey HSD procedure. The level adopted for achievement of statistical significance was p_ < .05. Results The results are presented according to four main questions, specifically: (1) Did performance of the required pattern improve with practice? (2) Did performance differences appear between segments? (3) Did a performance advantage emerge in no-feedback transfer tests for learners receiving limited relative motion feedback during acquisition (as compared to Experiment 1)? and finally, (4) Did extended practice on a bimanual coordination task reduce and eventually eliminate dual-task interference? Question 1: Did Performance of the Required Pattern Improve with Practice? Changes in Pattern Accuracy and Stability across Acquisition. A 9 x 2 x 5 (Day x Segment x Block) ANOVA, with repeated measures on all factors was administered to examine the impact of practice on the acquisition of the 90° out-ofphase coordination task for both RMSE and SD of RP. The segment variable referred to the first 8.5 s no-vision segment and the final 3 s vision segment. This section will focus on the results for day and block, while Question 2 will discuss 90° RP performance with respect to segment. Day 1 was not included in the analysis because several participants moved so slowly they did not complete one full movement cycle within a trial (as previously discussed in Experiment 1). Performance improved in accuracy as evidenced by main effects for day, F(8, 40) = 4.92, p. = .0002, and block, F(4, 20) = 10.20 , e = -0001. When compared to Day 2, RMSE improved by 9% on Day 3, 22% on Day 4, and 29% on Day 6. This trend continued as RMSE steadily improved from 39% on Day 7 to 49% on Day 10. Post-hoc analysis revealed that Block 5 (M = 21.1°, SD = 14.9°) was significantly more accurate than Blocks 1 (M = 24.7°, SD = 16.0°) and 2 (M = 23.5°, SD = 16.0°), while Blocks 3 (M = 22.5°, SD = 15.2°) and 4 (M = 22.2°, SD = 15.2°) were performed with greater accuracy when compared to Block 1. The performance  60  measure of RMSE and percent improvement scores are presented in the upper panel of Figures 2.1 and 2.2, respectively. The 90° out-of-phase pattern also became more stable with increasing practice. This observation is supported by significant main effects for day, F(8, 40) = 3.16, fi = .007, and block, F(4, 20) = 5.58, fi = .003. Large decreases in pattern variability occurred from Day 2 to Day 3, as stability improved by 25%. Improvements in SD of RP continued until the end of acquisition as evidenced by further increases in stability from 33% on Day 4 to 39% on Day 6, and finally to 51 % on the final day of acquisition, when compared to variability of the 90° out-of-phase task on Day 2. Performance of the 90° RP pattern was more stable in Blocks 3 (M = 16.6°, SD = 13.8°) and 5 (M = 15.7°, SD = 13.3°), in comparison to Block 1 (M = 18.6°, SD = 14.1°). Standard deviation of RP is presented in Figure 2.1 (lower panel), while percent improvement scores are illustrated in Figure 2.2 (lower panel). Changes in Cycle Frequency across Acquisition.  A 9 x 2 x 5 (Day x Segment  x Block) ANOVA, with repeated measures on all factors, revealed preferred rates of cycle frequency for the criterion task from Day 2 to Day 10 of acquisition. Results showed a significant main effect for Block, F(4, 20) = 7.77, fi = .001, only. Preferred rate of cycle frequency increased and was significantly faster in Blocks 3 (M = 0.726 Hz), 4 (M = 0.731 Hz) and 5 (M = 0.730), when compared to cycle frequency in Block 1 (M = 0.689 Hz). Mean preferred rate of cycle frequency ranged from 0.4 Hz to 1.2 Hz. On average, participants performed approximately 9 cycles per trial, which was maintained across the final nine days of practice.  Performance of the 90° out-of-phase task improved with practice.  These  findings demonstrate that accuracy and stability of the criterion task improved across the time course of learning. Given more practice trials, it is likely improvements in performance of the 90° pattern would have continued. Interestingly, cycle frequency did not increase as a function of practice. Rather, participants generally maintained their selected cycle frequency for the duration of the experiment.  61  Figure 2 . 1 . RMSE of goal relative phase (top panel) and SD of RP (bottom panel) across 9 days of acquisition (Error bars = SEM).  ^ $ fe  _ _0  40 38 36 34 32 30 28 26 24 22 H 20 18 H  § cn o "j  16 14 12 10  4? E  V $  0  0  -1  1  3  4  1  5  1  6  1  7  1  8  1  9  1  10  Day 40 i 38 -j "co" 36 $ 34 •o .E £ _  0  4—  O  Q  30 28 26 24 22 20 18 16 14 -I 12 10/  o  i  1  1  1  1  3  4  5  6  7  Day  1  8  1  1  9  10  63  Question 2: Did Performance Differences Appear between Segments? Changes in Pattern Accuracy and Stability across Acquisition. The 9 x 2 x 5 (Day x Segment x Block) ANOVA (see Question 1) revealed that performance during the no-vision segment was generally less accurate when compared to • performance under the vision condition (M = 25.6° vs. 19.8°; SD = 19.8° vs. 11.7°, respectively), but did not reach statistical significance (p. = .076). Accuracy of the bimanual coordination task continued to improve for both segments from Day 2 (M No-Vision = 37.2°; M_ Vision = 30.0°) to the final day of acquisition (Day 10: M NoVision = 16.2°; M Vision = 13.2°). Performance of RMSE is illustrated in Figure 2.3 for the 8.5 s no-vision segment (left panel) and the 3 s vision segment (right panel). Analysis of SD of RP showed a significant main effect for day, F(1, 5) = 7.35, p. = .042. The 90° pattern was more stable in the presence of on-line visual feedback (M = 14.9°, SD = 11.9°) in comparison to performance under the no-vision condition (M = 19.2°, SD = 15.4°). Results also revealed a significant interaction effect for day and segment, F(8, 40) = 2.46, p = .028. This interaction effect occurred because a larger gain in pattern stability occurred from Day 2 to Day 3 for the no-vision segment when compared to the vision segment (M improvement = 18.4° vs. 11.3°, respectively). Post-hoc analysis revealed that performance of the 90° RP task was not significantly different between segments after Day 2. The coordination pattern was performed with the lowest SD of RP on the final day of acquisition for both segments (Day 10: M No-Vision = 12.6°; M Vision = 10.0°). Standard deviation of RP is illustrated in Figure 2.4 for the 8.5 s no-vision segment (left panel) and the 3 s vision segment (right panel).  Performance in Relation to the Criterion Pattern. Although the preceding results showed that differences in performance between the two segments decreased as a function of practice, it does not reveal the type of pattern the participants produced. That is, did participants learn to coordinate the limbs in a phase offset of 90°? To determine performance in relation to the criterion pattern,  64  65  66  overall mean RP was examined. On Day 5, mean RP was 114° for the no-vision segment and 104° for the vision segment. By Day 10, this value was closer to the 90° RP criterion. The overall mean RP for the no-vision and vision segments were 96° and 94°, respectively. In general, participants learned to perform the 90° out-ofphase task within a bandwidth of ±10° of the required phasing relationship. However, learning progressions between participants varied widely. Individual Performance Data. With no instructions and a limited presentation of on-line visual feedback, a limb phasing of 90° was initially difficult for many participants to acquire. By the end of Day 1, only one participant was able to produce a phase offset of 90° under no-vision and vision conditions (Block 5, M RP = 90° and 91°, respectively), while 2 of the 6 participants demonstrated a pattern of performance within ±10° of the criterion task for both segments. In contrast, the remaining 3 participants demonstrated a strong bias toward the anti-phase pattern and were unable to produce a circular pattern after 100 trials in either segment of the trial. In fact, for one participant, performance within ±10° of goal relative phase only emerged during the vision segment after 600 trials, and was finally achieved in the no-vision segment after 800 trials. To demonstrate the differences observed in individual progressions of learning, sample data have been provided for two participants in Figures 2.5 and 2.6. In the top panel, mean RP data have been plotted in blocks of 20 trials across learning. Both the no-vision and vision segments have been plotted on the same graph. In the bottom panel, select trials have been presented in the form of a Lissajous figure to represent the phasing relationship between the limbs in 2-dimensional form. Figure 2.5 represents data from a participant who did not exhibit the same difficulty in learning the circle pattern as demonstrated by other participants. Performance within ±10° of the criterion was achieved on the first day of acquisition and maintained regardless of whether visual feedback was available or unavailable to guide performance. These data suggest that the individual was able to inhibit the attraction of intrinsic coordination tendencies and acquire a new phase relationship between the limbs. Providing 3 s of visual feedback, in addition to terminal feedback,  Figure 2.5. Mean RP for Participant 4 plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment (top panel). Performance is also presented in the form of Lissajous figures for select trials (bottom panel). The solid line represents the no-vision segment and the dotted line represents the vision segment (x-axis = right forearm displacement; y-axis = left forearm displacement). 180 170  J 1  160 : 150 \ in o 140 \ CJ 130 { L_ O) — 120 !  c  110 0) V) 100 5  ro 90 : JZ a.  a  >  80  *-  70 :  DC  60 :  0) re c  ro Q)  50 ]  - a — No-Vision Segment - • • - Vision Segment Goal Relative Phase  40 \ 30 ^ 20  j  10 \ 0  i  h- CM COW IO  CD CD CD iCOlCO  ooto  ICQ ICQ ICQ  3  T-lcM co!w[ir> T-lcMCO-tfLO T-cM cow-in T - CM COW LOT - C N co W to T-CM COW IO T - C M cow in C Q ICQ ICQ ICQ ICQ C Q ICQ ICQ ICQ ICQ  4  5  C Q ICQ ICQ ICQ ICQ  co  6  Day by Block  ICQ C Q ICQ  7  'CQ  co  ICQ ICQ ICQ ICQ  8  CQ  CQ CQICQ  9  CO CO  CO  CQ  10  CO  ICQ  68  was sufficient for learning the criterion task . 2  In contrast, Figure 2.6 represents data from a participant who showed a strong attraction towards anti-phase. This bias proved difficult for the individual to break-away from. A plateau was observed in performance from Day 2 to Day 5, whereby the phasing relationship between the limbs was maintained at approximately 160° (SD = 9.9°) and 130° (SD = 8.8°) for the no-vision and vision segments, respectively. Although large gains were made on Day 6, performance was again observed to plateau. On the final 4 days of learning, performance was maintained at approximately 112° (SD = 11.1°) during the no-vision segment, while the vision segment was sustained at approximately 102° (SD = 8.9°). Although performance differences between the no-vision and vision segments decreased, the participant was never able to coordinate the limbs within a bandwidth of ±10° of the criterion pattern in the absence of feedback . These data demonstrate the strong 3  influence of anti-phase coordination when learning a new coordination task, especially when attempting to perform a criterion task in the absence of feedback. However, one question that arises for all 3 participants exhibiting the latter trend in performance is whether 3 s of visual feedback was sufficient to facilitate a break from preferred coordination tendencies. Given that participants executed approximately 1-1V2 cycles in the vision segment, it may have been difficult to bring together enough information to optimize learning on subsequent attempts (For additional individual performance data see Figures 2.10 and 2.11 presented in Chapter Footnotes #4 and #5, respectively). Cycle Frequency. When comparing cycle frequency, no significant main effect for segment emerged (p. = .151). Mean cycle frequency for the no-vision segment was 0.706 Hz, while mean cycle frequency for the vision segment was 0.730 Hz. Individual data for mean cycle frequency is displayed in Table 2.1 for each segment collapsed across blocks.  69 Figure 2.6. Mean RP for Participant 3 plotted in blocks of 20 trials across nine days of acquisition for the no-vision segment and the vision segment (top panel). Performance is also presented in the form of Lissajous figures for select trials (bottom panel). The solid line represents the no-vision segment, the dotted represents the vision segment (x-axis = right forearm displacement; y-axis = left forearm displacement).  -s— No-Vision Segment Vision Segment ——— Goal Relative Phase  co  ICQ ICQ ICQ ICQ  2  cototo 3  r-csi cow- in ICQ'CQ  co  C M rtkrw T - C M co Win  CO C Q C O C O C Q ICQ ICQ ICQ ICQ  4  5  co  to 6  ICQ ICQ  Day by Block  CM  \mrt w  T - C M  0 0  Win 1- C N cokt \mr-  CM CO  Ut l O  to totoCQtoICQ toto totototocQtototoCQ  I C QC Q  ICQ  7  CQ  8  9  10  70  C> CO  co  §  >  T -  >  CO  >  CD  CD  •9o •£  CD Q .  CO  d  >  co d  CD  Q  LO  I  CD CO  C  o .CO  S  XJ  c  O  CM T-'  CN CN  CO  CO  LO  O)  d  d  d  CO  CO  "fr  O)  d  d  d  CD  CD  LO  d  d d d co r-- Tr o d d d in  co  d  d  lO  N  't CO  d  CN  V *  LO  CM  d  T-"  LO  T -  d -4-  9>  T-"  T-  1  d co d  d d r»- co d CO hd d d d co CO CD d d d d S  N  CO (D  d LO  d  d d  co oo co co d o d o  S o o CO o o  CO o CO o  (D CO o o ^ CO o o  0  JD CD  r-  g.  C CD  I  'o CD  CL  E  CD CL X  LU  CD c CD XJ CO  c o  CO CD  0  CO  'to CD  c CD  To CD CO CD  XJ CD XJ _ZJ  o c o c  2> £ CD JZ  — *- < CD +-» CD XJ CO  c  0  CO  2> CL  CD  2> CD XJ  0  c CD  CN  CD  CO oo  d  d d co d d  c  CO  d  d  O*  O)  d  >  >  d  d  CM  d  o>  d TT O) d d  LO  >  d  d o d o  CM co  d  O)  CO S  > z d  CN  d o  d o d o  CN  CD  .CD  d  > d z  c:  C  T-'  >  CO  1 &  T -  CN  CO  S•  Tf co d o d o LO  ^' d  > z d  >  CD  CD  1  cS cS c5  > CD CN CO hz d i-^ d d d  >  CD  .o  d  > z  c  C  d  T -  d  O O  C  d  CD  CD CO  CD CD  CD  CD  X?  •s  co eq -sr co  T ~  T-  co  CD  CN  d CD O  Z - Q  CO  CM CO ^J- LO CD  XJ CD J Z  to  0  71  Pattern Accuracy and Stability across Experiments. Separate analyses were conducted for the 8 . 5 s segment and the 3 s segment, whereby the acquisition data of each dependent measure were compared to Experiment 1 using a 2 x 9 x 5 (Group x Day x Block) ANOVA, with group as the between-subject factor. The group variable referred to Experiment 1 and Experiment 2 . Analysis of the 8 . 5 s segment showed that the group effect for both RMSE, F ( 1 , 1 0 ) = 7 . 0 3 , fi = . 0 2 4 , and SD of RP, F ( 1 , 1 0 ) = 7 . 9 0 , fi = . 0 1 8 , were significant. The 9 0 ° pattern was performed with greater accuracy and stability in the presence of concurrent on-line visual feedback (Experiment 1: M RMSE = 9 . 5 ° ; M SD of RP = 9 . 0 ° ) , as compared to performance without relative motion feedback (Experiment 2 : M RMSE = 2 5 . 9 ° ; M SD of RP = 1 9 . 2 ° ) . Analysis also revealed a significant group by day interaction for accuracy (RMSE, F ( 8 , 8 0 ) = 2 . 9 5 , fi = . 0 0 6 ) , as well as stability (SD of RP, F ( 8 , 8 0 ) = 2 . 2 4 , fi = . 0 3 3 ) . By Day 2 , participants in Experiment 1 already exhibited superior performance in comparison to the participants' performance in Experiment 2 . As such, the performance gains observed in Experiment 1 were smaller than the improvements observed for participants in Experiment 2 across days. A significant interaction effect for group by day was also demonstrated when analyzing the 3 s segment for RMSE, F ( 8 , 8 0 ) = 3 . 3 8 , fi = . 0 0 2 , and SD of RP, F ( 8 , 8 0 ) = 2 . 3 2 , fi = . 0 2 7 . Greater changes in performance were demonstrated across practice for participants in Experiment 2 , in comparison to gains observed in Experiment 1. There was no significant main group effect for RMSE (fi = . 5 2 3 ) or SD  of RP (fi = . 3 3 6 ) . Figure 2 . 7 represents RMSE of goal relative phase (top panel) and SD of RP (bottom panel) plotted separately for each trial segment (left panel = 8 . 5 s segment; right panel = 3 s segment) and compared across experiments. With respect to the present investigation, visual inspection of the graph shows that performance during the 8 . 5 s segment continued to improve across practice and eventually reached comparable levels with no-vision ( 3 s) performance in Experiment 1. Pattern accuracy and stability also improved with practice during the 3 s segment in Experiment 2 . However, performance never reached the level displayed in the 8 . 5 s  72  Figure 2 . 7 . RMSE of goal relative phase (top panel) and SD of RP (bottom panel) plotted separately for each trial segment and compared across experiments. The left panel represents the 8 . 5 s segment and the right panel represents the 3 s segment (Error bars = SEM). 45 i  5  73  segment in Experiment 1, even though participants in the present investigation received relative motion feedback for this portion of the trial. Performance of the 90° pattern generally improves in the presence of concurrent visual feedback. In the present investigation, wide ranges of performance were observed for the 90° out-of-phase task. For some individuals, performance was similar irrespective of whether continuous visual feedback was or was not available to guide performance. For other learners, large differences in performance were displayed between segments for the duration of the experiment. That is, performance of the 90° pattern was superior when on-line relative motion feedback was available to guide performance in comparison to conditions without visual feedback. However, even in the vision segment, these participants were often unable to readily overcome the influence of intrinsic coordination tendencies to produce the required pattern. As a result of the limited availability of visual feedback, performance in the vision (3 s) segment of the present investigation did not reach the level of performance displayed in the vision (8.5 s) segment of Experiment 1. In lieu of specific instructions regarding how to produce the to-be-learned pattern, these data demonstrate the important role of concurrent visual feedback for inhibiting the attraction of intrinsic coordination tendencies and for getting an idea of the required coordination pattern. Question 3: Did a Performance Advantage Emerge in No-feedback Transfer Tests for Learners Receiving Limited Relative Motion Feedback during Acguisition? No-feedback transfer tests were analyzed separately for RMSE and SD of RP using a Group (Experiment 1, Experiment 2) x Day (8) x Segment (8.5 s segment, 3 s segment) x Block (4) mixed design ANOVA, with group as the between-subject factor. Results revealed no significant difference between Experiments 1 and 2 for RMSE (E = .868) or SD of RP (p. = .408). However, accuracy and stability of the 90° pattern improved across practice, as evidenced by a significant main effect for day  74  (RMSE, F(7, 70) = 6.67, p = .0000; SD of RP, F(7, 70) = 6.14, p = 0000). A performance advantage did not emerge in no-feedback transfer tests for learners receiving reduced continuous visual feedback to guide performance. Figure 2.8 illustrates RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for Experiments 1 and 2. Question 4: Did Extended Practice on a Bimanual Coordination Task Reduce and Eventually Eliminate Dual-Task Interference? Probe-RT during Acquisition. The probe data were subjected to a Day (9) x Segment (2) x Task Condition (probe-RT, baseline RT) x Probe position (4) A N O V A , with repeated measures on all factors (following preliminary analysis of the data ). 6  Mean RT for the secondary task baseline condition was calculated from both preand post-acquisition measures (M = 423 ms; SD = 72 ms) and used for all cells of the baseline condition in the A N O V A . Baseline RT was computed from both measures because a one-way repeated measures ANOVA revealed no significant difference (p = .536) between pre-acquisition (M = 412 ms; SD = 70 ms) and postacquisition (M = 434 ms; SD = 73 ms) RT. Analysis revealed a significant two-way interaction for day x task condition, F(8, 40) = 7.19, p. = .000, and segment x task condition, F(1, 5) = 12.98, p = .016. Figure 2.9 (top panel) presents mean probe-RT for each segment across days compared to baseline RT. Findings revealed that probe-RT was significantly faster when responding to the auditory probe alone than when responding while performing the 90° out-of-phase task. Although mean probe-RT decreased from 666 ms (SD = 97 ms) on Day 2 to 557 ms (SD = 103 ms) on Day 10, it did not reach baseline levels. Post-hoc analysis of the segment by condition interaction also showed that probe-RT was significantly faster in the no-vision (8.5 s) segment versus probe-RT in the vision (3 s) segment of the trial. Mean probe-RT for both segments were significantly different than baseline RT on all days of acquisition. No significant main effect was found for probe position (p = .849).  75  Figure 2.8. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for immediate and delayed no-feedback transfer tests across days, Experiment 1 vs. Experiment 2 (Error bars = SEM).  » 45  -I  1  2  4  5  7  8  10  11  Day  Experiment 1  1  2  4  5  7  8  10  11  76  Figure 2.9. Mean probe-RT for acquisition (top panel) and transfer (bottom panel) plotted as a function of day and compared to baseline RT (M = 423 ms) (Error bars = SEM).  900  •  i  No-Vision Segment Vision Segment Baseline  800 «T 700 E  600 a> P  500 400 300  1  n  2  3  1  4  —\  -  5  6  7  1  8  1 —  9  10  Day 900  — —  800  Transfer Baseline  «T 700 E £ l-  600  j> 500  2 °-  400 300/ —i  1  1—  2  4  5  7  Day  8  i  r~  10  11  77  Probe-RT during Transfer. The data were submitted to a Condition (transfer probe-RT, baseline RT) x Day (8) x Probe (4) ANOVA, with repeated measures on all factors. Findings revealed a significant interaction effect for condition x day, F(7, 35) = 3.54, p = .006. Post-hoc analysis revealed that mean probe-RT was significantly slower than baseline measures on each day of transfer, although it improved from Day 1 through to Days 10 and 11 (M = 650, 564, and 549 ms, respectively). Figure 2.9 (bottom panel) displays mean probe-RT under transfer conditions compared to baseline RT. There was no main effect for probe position (p = .080). Dual-task interference was reduced, but not eliminated following extended practice of the 90° pattern. Probe-RT measures collected during both acquisition and transfer trials revealed that performance of the 90° pattern continued to demand attention after ten days of task-specific practice, and following a one-week retention interval of no practice. Therefore, dual-task interference was not eliminated under practice conditions of reduced visual feedback and explicit instructions. Results also showed that attention demands were greatest when performing the task in the presence of concurrent visual feedback in comparison to attention requirements in the absence of relative motion feedback. Discussion Previous research has demonstrated the effectiveness of providing continuous on-line feedback for the learning of a new bimanual coordination task (e.g., Lee et al., 1995). Due to its usefulness, one issue that has emerged is whether individuals become increasingly dependent on this type of information as learning progresses. The purpose of the present investigation was to examine the effects of reducing the availability of continuous relative motion feedback for the learning of a 90° out-of-phase task. This was manipulated by reducing the amount of feedback from 8.5 s, as provided in the beginning of the trial in Experiment 1, to the final 3 s of the trial. Decreasing availability of continuous visual feedback should encourage the processing of sensory information sources intrinsic to the task,  78  thereby facilitating performance when feedback is unavailable to guide coordinative movements. Overall, results showed that participants exhibited a slower rate of acquisition when compared to the results of Experiment 1. A large difference in performance was already evident by the second day of practice, whereby performance in Experiment 1 was superior during the 8.5 s segment, irrespective of whether the 90° pattern was performed in the presence or absence of relative motion feedback in the present experiment. A substantial gain in performance suggests that participants in the Experiment 1 acquired a rudimentary idea of the movement much earlier in the learning process, which is attributed to the greater availability of concurrent visual feedback. Subsequently, smaller gains in performance were evidenced from Days 2 to 10, which indicates that the learners were now more focused on refining the coordination pattern (e.g., increasing the consistency of each relative motion trace superimposed onto the criterion template). When examining individual data, results showed that getting an idea of the movement proved especially difficult for several of our participants under reduced conditions of visual feedback. Due to the self-regulatory nature of the learning environment, the performer required the information provided by real-time relative motion feedback, as well as terminal feedback, to formulate a rudimentary representation of the required task. In lieu of specific instructions addressing how to achieve the required pattern, visual feedback was essential for specifying how the system can be controlled. Therefore, an obvious contributing factor to the difficulties observed, was the self-selected cycle frequency of learners. With only one exception, all participants consistently moved at a frequency of less than 1 Hz. This is especially problematic because real-time relative motion feedback was only provided for 3 s, and in that time frame participants only completed 1-1 Vz movement cycles (on average). It is possible that individuals were not provided enough time to effectively utilize the relative motion feedback and learn the nature of the phasing relationship between the limbs. This ultimately influenced rate of acquisition, as well as the quality of the new coordination pattern. Given the importance of the 3 s segment, it is interesting that participants did not increase their rate of cycle  79  frequency in an attempt to obtain more exposure to the concurrent visual feedback. Instead, they maintained their preferred rate of frequency in both segments of the trial, as well as across practice. The negative effects of removing concurrent relative motion feedback for learning a new bimanual coordination pattern have been documented elsewhere in the literature. For example, Wishart, Lee, Cunningham, and Murdoch (2002) have reported that older adults were unable to learn the 90° out-of-phase task when feedback was reduced to terminal relative motion plots after every fifth trial. These findings emerged despite the fact that the older adults could verbally describe the requirements of the task (i.e., one limb leads the other by a % of a movement cycle) and a group of younger counterparts were successfully able to learn the task. Wishart et al. hypothesized that older individuals, in comparison to young, have more difficulty breaking away from, or inhibiting the influence of, pre-existing coordination tendencies. These predictions were based on the finding that the average pattern performed by the older adults was either a 20° pattern (i.e., a strong bias towards in-phase) or a 160° pattern (i.e., a strong bias towards anti-phase). Moreover, participants did not show improvements on any performance measure, even after three days of practice. Based on our findings, we would suggest that problems breaking away from the influence of pre-existing coordination tendencies also occurred for some of our participants. For example, after 3 days of practice, one participant continued to perform a 160° RP pattern, while another participant coordinated the limbs in a 140° RP pattern. In our study, we provided more feedback (i.e., a relative motion plot after every trial vs. every fifth trial; and concurrent visual feedback for 3 s of every trial vs. no concurrent feedback); yet, these participants continued to exhibit a strong coordination preference towards anti-phase. Therefore, sensitivity to the availability of continuous relative motion feedback may not be strictly age-related. An important question in the present investigation was to examine whether specificity of learning effects would emerge for continuous on-line visual feedback when learning a new bimanual coordination pattern. In recent years, Swinnen et al. (1997) were unable to provide evidence in support of the specificity of learning  80  hypothesis. They compared the performance of three groups on the acquisition of a 90° out-of-phase task. In addition to terminal feedback following every fifth trial (in the form of a Lissajous figure), one group received relative motion feedback during performance of the 90° pattern (enhanced vision), while a second group was only permitted normal vision of the limbs (normal vision). Finally, a third group was required to perform the pattern blindfolded (reduced vision). Transfer tests were administered under enhanced vision, normal vision, and blindfolded criterion test conditions. Findings revealed that the real-time feedback group was not only more successfulat the 90° out-of-phase task in acquisition, but also outperformed the other two groups under all criterion conditions. Swinnen et al. concluded that the enhanced vision group learned an action plan that generalized to various conditions. Therefore, concurrent relative motion information has beneficial effects for learning and does not make a learner vulnerable when it is no longer available to guide performance. In contrast to Swinnen et al. (1997), our findings provide some support for the specificity of learning hypothesis. When comparing the no-feedback transfer tests, analysis revealed no significant differences in pattern accuracy and stability between Experiments 1 and 2. The superior acquisition performance demonstrated by participants during the 8.5 s segment in Experiment 1 did not emerge when relative motion feedback was removed in transfer tests as Swinnen et al. have shown. Moreover, some interesting trends were also revealed between Experiments 1 and 2. Early in learning, participants in Experiment 1 generally performed the 90° pattern with less error in comparison to participants in Experiment 2. However, on the last four transfer tests, findings showed that the 90° pattern was performed with less error in Experiment 2, even after a one-week retention interval of no practice. This reversal effect is most likely a result of the large gains in performance that several participants made following the fifth day of practice in the present investigation. Thus, the benefits gained from performing the 90° pattern in the presence of concurrent visual feedback does not necessarily generalize to various performance contexts. Rather, reducing availability of relative motion feedback may have some  81  benefits for performance when the conditions of practice are similar to the conditions of transfer. These results also show that voluntary stabilization of the criterion bimanual coordination task incurred an attentional cost, as measured by probe-RT, even after extended periods of task-specific practice. Given the difficulty of learning the new bimanual task under conditions of reduced feedback, performance gains were generally displayed throughout the time course of practice. In fact, for some participants, large gains were not exhibited until later stages of the experiment (i.e., after 500 practice trials). Whenever learning takes place and motor performance continues to improve, interference between a primary and a secondary task is predicted to persist (Blischke, 1998). The findings also suggest that the provision of relative motion feedback incurred a greater attentional cost in comparison to the central costs associated with performing the 90° pattern in the absence of real-time feedback. These results are contrary to Experiment 1, whereby attention demands were greatest when relative motion feedback was unavailable to guide performance. This indicates that attentional demands increase when the learner is required to switch between feedback sources within a trial (e.g., from visual sources to proprioceptive sources, or vice-versa). However, when considering that the final segment of the trial was only 3 s in both experiments, it remains unclear whether this central cost is a temporary effect, or more permanent in nature. It is possible that adding or removing augmented feedback within a trial induces a momentary increase in attention demands. Once the individual adjusts to the new information source, central costs may subsequently decrease. The attention demands of visual and proprioceptive feedback sources are discussed in more detail in Experiment 5 of this thesis.  82  Chapter Footnotes 1  The no-vision segment in Experiment 1 was initially designed to only be 3 s in duration because we were mainly interested to see whether the individual could maintain the criterion pattern when visual feedback was removed. With practice, we predicted that individuals would be able to maintain the 90° pattern. Contrary to our prediction, several individuals displayed coordination preferences towards anti-phase in the absence of relative motion feedback, even after 1, 000 practice trials. One possible explanation for this finding suggests that removing augmented feedback may act as a perturbation, inducing a temporary disruption in coordination performance. Given that the last segment of the trial was only 3 s, and participants performed at an average frequency of 0.6 Hz, it is possible that we did not provide the performer an adequate opportunity to recover from a possible perturbation. However, for purposes of analysis we maintained the time duration of each segment for Experiments 2 and 3, and further addressed this design limitation in Experiment 5 of this thesis.  For Footnotes #1 and #2, refer to Appendix B - Using the Scanning Procedure for Mapping Dynamical Landscapes: Methodological Implications. 2  The ease at which Participant 4 acquired the 90° pattern may be explained by discussing the data collected for pre-acquisition scans of the coordination dynamics. The pre-scan data revealed a preferred tendency towards 0°, 90°, and 180° RP. This suggests that the individual may have already possessed a multi-stable regime prior to practice (see Appendix B, Figures B4-B5).  3  Post-scans revealed no development of an attractor at the 90° pattern (see Appendix B, Figures B6-B7).  83  Figure 2.10. Mean RP for Participant 2 (top panel) and Participant 5 (bottom panel) plotted in blocks of 20 trials across 9 days of acquisition for the novision segment and the vision segment. Performance within a bandwidth of ±10 of the criterion pattern did not occur until late stages of learning. 180 170 160 150 In 140 £ 130 O)  « 120 £ 110  « T to  100 90 80 70 > 60 re » 50 c 40 re « S 30 20 -\ 10 0 re £ Q. tt  No-Vision Segment Vision Segment Goal Relative Phase  M CO W LO h—I calmT-CMCOht|LOCQtoCO lea ICQ W- C toko ICQ Icolcofo co ICQ ICQ  •CMCOhtW)  htio'  |m!mimlm!m|m  5  2  T - CM CO  |  6  j  W ION-  LOCDO siC CC O W T-CM iCQ lO lQ O l"QIO  CM CO CQ CO C Q CO CQ ICQ ICQ  7  CoWlO  mtototom 10  9  Day by Block  180 170 160 150 1 ? 140 £ 130 H •S 120 £ 110 « 100 CA re 90 80 70 60 tc 50 c re tt S 40 30 20 10 0  .1 ra «  • No-Vision Segment Vision Segment •Goal Relative Phase  h-CMCO^-Or^  XICQCOCQCQ  mmmmmxi  2  T-klbokfloT-klkokj-fO*-klkoW-ko<-k|bk|-ko cotommcQ cammmco mmmmco 3  4  5  6  Day by Block  -kN^oU-Lo almlScoco< mmmlmm n m m m m 7  8  9  10  84  Figure 2.11. Mean RP for Participant 1 (top panel) and Participant 6 (bottom panel) plotted in blocks of 20 trials across 9 days of acquisition for the no-vision segment and the vision segment. Relatively accurate performance of the 90° pattern emerged early in learning.  — • — No-Vision Segment - © - Vision Segment Goal Relative Phase  •<-CM  CM co w in •"-CMcow-m mm co co !m co  cow-m T -  ICQ ICQ ICQ ICQ CO CM  lea  T - CM CO CO CO CO  cow m ICQ ICQ ICQ  5  |  t co w m cow- m -CM cow-in T-cMco^m com cameo coco com Immimmmico co com mm  6  LO T - CM  |  CM ICQ ICQ  7  10  Day by Block  $140 |130 H  •8120  9-  70 60 50 40 -i 30 20 T-CMCO  i— No-Vision Segment 1  - Vision Segment  — G o a l Relative Phase  w-in T -  cow m T - C M cow-in T-cMcow-m * - C M cow-in T-CMcow-m r-  CM ICQ ICQ CQ [CQ ICQ ICQ ICQ CQ ICQ ICQ ICQ ICQ  co mm  3  4  mmlmmm co ico  5  ICQ ICQ  6  Day by Block  ICQ  cow- inT - C M cow-in •*-ki cow-in  CM GO ICQ ICQ m ICQ  coco to Ico Ico  7  8  CQCQlcQmlcQ  9  mtommm  10  85  6 As previously discussed in Experiment 1, a preliminary analysis was conducted to verify that performance of the 90° pattern was maintained on trials where a probe occurred, in comparison to no-probe trials. RMSE was calculated and submitted to a Trial Type (no-probe, probe) x Day (9) repeated measures ANOVA. Findings revealed no significant difference for trial type (fi = .113). Mean RMSE was 24.8° and 25.6° for the no-probe and probe trials, respectively. We also conducted a second analysis to verify that performance of the 90° pattern was the same 1 cycle before and 1 cycle after the probe. Mean RMSE data was submitted to a Time (pre-probe, post-probe) x Day (9) repeated measures ANOVA for the no-vision segment only. Results also showed no significant difference (fi = .192) in RMSE prior to (M = 25.0°) and following an auditory probe (M = 27.3°). Taken together, these findings indicate that participants did not share attentional priority with the secondary task, but maintained attentional focus on the 90° out-of-phase task.  86  CHAPTER III EXPERIMENT 3 Effects of Manual Guidance on Attention Demands and Extended Learning of a New Bimanual Coordination Task Introduction In the applied setting, it is common to employ instructional techniques that provide a guiding function. These might include the use of verbal cues or knowledge of results (KR), as well as physical or manual guidance. The rationale behind use of any guidance technique is that it helps learners gain a better idea of the criterion movement, helps reduce the number of errors that the learner makes, and provides more security for the individual in learning environments that are potentially dangerous by the very nature of the tasks performed (e.g., gymnastics, weightlifting) (Schmidt & Lee, 1999; Tsutsui & Imanaka, 2003; Wulf, Shea, & Whitacre, 1998). Based on these ideas, instructors generally assume that guidance facilitates the acquisition of a new motor skill. In the present investigation, the effectiveness of manual guidance on acquiring a new bimanual coordination task was examined. Manual Guidance Manual guidance is described as a procedure used to help learners experience sensory information that is derived from correct movements of the goal task (Tsutsui & Imanaka, 2003). For example, an instructor can provide manual guidance by moving the learner's passive limb(s) through the desired movement. Although practitioners often advocate the use of manual guidance in the instructional setting, research investigating the effects of manual guidance on motor skill learning has been sparse. Moreover, the results of investigations that have examined the effectiveness of manual guidance techniques have generally shown that guidance provides only temporary effects on motor performance (e.g., Armstrong, 1970; Hagman, 1983; Winstein, Pohl, & Lewthwaite, 1994).  87  Reasons provided for the ineffectiveness of manual guidance generally relate to the fact that the limb is passive when it is moved through the criterion task (Schmidt & Lee, 1999). Performance benefits are limited because the learner receives different sensory motor information than what would be received if the person was actively performing the task. Importantly, when experiencing a correct movement in a passive way, the individual is unable to obtain critical error information about the task or generate his or her solutions to the movement problem by way of discovery or trial and error learning. Of the few experiments that have examined manual guidance, simple tasks have generally been employed, which involve relatively few degrees of freedom. For example, Armstrong (1970) asked participants to move a lever in a specified spatialtemporal pattern, while Hagman (1983) and Winstein et al. (1994) required individuals to perform positioning movements to a mechanical stop. More recently, Wulf, Shea and Whitacre (1998) have advocated an approach that examines guidance using coordination tasks involving large numbers of degrees of freedom. Examining the effects of physical guidance on the learning of a complex skisimulation task, Wulf et al. provided support for the use of physical guidance for learning complex coordination tasks. When learners were provided a pair of ski poles to use as a guidance tool, they were able to produce a pattern that they would not have otherwise been able to perform until more advanced stages in learning. Most importantly, efficiency of the movement pattern was maintained when the ski poles were removed in immediate and delayed retention tests. Wulf et al. concluded that the ski poles were advantageous to the learning process because it provided the learner with an idea about how the goal movement pattern should feel. In addition, the learner actively produced the required movement. Used in this manner, guidance tools may be highly effective for motor skill learning. In a more recent investigation, Tsutsui and Imanaka (2003) examined the effectiveness of manual guidance for learning a 90° out-of-phase pattern. Findings showed that participants receiving manual guidance only, did improve their performance, although this method was less effective than just physical practice. Interestingly, no performance advantage emerged for participants receiving both  88  manual guidance and physical practice. As such, Tsutsui and Imanaka concluded that the learning benefits of manual guidance were limited for several reasons. In their study, participants were passively guided through the bimanual coordination pattern by an expert performer. Once again, the learner was not engaged in active practice of the movement, nor did (s)he experience movement errors. For guidance to be effective in the learning of a bimanual coordination task, Tsutsui and Imnaka postulated that the learner needs to be engaged in the problem-solving experiences that arise from actively executing the movement. Overview of Experiment The most difficult component of learning a new bimanual coordination task is realizing how to break away from pre-practice behaviours, as well as resisting the influence of intrinsic patterns when task conditions change. For the practitioner or motor learning theorist, determining practice variables that will decrease the effects of pre-practice behaviours is important for the instructional setting. In Experiment 1, we found that participants were highly successful at producing a 90° out-of-phase pattern when concurrent visual feedback was provided to guide performance; however, they were less adept at modifying the influence of pre-practice behaviours when visual feedback was removed. Because participants received real-time relative-motion feedback for 65% of each and every trial, we concluded that participants became over-reliant on the visual feedback and were unable to process critical proprioceptive features of the task. As a consequence, these individuals performed poorly when visual feedback was no longer available to guide the coordination movement. To further examine the influence of continuous on-line visual feedback on learning a new coordination pattern, we conducted an additional experiment where availability of visual feedback was restricted to a total of 4.5 s (or 35%) of each trial. The purpose of reducing availability of relative motion feedback was to encourage attention to critical sources intrinsic to the task, thereby facilitating performance of a 90° out-of-phase pattern when visual feedback is unavailable. Results showed that reduced availability of on-line visual feedback had a critical impact on rate of acquisition and several of our participants experienced great  89  difficulty overcoming pre-existing behavioural tendencies to acquire the criterion pattern. Due to the self-regulatory nature of the learning environment, we suggested that learning was hindered in the early stages of acquisition because participants experienced difficulty formulating a general understanding or rudimentary representation of the required task. Therefore, the purpose of the present investigation was to examine the manual guidance technique for learning the 90° out-of-phase pattern. Participants engaged in 10 sessions of 100 practice trials across a two-week period (i.e., 1, 000 trials). Each trial was broken up into three segments as previously described in Experiment 2. Continuous on-line feedback was provided for only initial calibration to the computer monitor (1.5 s) and the final (3 s) segment of the trial. Participants received no feedback for the middle (8.5 s) segment. Following each trial, terminal feedback of the just-performed no-vision segment was provided. Unlike Experiment 2, however, participants also received manual guidance. This was accomplished via a set of servo torque motors attached to the manipulanda. Immediately following provision of terminal feedback, the servo motors became operational and moved the participant's limbs for a period of 12 s, which was subdivided into two time intervals. For the first 6 seconds, participants received a summary playback of their own limb movements from the just-performed trial. The motors then moved the participants' limbs in the required phase offset for the final 6 s. This manipulation provided the learners an opportunity to identify their own movement errors by comparing their performance against the goal movement. We postulated that moving the participant's limbs in the required pattern would help the learner get a general idea of the goal movement, as well as experience with error information, thereby facilitating rate of acquisition and quality of the required coordination pattern. Identical to Experiments 1 and 2, we also assessed the attentional costs of performing a 90° out-of-phase task. We postulated that performance would continue to improve across practice (as previously demonstrated in our previous two investigations); therefore, attention demands of the 90° pattern were predicted to decrease as a function of practice, but elimination of dual-task interference was not expected to emerge.  90  Method All methods were identical to the previous experiments with the following exceptions. Participants Informed consent was received from six self-professed right-handed individuals from a university population (n = 5 females; n = 1 male). The mean age of participants was 22.7 years (SD = 3.7, range = 19-29 years). All individuals were naive to the purpose of the experiment and none had previous experience as a participant in a bimanual coordination investigation. Upon completion of the final session, all participants received remuneration. The experiment was carried out according to the ethical guidelines set by the University Behavioural Science Screening Committee for research involving human participants. Apparatus As previously described, participants were seated at a table facing a colour monitor with their forearms strapped comfortably into two angular manipulanda. In the present experiment, however, each manipulandum was connected to a servo torque motor (Mavilor MT-600) controlled via a motion control card (Tech-80 model 5638) and servo amplifier (Infranor SMVEN 2410). Angular positions of the manipulanda were recorded using two optical encoders (Dynapar, E2025001303), one mounted on each servo motor. Both optical encoders were attached to an interface card (Advantech Quadrature, PCI-833) giving 10, 000 counts per revolution, or a resolution of 0.036 degrees. Procedure During acquisition, participants received the same schedule of visual feedback as described in Experiment 2. Visual feedback was provided for the first 1.5 s of each trial, whereupon vision was removed and participants received no relative motion feedback for the next 8.5 s. At the 10 s mark, visual feedback was restored and remained on the screen for the final 3 s of the trial. At the completion of  91  each trial, participants received terminal feedback of the just-performed no-vision segment (8.5 s) trial segment, which was superimposed over the criterion template and the orthogonal plot produced during the vision (3 s) segment of the trial. Each segment was plotted in different colours in the form of a Lissajous figure. During each 13 s trial, the participant solely drove movements of the limbs. The servo torque motors were inoperative, acting only as small passive inertial and frictional loads on each manipulandum during participant-driven movement. However, immediately following the provision of terminal feedback, the servo motors became operational, physically moving the participant's limbs for 12 s. This 12 s time interval was divided into two segments. First, participants' received a playback of their own limb movements from the just-performed trial. The last 3 cycles of recorded movement in the no-vision segment was selected by the computer running customized software. The average cycle speed for the participant's movement was calculated, then speeded or slowed to play back at 0.5 Hz for 6 s. Following playback of the participant's movements, the servo torque motors then moved the right and left limbs at the required phase offset of 90°. Playback of the criteria movement consisted of three cycles of movement at 0.5 Hz over the final 6 s of the time interval. Following explanation of the 12 s playback component of the experiment, participants were reminded that they did not have to move the limbs at the speed of the servo torque motors (i.e., 0.5 Hz); rather they could move at their own self-selected frequency. All procedures for immediate and delayed transfer tests were identical to Experiments 1 and 2. The servo torque motors were inoperative during all transfer phases. No external instructions were provided to the learner at any time during the experiment. Results In the same manner as Experiment 2, the results are presented according to four main questions. These are: (1) Did performance of the required pattern improve with practice? (2) Did performance differences appear between segments? (3) Did a performance advantage emerge in no-feedback transfer tests when compared to  92  Experiments 1? and finally, (4) Did extended practice on a bimanual coordination task reduce and eventually eliminate dual-task interference? Question 1: Did Performance of the Required Pattern Improve with Practice? Changes in Pattern Accuracy and Stability across Acquisition. The dependent measures of RMSE and SD of RP were submitted to a 9 x 2 x 5 (Day x Segment x Block) ANOVA, with repeated measures on all factors. As previously discussed, Day 1 was not included in the analysis because several participants did not complete one full movement cycle within the initial blocks of trials. Practice improved performance towards the criterion pattern (i.e., 90° RP) as confirmed by a main effect for day, F(8, 10) = 4.34, p_ = .0008. The evolution of task performance is illustrated in Figure 3.1 across days for RMSE (top panel). When compared to Day 2, RMSE improved by 9% and 12% on Days 3 and 4, respectively. A gradual rate of improvement continued through Day 5 (17%) and Day 6 (16%), followed by a maximal gain of 29% on Day 7. On the final three days of acquisition performance was 23% (Day 8), 26% (Day 9), and 27% (Day 10) more accurate as compared to Day 2. Percent improvement scores are illustrated in Figure 3.2 for RMSE. Analysis also revealed a significant main effect for block, F(4, 20) = 8.05, p_ = .004. Blocks 3, 4, and 5 (M = 15.5°, 15.3°, and 14.6°; SD = 5.4°, 4.9°, and 4.2°, respectively), were performed with significantly less error as compared to Block 1 ( M = 17.2°, SD = 5.7°). The 90° relative phase task became more stable with increasing practice, as evidenced by a main effect for Day, F(8, 40) = 9.77, p_ = .0000. Standard deviation of RP is presented in Figure 3.1 (bottom panel). Increases in pattern stability continued at a gradual rate from 8% on Day 3 to 15% on Day 5, and finally to 30% on Days 9 and 10, when compared to variability of the 90° RP pattern on Day 2. Percent improvement scores for SD of RP are illustrated in Figure 3.2. A significant block effect was also observed (SD of RP, F(4, 20) = 9.49, p. = .0001). Performance of the 90° out-of-phase task showed a general increase in stability across blocks, whereby  95  Blocks 2 (M = 10.1°, SD = 3.2°), 3 (M = 9.9°, SD = 3.0°), 4 (M = 9.7°, SD = 2.8°) and 5 (M = 9.3°, SD =2.4°) were significantly more stable than Block 1 (M_ = 10.9°, SD = 3.4°). Changes in Cycle Frequency across Acquisition. Cycle frequency was  submitted to a 9 x 2 x 5 (Day x Segment x Block) ANOVA, with repeated measures on all factors. Findings revealed no significant main effect for day (p. = .461) or block (rj = .567). Mean preferred rate of frequency was 0.5 Hz, and ranged from 0.40 to 0.70 Hz. On average, participants performed 6.5 movement cycles per trial. Individual Performance Data. Of special interest is the learning strategy the individual self-employed to learn the 90° pattern. In contrast to Experiments 1 and 2, visual inspection of the Lissajous figures revealed that the majority of participants attempted to learn the phasing requirements by first making a box on the computer screen. This was accomplished by decoupling the limbs and moving each limb independently (see Figure 3.3 for a schematic illustration of the following description). First, the individual made a horizontal line across the screen by moving the right limb from the 'IN' to 'OUT' marker (panel A). This was followed by displacement of the left limb (flexion to extension), which produced the vertical component of the box configuration (panel B). A full square was completed by returning the right limb from the 'OUT marker to the 'IN' marker (panel C), followed by the left limb (panel D). Once individuals could produce a square configuration, their focus shifted towards 'smoothing' or 'rounding' the corners of the box to create more of a circular shape. At this point, instead of momentarily stopping the limb at each reversal point, the individual attempted to keep the limbs moving continuously. Figure 3.4 demonstrates the evolution of 90° RP performance across the first 100 trials for a participant employing the 'square box' strategy for learning the criterion pattern. On the very first trial of acquisition, the individual coordinates the limbs in an anti-phase mode of coordination. However, this influence is only temporary as the individual immediately attempts to uncouple the limbs, as demonstrated in Trial #5 by the high variability and widening of the Lissajous trace.  96  Figure 3.3. Schematic description of the 'box strategy' employed by participants to learn the 90° pattern.  c , C P >> C CO D -. _ CL C D 8> E  CD o .£ ro I  o CO  «  O CO TJ CD E J= ro  \-  CD  16  O  o ~m h-  I  £  •4—» J_J  ro co CO  LO  C L -Xc CO  !2 nCO  TO  CO  «§ ra «= CO  CD CO  1 ° ro CO co -S ££ CO  _J  LO  >  0  =tfc  c o £ CO  T3 CD >^ CO  ro  co -t—•  r—  c © CO  -i"  0  i -  0  €1 C L >=-  ro  —  ^ 0 E ! - §0 c ro "O TJE0 CO  CD  —  C  O  co co _ro  2= +J-  CL  —  CO  C  ro 0 no ZJ  > C  ii  © ro CO  c  <D  — S g .0 -fr co it; CO CD  > 0 o C  co  o LO  ro  TO h-  98  By Trial #6, a square configuration emerged, which continued to increase in consistency and amplitude (Trial #15). By Trial #30, a configuration more circular in nature appeared. Improvements in accuracy and stability of the circle shape continued across practice. By the end of Day 1, the individual was able to produce a relatively accurate and consistent 90° pattern. Performance of the 90° pattern improved with practice. The criterion task  continued to improve in accuracy and stability over the final nine days of acquisition. These data also suggest that the use of servo torque motors influenced the strategies employed by participants to learn the new coordination pattern. Once again, cycle frequency did not increase from Days 2 to 10 of acquisition. Instead, participants generally maintained their self-selected frequency for the duration of the experiment, which happened to be the same frequency the servo torque motors moved at. Although mean cycle frequency in Experiments 1 and 2 were also slow (0.6 Hz and 0.7 Hz, respectively), individual data from the former investigations exhibited a greater range in selected movement frequency between participants in comparison to Experiment 3 (see Question 2, Table 3.1 of this chapter). Therefore, it is possible that rate of cycle frequency in the present investigation was influenced by the speed of the servo torque motors. Question 2: Did Performance Differences Appear between Segments? Changes in Pattern Accuracy and Stability across Acquisition. The dependent  measures of RMSE and SD of RP were separately analysed by way of a 9 x 2 x 5 (Day x Segment x Block) ANOVA, with repeated measures on all factors (refer back to Question 1). Results showed that performance was more accurate, F(1,5) = 31.88, p_ = .002, and more stable, F(1, 5) = 249.08, p. = .0000, under visual feedback conditions (i.e., the 3 s segment), in comparison to performance in the absence of relative motion feedback (i.e., the 8.5 s segment) (see Figures 3.5 and 3.6). Inspection of mean RP values revealed that all participants performed within a  99  co  (S99j6ep ui) aseud 8AijB|aj p as  '>  6  A I  101  bandwidth of ±10° of the criterion pattern by the end of the third day of acquisition in both trial segments. Cycle Frequency. No significant main effect occurred between segments for mean preferred cycling frequency (p = .616). Selected cycle frequency for the novision segment was 0.481 Hz, while mean movement frequency for the vision segment was 0.475 Hz (see Table 3.1). Pattern Accuracy and Stability across Experiments. Acquisition data for each segment was examined in a Group (Experiment 1, Experiment 2, Experiment 3) x Day (9) x Block (5) ANOVA, with group as the between-subject factor. Experiment 1 was referred to as the enhanced feedback group (enhanced FB) because participants received relative motion feedback for 8.5 s of the segment. In contrast, Experiment 2 was referred to as the reduced feedback group (reduced FB) as they received concurrent visual feedback for the last 3 s of the trial only. Finally, the present investigation was referred to as the manual feedback group (manual FB) because servo torque motors, in addition to 3 s of concurrent visual feedback, passively moved participants' limbs. Separate analyses were conducted for RMSE and SD of RP. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) are presented in Figure 3.7. Each trial segment is plotted separately, whereby the left panel represents the 8.5 s segment and the right panel represents the 3 s segment. Analysis of the 8.5 s segment revealed a significant difference between groups for both RMSE, F(2, 15) = 4.98, p. =.021, and SD of RP, F(2, 15) = 6.15, p = .011. The 90° pattern was performed with least error by the enhanced FB group (M = 9.4°, SD = 2.5°), followed by the manual FB group (M = 16.9°, SD = 5.0°) and reduced FB group (M = 25.9°, SD = 18.0°), respectively. With respect to SD of RP, the enhanced FB group and the manual FB group performed the required pattern with the least variability (M = 8.9° and 12.1°; SD = 2.4° and 2.5°, respectively) in comparison to the reduced FB group (M = 19.2°, SD = 15.0°). Findings also revealed a significant day  >  LO  > z  >  rt  rt;  rt  rt  d  d> d> d> ci  CD  LO  LO  LO  d  d  d  d d  rt  LO  CD  LO  rt  rt  d  LO  d d d > co LO r t z d d d  o -Q  CO CO  §  oo  CO  "O CD CO Q.  LO  LO  d  LO LO LO rt > CD d d d d d  rt  > CD LO LO LO r t z d d d d d CD  d  LO  LO  d d  LO  >  CO  "§  CO CD  CD >s CO  Q  LO  rr  d d  LO  d  > CD LO LO LO z d d d d  >  c:  d  CD  d  LO  LO  d d  LO  d  rt  o  CD  d  rt  d d  > LO LO LO LO r t z d d d d d  •fc  d  d  d  c  CO .Q.  rt  d d LO  >  8  d  LO  d  d LO  d rt  rt  O  d  rt  LO  O  d  LO  rt  d  d  E  > CD LO LO LO LO LO z d d d d d d  CD CO C  >  CD  Dl  .o  S  "O  CO  LO  CO  S• o c  CN  CD =3  Cr  -32 C CO  I  d  lO  lO  LO  d d o d o  rt  LO  LO  d d  LO  lO  rt  d  d  d  > LO LO LO LO LO LO z d d d o d o LO  CD  c •o c^ c:  LO  > LO LO IO LO LO LO z d d d d d d  .CO  c co c .o  LO  d  Tt  d  rt  LO  lO  rt  d d d d d > r t rr LO LO LO LO z d d d> d> d> d>  >  co r*- LO LO h d d cS d d  >  d  c co Q. 'o "tr CO  0_  LO  LO  LO  LO  d o  LO  d  o  CN  rt  LO  CO  Tt  d LO  o  CD  103  Figure 3 . 7 . RMSE of goal relative phase (top panel) and SD of RP (bottom panel) plotted separately for each trial segment and compared across experiments. The left panel represents the 8 . 5 s segment and the right panel represents the 3 s segment (Error bars = SEM). • Experiment 1 Experiment 2 • Experiment 3 35 -\  104  by group interaction for both pattern accuracy (RMSE, F(16, 240) = 5.25, p = .0000) and stability (SD of RP, F(16, 240) = 4.23, p = .0000). These interactions occurred because the reduced FB group made the greatest gains in performance across days. The provision of continuous relative motion feedback and manual guidance during the 8.5 s segment appeared to facilitate acquisition of the 90° pattern on the very first day of practice. By Day 2, participants in both the enhanced and manual FB groups had already made large gains in performance. Thus, further performance gains were smaller than the improvements observed for the reduced FB group across the final 9 days of practice. Analysis of the 3 s segment showed no significant differences between groups for both RMSE ( p = .461) and SD of RP ( p = .083). However, results did reveal a significant group by day interaction for both dependent measures (RMSE, F(8, 120) = 8.69, p = .0000; SD of RP, F(16, 120) = 2.06, p = .014). Once again the greatest gains in performance were exhibited by the reduced FB group in comparison to both the enhanced FB and manual FB groups. Performance of the 90° pattern is superior in the presence of concurrent  visual feedback. These data indicate that manual guidance had a beneficial effect on the learning of a new coordination pattern (in comparison to Experiment 2). However, accuracy and stability of the 90° out-of-phase task was generally superior when continuous visual feedback was available to guide performance in comparison to conditions without visual feedback. Question 3: Did a Performance Advantage Emerge in No-Feedback Transfer Tests when Compared to Experiment 1? Each dependent measure was submitted to an Experiment (FB Group: Enhanced, Manual) x Day (8) x Segment (2) x Block (5) mixed design ANOVA, with experiment as the between-subject factor. Figure 3.8 illustrates RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for Experiment 1 versus 3. Performance improved for both groups with practice, as indicated by a significant  105  Figure 3 . 8 . RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for immediate and delayed no-feedback transfer tests across days and experiment (Error bars= SEM).  106  day effect for RMSE, F(7, 70) = 4.09, p = .0008, and SD of RP, F(7, 70) = 5.73, p = .0000. Results also revealed a significant difference between groups for RMSE, F(1, 10) = 5.38, p = .043. Mean RMSE was significantly lower for the manual FB group (M = 17.9°, SD = 6.7°) when compared to the enhanced FB group (M_ = 27.1°. SD = 13.0°). In addition, performance of the 90° pattern was generally less variable for the manual FB group (JM = 8.1°, SD = 4.6°), as compared to the enhanced FB group (M = 12.1°, SD = 9.5°), but did not reach statistical significance (p = .061). These data demonstrate that a performance advantage emerged in the no-feedback transfer tests for learners receiving manual guidance in lieu of extended availability of continuous on-line feedback. Question 4: Did Extended Practice on a Bimanual Coordination Task Reduce and Eventually Eliminate Dual-Task Interference? Probe-RT during Acquisition. Following preliminary analysis , the probe data 2  were submitted to a Day (9) x Segment (no-vision, vision) x Task Condition (acquisition probe-RT, baseline RT) x Probe position (4) ANOVA, with repeated measures on all factors. A one-way repeated measures ANOVA revealed no significant difference between pre-acquisition (M = 426 ms; SD = 99 ms) and postacquisition (M = 400 ms; SD = 89 ms) measures of baseline RT (p = .221). Therefore, mean RT for the secondary task baseline condition was calculated from both measures (M = 411 ms; SD = 95 ms), and used for all cells of the baseline condition in the ANOVA. Figure 3.9 (top panel) presents mean probe-RT for each segment across days compared to baseline RT. Results revealed a significant two-way interaction for day x task condition, F (8, 40) = 7.19, p = .000, and segment x task condition, F(1, 5) = 2.70, p = .018. Probe-RT decreased from 674 ms (SD = 122 ms) on Day 2 to 591 ms (SD = 113 ms) on Day 10, but did not reach baseline levels. These observations were confirmed by significant main effects for day, F(8, 40) = 2.70, p = .018, and condition, F(1, 5) = 53.92, p = .0007. Post-hoc analysis showed that probe-RT on Day 10 was significantly faster than probe-RT on Days 2, 3 and 4. Further post-hoc  107  Figure 3.9. Mean probe-RT for acquisition (top panel) and transfer (bottom panel) plotted as a function of day and compared to baseline RT (M = 411 ms) (Error bars = SEM).  900  • No-Vision Segment — ^ — Vision Segment Baseline  800 «T  700  o  600  E  a> 500  2 °- 400  zoo/  ~i  3  1  4  1  5  1  1  6  1  7  8  p  1  9  10  Day 900  — —  800 H «T  700  £  m  600  i 2  500  E l-  Transfer Baseline  °- 400 300 / —i  1  1—  2  —i  4  1—  n  5  7 Day  r  8  n  10  r  11  108  analysis of the segment x task condition interaction demonstrated that probe-RT was significantly faster in the no-vision (8.5 s) segment (M = 618 ms, SD = 112 ms) as compared to the vision (3 s) segment (M = 655 ms, SD = 126 ms), and both segments were significantly different than baseline RT. No significant main effect was found for probe position (p = .292). Probe-RT during Transfer. The data were submitted to a Condition (transfer probe-RT, baseline RT) x Day (8) x Probe (4) ANOVA, with repeated measures on all factors. Figure 3.9 (bottom panel) displays mean probe-RT under transfer conditions compared to baseline RT. Findings revealed a significant main effect for condition, F(1, 5) = 46.49, p = .001, and day, F(7, 35) = 11.09, p = .0000, as well as a significant interaction effect for condition x day, F(7, 35) = 11.09, p = .0000. Mean probe-RT improved across transfer tests from 707 ms (SD = 104 ms) on Day 1 to 593 ms (SD = 114 ms) and 539 ms (SD = 122 ms) on Days 10 and 11, respectively. Despite these improvements, post-hoc analysis revealed that mean probe-RT was significantly slower than baseline RT on each day of transfer. There was no main effect for probe position (p = .446).  Performance of a 90° out-of-phase task remains attention demanding  following extended practice. Identical to our previous investigations, these findings demonstrated a central attentional cost when responding to a discrete secondary probe during the simultaneous performance of a 90° out-of-phase task. Reaction time to the probe decreased from Day 1 to Day 10, but never reached RT measures collected when responding to the auditory probe alone. Results also showed that the attention demands of the 90° pattern were greatest when visual feedback was available to guide performance, as compared to performance in the absence of online relative motion feedback.  109  Discussion The purpose of the present investigation was to examine the effects of manual guidance for learning a 90° out-of-phase bimanual coordination pattern. Results demonstrated that using servo torque motors to physically guide the learner's limbs provided some learning benefits for acquiring the new relative phase task. In comparison to Experiment 2, all participants were able to coordinate the limbs near a phase offset of 90° much earlier in the learning process. Given that the only difference between Experiments 2 and 3 was the addition of manual guidance, we suggest that the servo torque motors provided the learner with a general idea of the task requirements. This occurred by giving the learner an indication of how each limb should move in relation to the other to produce a phase offset of 90°. This is important when instructions regarding how to produce the criterion task are not provided. For some individuals, manual guidance may have enabled the learner to produce the required pattern, which otherwise he or she may not have been able to effectively produce until later stages of learning, or not at all. Although individuals were passively moved by the servo torque motors in the required phase offset, they also received motor playback of their own movements. This was advantageous in that learners were provided the opportunity to make direct comparisons between their movements and the criterion pattern. The participant not only experienced making the error, but also received sensory feedback regarding the error. The individual was then free to generate a movement solution on the subsequent trial. This problem-solving experience may have contributed to the effectiveness of manual guidance in learning a new coordination pattern (see Tsutsui & Imanaka, 2003). Visual inspection of the relative motion plots showed that the servo torque motors influenced the type of strategy the individual employed to learn the new pattern. Although the movement of each manipulandum was smoothed as much as possible, to give a natural sinusoidal movement profile, the resulting servo motor movements may have appeared "robotic", with a triangular movement profile. The main characteristic of a triangular movement profile is very high acceleration at movement reversals and low acceleration 'coasting' between transitions. High-  110  acceleration movement reversals resulted in the majority of participants using a triangular movement profile (i.e., the 'square box' strategy) to learn the phasing requirements of the task - a strategy that did not spontaneously emerge in any of our previous investigations. It remains unclear, however, whether movement of the servo torque motors prompted participants to make a concerted effort to decouple the limbs on their own accord, or whether they were merely trying to imitate the movements of the manipulandum. Nonetheless, the high-acceleration movement reversals by the torque the motors did influence the phase relation between the right and left forearms. The cycle frequency of the servo torque motors may also have influenced the participants selected movement frequency. Although participants have shown a tendency to move at a frequency of less than 1 Hz (Experiments 1 & 2), all participants in the present study moved at a frequency of approximately 0.5 Hz, the same frequency the servo torque motors moved at. This occurred despite initial instructions to individuals stressing that they could move at their own self-selected frequency. An obvious suggestion for future research would be to increase the frequency of the torque motors to examine the effects on learning a new bimanual coordination pattern . 1  Despite the benefits of manual guidance, performance in the no-vision segment did not reach the same levels of performance displayed in the vision segment. This finding has been demonstrated in all three of our experiments, which leads us to conclude that performance of the 90° task is more successful when concurrent visual feedback of the criterion pattern is available to direct performance. Under visual conditions, participants are provided information regarding the displacement of each limb as it moves in real-time. This facilitates the learning of a new coordination pattern because the relative motion plot identifies the task goal, and provides learners with non-prescriptive feedback alerting them to the properties and relationships inherent within the task. This specifies how the system can be controlled without specific instructions on how to achieve the required pattern. However, the biggest benefit of manual guidance may be with respect to transfer. Performance of the 90° pattern transferred most successfully to conditions of no-  Ill  feedback for participants in the present experiment, in comparison to Experiment 1. Therefore, these data suggest that supplementing the learning environment with proprioceptive information may benefit performance when concurrent visual feedback is no longer available to guide performance. In light of increasing evidence emphasizing the importance of proprioception for interlimb coordination (e.g., Baldissera, Cavallari, & Tesio, 1994), further investigation into the benefits of manual guidance as an instructional aid are clearly warranted. With respect to attention demands, the present investigation supported the findings of both Experiments 1 and 2. Performance of a 90° out-of-phase task incurred an attentional cost, even after extended periods of task-specific practice. Results also revealed that reaction time to a secondary probe was faster when performing the criterion task in the absence of real-time feedback, in comparison to reaction time collected when visual feedback was available to guide performance. Similar to our two previous investigations, these data support the idea that attention demands increase when the learner is required to switch between feedback sources within a trial. This occurs irrespective of whether the transition is from a proprioceptive feedback source to a visual feedback source, or vice-versa.  112  Chapter Footnotes 1  In pilot work, the servo torque motors were originally set at a frequency of 1 Hz. However, at this speed the manipulanda began to vibrate noticeably and the movements of the torque motors were extremely jerky. Therefore, we traded-off speed for increased smoothness of movement.  2 A preliminary analysis was conducted to verify that performance of the 90° pattern was maintained on trials where a probe occurred, in comparison to no-probe trials. RMSE was calculated and submitted to a Trial Type (noprobe, probe) x Day (9) repeated measures ANOVA. Findings revealed no significant difference for trial type (p. = .224). Mean RMSE was 15.8° for the no-probe trials and 14.8° for the probe trials. A Time (pre-probe, post-probe) x Day (9) repeated measures ANOVA was also administered to verify that performance of the 90° pattern was the same 1 cycle before and 1 cycle after the probe. Results showed no significant difference (p = .679) in RMSE prior to (M = 14.1°) and following an auditory probe (M = 14.0°). These data confirm that participants maintained attentional priority on the 90° out-ofphase task.  113  CHAPTER IV EXPERIMENT 4 Attention Demands and the Performance of Intrinsic and Learned Bimanual Coordination Patterns Introduction The primary purpose of the preceding line of investigations was to examine changes in attention demands during the learning of a new bimanual coordination task. More specifically, acquisition of the 90° pattern was investigated over the time course of extended learning episodes and under different conditions of continuous visual feedback. In each of the three experiments, we found that voluntary stabilization of a newly learned bimanual coordination task incurred an attentional cost even after extended periods of task-specific practice. In the present investigation, we asked participants to return to the laboratory to perform the bimanual coordination patterns of in-phase and anti-phase. The purpose of this experiment was to examine the attentional costs associated with maintaining performance of pre-existing coordination tendencies in comparison to the newly learned bimanual coordination task of 90° RP. The coordination states of in-phase and anti-phase were chosen because they are intrinsic patterns of the human motor system, and can be performed skilfully, without practice (Kelso, 1984) (refer to Appendix A - Motor Dynamics: Fundamental Concepts and Terminology). We were  primarily interested in examining whether dual-task interference is eliminated when performing coordination patterns thought to be highly automatic in nature (also see Lee, Wishart, & Murdoch, 2002).  Attention Demands of Preferred Coordination Tendencies According to the dynamical systems perspective, smooth coordinated behaviour emerges from a mutual coupling between the components of the system. At a behavioural level, this is captured by the dynamics of relative phase, which results from a nonlinear coupling between the homologous limbs that specifies the positioning and velocity of each limb relative to the other. To determine the motion of  114  each limb in space and time, it is further proposed that an informational linkage exists between the limbs, which serve to inform each limb of the contralateral limb's whereabouts (see Temprado, Zanone, Monno, & Laurent, 1999). This informational linkage is mediated by the central nervous system (CNS). Conceptually, the CNS contributes to the coordination dynamics in controlling (or linking) the limbs, so that required patterns may be assembled and stabilized (Schoner & Kelso, 1988). This informational linkage has been recently viewed as an 'energy expense' necessary for maintaining coordination at a given frequency. The energy involved in stabilizing the system (determined at the behavioural level by fluctuations of relative phase) may be viewed in terms of mechanic and metabolic energy, as well as some kind of 'mental energy' or 'effort' devoted by the CNS to perform a preferred pattern of coordination. Referred to as attentional load, research has recently focused on assessing the relationship between coordination dynamics and the attentional activity of the CNS - concepts originally considered to be irreconcilable (see Swinnen & Carson, 2002 for a historical perspective on interlimb coordination). It is postulated that attention measures provide an approximate index of the coordinative activity of the CNS. As such, the focus of much of this work is on understanding the central costs incurred by the CNS to stabilize coordination patterns against perturbing forces. As previously discussed (see Experiment 1), attentional load is typically assessed using a dual-task paradigm, whereby participants simultaneously perform a bimanual coordination task (i.e., in-phase or anti-phase) and a discrete RT task. Moreover, participants are instructed to maximize the stability of the required pattern by giving attentional priority to the coordination task, which serves to increase the amount of attentional resources allocated to its performance. As a result, it is hypothesized that coupling strength will be maintained or increased (as measured by the collective variable of relative phase), while available resources to the secondary task are depleted (as indicated by increases in RT). Using the dual-task paradigm, Temprado et al. (1999) provided empirical evidence showing that intrinsic coordination patterns can be intentionally stabilized, improving performance beyond the level of stability attained spontaneously (i.e.,  115  when no special attention was given to the coordination task). Moreover, this extra stabilization was shown to increase the coordinative activity of the CNS, leading to a performance trade-off between relative phase variability and RT. Findings also revealed that attentional cost was dependent on the respective stability of the pattern. Assembling and maintaining the required pattern at a given level of stability was less attention demanding the more stable the pattern. As such, Temprado et al. found a greater performance trade-off between relative phase variability and RT for anti-phase coordination than for in-phase coordination (also see Monno, Chardenon, Temprado, Zanone, & Laurent, 2000; Zanone, Monno, Temprado, & Laurent, 2001). A tentative origin for the high cost incurred in maintaining the anti-phase pattern can be explained by the recent work of Mayville, Bressler, Fuchs, and Kelso (1999) who examined activation of primary sensorimotor areas while executing preferred coordination tendencies. Results revealed that the execution of in-phase and antiphase patterns involved activation of parietal and frontal cortical areas. However, to sustain performance of anti-phase coordination, greater activation of the sensorimotor areas was demonstrated. A dual-task paradigm was employed in the present investigation to examine the attentional costs associated with performance of the preferred patterns of inphase and anti-phase. Participants engaged in one session of 160 performance trials, which was comprised of 80 trials for the in-phase pattern and 80 trials for the anti-phase pattern. During the performance of in-phase or anti-phase, participants were required to vocally respond to a randomly computer-generated auditory signal on one-third of trials. Participants were instructed to maximize performance on the required pattern by giving attentional priority to the coordination task. Stability of the coordination pattern was quantified by the collective variable relative phase, whereas RT to the secondary task was used as an index of the central processing activity needed to maintain a coordination pattern at a given level of stability. Both primary and secondary task performance for in-phase and anti-phase modes of coordination were then compared to performance of the 90° out-of-phase task collected on the final day of practice in our three previous investigations (i.e., Experiments 1, 2, 3). The primary purpose of this analysis was to compare central  116  costs associated with intentional stabilization of a newly learned coordination pattern to bimanual coordination patterns that are inherently stable. Since participants were asked to give attentional priority to the coordination task, it was hypothesized that RT should be slower in the dual-task condition, when compared to RT values obtained when performing the secondary task alone. Furthermore, the central costs associated with stabilizing each coordination tendency were predicted to be smaller for the in-phase pattern, compared to the anti-phase pattern because in-phase is demonstrated to be a preferred (more stable) mode of coordination. Therefore, in-phase coordination is less costly to the CNS when compared to anti-phase performance. In keeping with this rationale, inphase coordination is also predicted to be less costly to stabilize when compared to stabilization of the newly learned 90° pattern. Method Participants Ten participants were recruited from our three previous investigations: Experiment 1 (n = 4), Experiment 2(n = 3) and Experiment 3 (n = 3). All participants had formerly received 1, 000 practice trials where the goal relative phase was an offset of 90°; however, participants were exposed to different types of experience at the 90° task depending on which experiment they were recruited from. Individuals participating in Experiment 1 were provided concurrent information feedback of the relative motions of their limbs for 10 s of every 13 s practice trial, while participants from Experiments 2 and 3 received concurrent feedback for only 4.5 s of each 13 s trial. In addition, the limbs of each individual participating in Experiment 3 were also moved passively in the required relative phasing following each practice trial by servo torque motors. Although participants had substantial experience with the 90° RP task, they were naive to the purpose of the present experiment (n = 8 females; n = 2 males). The mean age of participants was 28.5 years (SD = 6.0; range = 24-44). All individuals signed informed consent forms prior to the experiment and upon completion received remuneration for participation. The experiment was carried out  117  according to the ethical guidelines set by the University Behavioural Sciences Screening Committee for research involving human participation at the University of British Columbia. Primary Task and Apparatus Using the same apparatus described in Experiment 1, the goal of the task was to flex and extend the limbs in either an in-phase or an anti-phase pattern of coordination in the horizontal plane. Displacement of the right limb resulted in horizontal movements on the screen, while displacement of the left limb resulted in vertical movements on the screen. Concurrent visual feedback was provided on the computer monitor for each participant by plotting the real-time displacement of the right limb on the ordinate against the displacement of the left-limb on the abscissa superimposed onto the template of the criterion task. This plot was referred to as a Lissajous figure. When the limbs were correctly moved in 0° of relative phase, a diagonal line between the lower left and upper right corners of the computer screen was produced (Figure 4.1, panel A). Conversely, a diagonal line between the lower right and upper left portions of the screen was displayed for 180° of relative phase (Figure 4.1, panel B). Secondary Task and Apparatus While performing the task, participants were required to emit a vocal response if at any time an auditory 'beep' sounded. The apparatus used to generate and present the auditory signal, as well as the microphone used to pick-up the participant's vocal response, was the same as previously described in Experiment 1. Procedure The present experiment consisted of one testing session in which participants performed a total of 80 in-phase trials and 80 anti-phase trials. All participants executed the 80 trials of each coordination mode in 2 sets of 40 trials, which was further subdivided into 4 blocks of 10 trials. Half of the participants performed in an  118  Figure 4 . 1 . Illustration of in-phase and anti-phase coordination. (Top panel) A visual presentation of the Lissajous templates for in-phase and antiphase patterns of coordination, respectively. (Middle panel) Performance of in-phase and anti-phase, respectively, plotted in the form of a Lissajous figure. (Bottom panel) A schematic diagram depicting limb movements for in-phase and anti-phase patterns of coordination, respectively.  119  120  in-phase set first, while the other half received an anti-phase set first. Subsequent presentation of the coordination sets alternated between in-phase and anti-phase. All trials were 13 s in duration. At the beginning of the session, participants were given an opportunity to refamiliarize themselves with the apparatus. Criterion templates representing in-phase and anti-phase patterns of coordination were then displayed on the screen, one after the other. Participants were informed that they could match the criterion template by moving the limbs in a particular phase relation. When the template for in-phase was displayed on the computer screen, participants were informed that they could achieve the pattern by coordinating the limbs so that they moved symmetrically in opposite directions between the two boundary markers. When the template representing the coordination state of anti-phase was displayed, participants were told that this pattern was accomplished by moving the limbs in alternating fashion (like a windshield wiper) in the same direction. Using the same instructions provided in the previous experiments, participants were told that they could move at their own pace, but the goal of the task was to produce continuous movement cycles within each trial that matched as closely as possible to the criterion template. One movement cycle was explained to consist of the moving distance between the boundary markers so that each hand returned to the same starting position. Participants were also informed that they would receive the same augmented feedback schedule previously provided to them while learning the 90° RP task. Specifically, four participants (Experiment 1) received concurrent visual feedback for the first ten seconds of the trial, and then no feedback for the final three seconds. This group of participants was referred to as the Vision group. In contrast, six participants (Experiments 2 and 3 ) received concurrent feedback for initial 1  calibration to the screen (1.5 s), as well as for the final 3 s segment. Visual feedback was removed for the middle (8.5 s) segment of the trial. This group was referred to as the No-Vision group. For all individuals, terminal feedback was provided immediately following the trial, which displayed the participants' performance during their respective no-vision segment against their performance in the vision segment.  121  This information was superimposed (in different colours) on the criterion template in the form of a Lissajous figure. If the stimulus for the secondary task occurred during the trial, participants were instructed to respond by saying the word 'pop' as .quickly as possible. Furthermore, participants were informed that there was the potential of only one probe to occur in a trial. Instructions also stressed that performance of the coordination pattern should be given maximal priority. For each coordination pattern, auditory probes occurred on 30 of the 80 trials. Twenty-two of the probes occurred at variable times during the first ten seconds of the trial, while 8 probes occurred in the last 3 seconds. Probes were presented randomly for each participant across the session. Prior to, and at the end of acquisition, each participant performed 15 control trials during which RT to the auditory tone were collected in the absence of the primary task. This measure served as the baseline measure of RT. Baseline RT were collected for later comparison to probe-RT gathered during experimental trials, as well as baseline RT collected from Experiments 1 -3.  Data Analyses Performance of Bimanual Coordination Patterns To assess the performance of the coordination task, continuous measures of relative phase (RP) were calculated (see Experiment 1). From the relative phase values, the standard deviation around the mean relative phase (SD of RP) was computed to obtain an estimate of the variability in relative phase. To provide an overall measure of performance error, root-mean-square error (RMSE) was calculated. Finally, cycle frequency was also computed to determine the average self-paced frequency of each movement cycle on every trial. Central Costs Reaction time (in ms) was measured as the interval between the onset of the auditory stimulus and the moment at which the microphone detected the  122  participants' vocal response. To determine baseline RT, a mean RT was determined for pre- and post- acquisition control conditions and submitted to a one-way repeated measures analysis of variance (ANOVA). No significant difference was found between pre-test (M= 440 ms, SD = 86 ms) and post-test (M = 480 ms, SD = 118 ms) baseline RT (p = .064) . Therefore, mean RT for the secondary task control 2  condition was calculated using both pre- and post-test measures and then compared to participants' baseline RT time from the respective investigations collected previously. A one-way repeated measures ANOVA revealed no significant difference in baseline RT across experiments (p = .226). Mean RT for the secondary task control condition was then calculated using all baseline probe-RT measures obtained across experiments for each participant (M = 442 ms; SD = 95 ms). Results Performance of Bimanual Coordination Patterns Pattern Accuracy and Stability. The primary measure of performance was the RMSE relative to the respective coordination goal (i.e., 0°, 180°), as well as SD of RP. The 80 trials performed in each coordination mode were divided into 4 blocks of 20 trials. These data were compared to performance on the 90° coordination pattern from the final session of acquisition from each of the participants' respective learning investigation (i.e., from Experiment 1, 2, or 3). The last 80 trials, divided into 4 blocks of 20 trials, were used from Day 10 of learning for comparison purposes. Dependent measures were submitted to a Group (vision, no-vision) x Pattern (0°, 180°, 90°) x Segment (8.5 s, 3 s) x Block (4) analysis of variance (ANOVA), with repeated measures on the last three factors. Figure 4.2 illustrates RMSE performance for the coordination patterns of inphase, anti-phase, and 90° RP across all participants for each trial segment. No significant main effect for pattern emerged from the data (p = .123). Mean RMSE for the in-phase pattern was 9.0° (SD = 2.0°), while mean RMSE for the anti-phase pattern and the 90° pattern were 11.0° (SD = 3.6°) and 11.8° (SD = 4.4°),  Figure 4.2. RMSE of goal relative phase for pattern (0°, 180°, and 90°) by trial segment (Error bars = SEM).  • 8.5 s Trial Segment H 3 s Trial Segment  13 oal  CO  987-  o> 6o LU  U)  S or  5432 100-deg RP  180-deg RP Required Pattern of Coordination  90-deg RP  124  respectively. Results also revealed no significant main effect for group (p = .441) or block (p = . 173). Analysis did reveal a significant group x segment interaction, F ( 1 , 8) = 31.08, P = .0001. The vision group displayed lower RMSE values in the 8.5 s segment (M = 8.8°; SD = 2.5°) versus the final 3 s segment (M = 11.3°; SD = 3.1°), while a smaller difference emerged between segments for the no-vision group (8.5 s segment: JM = 10.7°; SD = 3.8°; 3 s segment: M_= 11.6°; SD = 4.9°). Taken together, these findings indicate that performance of the 90° pattern is generally more accurate when receiving visual relative motion feedback of the limbs in contrast to performing the criterion task in the absence of a visual information source. Figure 4.3 illustrates SD of RP performance across all participants for each coordination pattern by trial segment. A main effect for pattern emerged, F(2, 16) = 4.68, p = .025, whereby in-phase coordination was performed with the greatest stability (M_ = 7.1°; SD = 1.4°), followed by 90° RP performance (M_ = 8.2°; SD = 2.3°) and anti-phase performance (M = 9.5°; SD = 3.1°). Post-hoc analysis revealed that in-phase and anti-phase were significantly different from each other. The analysis showed no main effect for group (p = .421) or block (p = .614). Findings also revealed a significant group by pattern by segment interaction, F (1, 8) = 10.63, p = .001. In the 8.5 s trial segment, the no-vision group displayed greater variability in performance of the bimanual coordination patterns (M = 9-6°; SD = 3.1°) when compared to performance during the final 3 s of the trial (M = 7.9°; SD = 2.8°) and performance of both segments for the vision group (8.5 s segment: M = 7.8°, SD = 1.9°; 3 s segment: M = 7.8°; SD = 2.0°). Performing a large segment of the trial in the absence of a visual information source increased the variability of coordination performance compared to pattern stability when visual feedback was available to monitor on-going movements.  Cycle Frequency. To determine the self-paced cycle frequency for each coordination pattern, an average frequency was determined (in Hz) for each trial. The 80 trials of each coordination mode were then divided into 4 blocks of 20 trials  125  Figure 4.3. Standard deviation of relative phase (SD of RP) for pattern (0°, 180°, and 90°) by trial segment (Error bars = SEM).  14 n • 8.5 s Trial Segment  13 -  • 3 s T r i a l Segment  es)  12 -  egi  »  •a  c  a.  11 10 9 -  0£  8 -  o c o  7 -  > « TJ TJ (0 TJ C rs *J  V)  6 5 4 3 2 1 0 0-deg RP  180-deg RP Required Pattern of C o o r d i n a t i o n  90-deg RP  126  and compared to average cycle frequency of the last 80 trials of the 90° pattern from Day 10 of acquisition for each participant (i.e., Experiment 1, 2, or 3). Cycle frequency for performing 90° RP was also divided into 4 blocks of 20 trials and submitted to a Group (vision, no-vision) x Pattern (0°, 180°, and 90°) x Segment (8.5 s, 3 s) x Block (4) analysis of variance, with repeated measures on the last three factors. There was a main effect for pattern, F(2, 12) = 11.55, p =.002, and block, F(3, 18) = 5.70, p_ = .006. Post-hoc analysis revealed that cycle frequency for 90° RP was significantly slower (M = 0.581 Hz) when compared to natural modes of coordination (0° RP = 0.758 Hz; 180° RP = .745 Hz). Analysis of the main effect for block revealed that block 1 (M = 0.649 Hz) was significantly different from blocks 3 (M = 0.731 Hz) and 4 (M = 0.729). Figure 4.4 further demonstrates that cycle frequency increased across blocks for both in-phase and anti-phase, but remained consistent for 90° RP. Although the block x pattern interaction approached significance (p. = .052), no other significant main effects or interactions were found. Probe-RT The probe data were subjected to a Group (vision, no-vision) x Pattern (0°, 180°, 90°) x Segment (8.5 s, 3 s) x Condition (probe-RT, baseline RT) ANOVA, with repeated measures on the last three factors. Probe-RT for the 3 coordination patterns is illustrated in Figure 4.5. The analysis showed no significant main effect for group (p = .119), pattern (p = .600), or segment (p = .160). However, there was a significant difference between the two conditions, F(1, 8) = 38.69, p = .000. Attentional costs were exhibited for all three bimanual coordination patterns in the dual-task condition (0° RP = 599 ms; 180° RP = 584 ms; 90° RP = 578 ms), compared to performing the secondary task alone (M = 442 ms).  127  Figure 4.4. Self-paced cycle frequency for pattern (0°, 180°, and 90°) by block (Error bars = SEM).  Block 1  Block 2  Block 3 BLOCK  Block 4  Figure 4.5. Mean probe-RT for 0°, 180°, and 90° of relative phase compared to baseline RT (M = 442 ms). An asterisk indicates statistical significance when baseline RT was compared with the corresponding bimanual coordination pattern (Error bars = SEM).  129 Discussion The purpose of the present investigation was to evaluate central attentional costs associated with intentional stabilization of a newly learned coordination pattern to bimanual coordination patterns that are naturally stable. This issue was examined by comparing probe-RT between conditions when the secondary task was presented alone versus conditions in which the task was presented in parallel with one of three required bimanual coordination patterns: in-phase, anti-phase, and 90° out-of-phase. In addition, primary task performance of both in-phase and anti-phase coordination modes were compared to performance of the 90° pattern. This was included to evaluate the level of pattern accuracy and stability achieved for the 90° pattern in relation to pre-existing coordination tendencies of the human motor system. With respect to primary task performance, findings revealed no significant difference between the newly acquired 90° pattern and the inherently stable attractor states of in-phase and anti-phase. Moreover, the lack of a significant difference between coordination patterns appears to be related to movement frequency. When performing each required pattern, participants were permitted to move at a selfselected (or preferred) rate of movement. Findings revealed that the self-paced nature of the task varied depending on whether the individual was performing a newly learned 90° pattern or a pre-existing coordination tendency. More specifically, the 90° pattern was performed at a mean frequency of 0.6 Hz, while in-phase and anti-phase coordination patterns were performed at a mean rate of 0.8 Hz. Clearly, participants adopted a slower speed of movement when performing a newly learned pattern in comparison to pre-existing coordination tendencies. However, an obvious question that arises from the present data is whether stability of the 90° pattern can be maintained in the face of a changing environment. Would our participants, for example, be able to maintain the newly learned pattern at higher cycling frequencies? Furthermore, how would performance at higher cycling frequencies compare to the performance of intrinsic coordination patterns? First, we predict that differences between in-phase and anti-phase would occur, whereby in-phase coordination would emerge as the more accurate and  130  stable mode of coordination. This conclusion is based on well-documented evidence, which shows that the anti-phase pattern loses stability and an unavoidable switch from anti-phase to in-phase takes place when movement frequency is progressively scaled beyond a certain critical region (Kelso, 1984). Second, stability of the 90° pattern is also expected to break down at higher movement frequencies. However, instead of a loss of stability to the in-phase pattern, it is expected that a transition to the attractor state of anti-phase would occur. This prediction is based on the evidence of Smethurst and Carson (2001) who have shown that there is a general tendency to switch from 90° relative phase to anti-phase when movement frequency is progressively scaled. The latter prediction is contrary to the theoretical predictions of Schoner and colleagues (Schoner, 1989; Schoner, Zanone, & Kelso, 1992) who postulated that learning a 90° pattern leads to de-stabilization of the anti-phase pattern because it is the weaker of the two initial attractors (see Zanone & Kelso, 1997). In the present investigation, the anti-phase pattern and the 90° pattern were performed with similar accuracy and stability. Therefore, learning a new coordination pattern in our three previous investigations did not 'annihilate' the anti-phase pattern. Instead, practice of the criterion pattern permitted the learner to break away from pre-existing coordination tendencies, and merely suppress the attraction of the anti-phase mode (also see Shadmehr & Holcomb, 1999). This implies that the anti-phase pattern is still available as an attractor state even when the 90° pattern is being performed. Fontaine, Lee, and Swinnen (1997) and Lee, Swinnen, and Verschueren (1995) have also provided evidence showing that naturally occurring coordination tendencies do not de-stabilize when a new coordination pattern is learned. In the present study, findings revealed that probe-RT for in-phase and antiphase were significantly slower than RT at the single-task level. This demonstrates that dual-task interference not only persists for newly learned patterns of coordination (Experiments 1-3), but also exists for coordination tendencies intrinsic to the system. Stated differently, the intentional stabilization of preferred coordination tendencies demands at least a minimal level of cognitive control. Despite the costs associated with increasing coupling strength between the limbs,  131  intentional stabilization allows for flexibility within the natural dynamics of the system. The effects of intentional processes have now been demonstrated in a number of investigations, which show that people are highly adept at deliberately switching between intrinsic patterns (e.g., Carson, Byblow, & Goodman, 1994; Carson, Goodman, Kelso, & Elliott, 1994; Carson, Byblow, Abernethy, & Summers, 1996; Scholz & Kelso, 1990), as well as in volitionally delaying or inhibiting spontaneous phase transitions from one intrinsic pattern to the other (e.g., Lee, Blandin, & Proteau, 1996). However, in the latter observation, Smethurst and Carson (2003) have recently suggested that the ability to delay the onset of a phase transition needs to learned. Within the literature, dual-task interference has also been investigated by assigning different attention priority requirements to the tasks at hand. For example, Temprado et al. (1999) instructed participants to share priority between the required bimanual coordination task and a discrete RT task (non-priority focus condition), as well as give priority to either the intended pattern or the RT task (priority focus conditions). As previously stated, findings showed that pattern variability decreased, while response to the auditory probe increased when assigning attentional priority to the coordination task. However, similar findings were demonstrated, although not to the same extent, when participants were instructed to share their attention (i.e., do your best in each task). According to Temprado et al., this demonstrates the important role of the CNS in spontaneous situations, where the level of stability exhibited by the coordination patterns is not specified in any manner. These results lend support to the notion that dual-task interference persists in tasks that are considered to be preferred tendencies of interlimb coordination. Investigation of the relationship between attention demands and the coordination of walking has also produced similar findings with respect to dual-task interference. Traditionally, a dominant perspective within the motor control literature has been to view walking as an automatic process given the extensive level of practice this fundamental motor skill receives across the lifespan. Despite the relative ease with which we manage walking, recent evidence suggests that gait control is not an automatic process and requires at least a minimum of cognitive or  132  attentional involvement for movement control. For example, research has shown an increased attentional cost for walking when compared to standing and sitting (Bardy & Laurent, 1991; Lajoie, Teasdale, Bard, & Fleury, 1993), as well as secondary-task interference to gait control when a concurrent cognitive task is introduced (Evbersbach, Dimitrijevic, & Poewe, 1995). Relative to no-walking baseline measures, Sparrow, Bradshaw, Lamoureux, and Tirosh (2002) have recently demonstrated attentional costs associated with normal (unconstrained) walking and walking in conditions requiring participants to regulate step length on approaching a target. Attentional load was also shown to fluctuate incurring the greatest costs at gait initiation and when an obstacle was encountered, in comparison to RT during steady state or unconstrained walking. Finally, recent evidence has also demonstrated that increased attentional resources are needed for gait control in different populations including the elderly (Brown, Shumway-Cook, & Woollacott, 1999; Marsh & Geel, 2000), as well as persons with Parkinson's (Morris, lansek, Summers, & Matyas, 2000) and Alzheimer's disease (Camicioli, Howieson, Lehman, & Kaye, 1997). Taken together, these findings indicate that control of gait and control of preferred bimanual coordination states, as shown in the present investigation, are not automatic, but require at least a minimum of cognitive or attentional involvement. When considering that people execute motor skills to achieve a specific goal, it seems counterintuitive that the system would operate independently of central processes. Interestingly, results of the present investigation showed no significant difference in attention demands between the three coordination patterns. This finding was contrary to our initial predictions, as well as the work of others (Monno et al., 2000; Temprado et al., 1999; Zanone, Temprado, & Monno, 1999; Zanone et al., 2001). When comparing in-phase and anti-phase patterns, a greater performance trade-off (between the coordination pattern and RT task) is typically reported for antiphase coordination. This has also been found when scaling the frequency of oscillation under different attentional priorities. Manipulating oscillation frequency away from a participant's spontaneous frequency incurs a greater cost at the central  133  level for anti-phase, in comparison to in-phase, irrespective of attentional focus (i.e., a non-priority focus vs. a priority focus). Once again, a lack of a differential effect between coordination patterns in the present investigation may be a function of movement frequency. We predict that the attention demands required to sustain anti-phase performance should increase when movement frequency is progressively scaled beyond preferred modes, thereby increasing probe-RT. As cycling frequency increases beyond critical levels, attentional load needed to sustain the pattern should be at its greatest. Similar results are also predicted to occur for the 90° pattern; although it is speculated that reaction time to the probe task will be slower for the 90° pattern in comparison to the anti-phase mode. The 90° out-of phase task is not an inherently stable pattern of the human motor system and phase transitions from 90° of RP to 180° of RP are generally reported when movement frequency is progressively scaled (Smethurst & Carson, 2001). In essence, phase transitions from anti-phase to in-phase coordination, as well as from performance of the 90° pattern to the anti-phase mode, are predicted to correspond to increases in probe-RT. This idea is based on the findings of Monno et al. (2000) who examined the effects of attentional focus on dynamic stability and phase transitions between coordination states. In this experiment, participants were required to execute an in-phase pattern of coordination while gradually increasing frequency of oscillation under a non-priority and a priority condition. First, findings revealed that phase transitions occurred for both conditions, but were delayed when attentional priority was allocated to the bimanual coordination task. This delay in transition corresponded to an increase in pattern stability. Second, the data revealed longer RT for trials in the priority condition where a phase transition from anti-phase to in-phase occurred. From these findings, Monno et al. maintain that (a) pattern stability incurs a cost at the level of the CNS, and (b) phase transitions may be induced by two different mechanisms. In non-priority shared conditions, transitions may result as a loss of pattern stability, while transitions in a priority focused condition may result from too high of an attentional load needed to sustain the pattern. As such, Monnno et al. suggested that central costs associated with  134 attentional focus may act as a control parameter triggering phase transitions in bimanual coordination. Clearly examining the relationship between cycle frequency, probe-RT, and intentional focus warrants further investigation for understanding the coordination dynamics of preferred tendencies, as well as newly learned bimanual coordination tasks.  Chapter Footnotes 1.  Participants recruited from Experiment 3 did not receive feedback from the servo torque motors following trials of in-phase and anti-phase. The torque motors were used in Experiment 3 to facilitate acquisition of the 90° RP task, and therefore, were not required for the purposes of the present experiment.  2.  Although no statistical differences were revealed between pre- and post-test RT measures, results did show the p-value approaching significant levels (p_ = .064). Upon visual inspection of the data, it was found that one participant revealed large differences between pre-and post-test measures. For this particular individual, a general increase in RT occurred from pre- to post-test. Prior to in-phase and anti-phase performance, mean RT was 509 ms (SD = 50 ms) versus 656 ms (SD = 120 ms) following coordination performance. This was an absolute difference of almost 150 ms. One possibility is that RT to the secondary task increased as a result of fatigue, or perhaps, a loss of motivation following160 trials of in-phase and anti-phase coordination. To demonstrate the impact of this individual on baseline RT, the data were again submitted to a one-way repeated measures ANOVA with the data of this participant excluded from the analysis. Findings showed that the p-value no longer approached statistical significance (p_ = .171). Therefore, an increase in RT from pre- to post-test measures was not a general trend of all participants.  135  CHAPTER V EXPERIMENT 5 Removal of Continuous On-Line Visual Feedback Degrades Performance and Increases the Attention Demands of a Bimanual Coordination Pattern Introduction In the previous series of investigations, we manipulated the provision of relative motion feedback during performance of the 90° RP pattern. The reason for this manipulation was to determine the relative contribution of continuous on-line visual feedback to the acquisition and performance of a new bimanual coordination task. Findings revealed that changing the availability of visual feedback in the final 3 s of each practice trial influenced coordination performance. Providing continuous relative motion feedback improved the accuracy and stability of the 90° pattern (Experiments 2 and 3), while performance deteriorated when visual feedback was removed (Experiment 1). Results also showed that attention demands increased when the dominant source of feedback was changed in the final segment, irrespective of whether the switch was from a visual source to a proprioceptive source, or vice-versa. What remains unclear, however, is whether the changes observed in coordination performance and attention demands were temporary effects on behaviour, or whether these changes were more permanent in nature. The present investigation was designed to address this issue. Limitations in Experimental Design: The 'Segment Problem' In each of the four previous experiments, participants received a 13 s practice trial subdivided into 3 segments of different durations (see Figure 5.1 for a review of the breakdown of trial segments). Segment 1 consisted of the first 1.5 s of the trial. During this time, all participants were provided with relative motion feedback. This permitted individuals an opportunity to appropriately position their limb movements to the criterion template provided on the computer screen. Because this segment was used for calibration purposes, it was not included in analyses of the data. In  136  Figure 5.1. Breakdown of the 13 s practice trial in Experiments 1 through 4. The top panel represents the trial design used in Experiment 1 and for the Vision group in Experiment 4. The bottom panel illustrates the trial design used in Experiments 2 and 3, as well as for the No-Vision group in Experiment 4.  13s Trial  i  1.5 s:  8.5 s  Calibration  Vision  <  3s  No-Vision  13s Trial  UJ  I  I  I  I  I  1.5 s:  8.5 s  Calibration  No-Vision  1  I  I  : •  I  I 3s  Vision  137 contrast, we were specifically interested in performance during Segments 2 and 3. The second segment began at the 1.5 s mark and continued for the next 8.5 s, ending 10 s into the trial. At the 10 s mark, the third segment began and continued for 3 more seconds, until an auditory tone signalled the end of the trial. We were able to examine the influence of concurrent visual feedback on the performance of the 90° RP pattern because participants in our various investigations received relative motion feedback for only one of these two segments. More specifically, participants in Experiment 1 only received continuous visual feedback during the 8.5 s segment of the trial, while participants in Experiments 2 and 3 received relative motion feedback for only the last 3 s (not including calibration). Analysis of the data in Experiment 1 revealed that continuous on-line visual feedback had a critical impact on the acquisition and performance of a 90° RP task. When visual feedback was withdrawn for the last 3 s, many participants were unable to maintain accuracy and stability of the required relative phase pattern. Instead, performance drifted away from the criterion pattern to more intrinsically stable modes of coordination, even after 1, 000 practice trials. This was contrary to our original predictions. Once acquiring the 90° RP task, we expected that individuals would be able to maintain their performance in the absence of feedback, especially considering it was for 3 s only. Given that participants were exposed to relative motion feedback for a greater portion of time, we suggested that some performers might have developed a control strategy that was dependent on their capability to use the continuous on-line visual feedback to guide on-going movement. When visual feedback was removed in the 3 s segment, these participants were unable to use other information sources to maintain phasing requirements of the criterion task. There are, however, alternative explanations for the findings revealed in Experiment 1. The original interpretation assumes a relatively permanent change in behaviour when visual feedback is removed for the last few seconds of a trial. However, adding or removing augmented feedback from the environment (while simultaneously performing a complex task) may only act as a perturbation, inducing a temporary disruption in coordination performance. With enough time, the system  138  may be able to recover to the same levels of pattern accuracy and stability exhibited prior to the perturbation. Considering that (1) the no-vision segment in Experiment 1 was 3 s, and (2) participants only performed 1-1 14 movement cycles during this segment, the opportunity to recover from the perturbation was not adequately provided. In Experiments 1-3, findings also revealed that probe-RT was slower in the last 3 s of the trial regardless of whether individuals were performing in the absence or in the presence of continuous on-line visual feedback. Increases in probe-RT indicate a greater attentional cost at the central level. Therefore, in Experiment 1, this cost may be incurred as the individual attempts to maintain a required coordination pattern without the use of visual feedback. When an individual has limited experience or is unable to effectively use information sources other than vision, greater attention demands will be required to optimize performance. As a result, probe-RT increases when relative motion is removed for the final 3 s of the trial. But why might probe-RT increase when visual feedback is provided? After initial calibration to the computer monitor in Experiments 2 and 3 , the individual was unable to check his or her performance until the last 3 s of the trial. Therefore, the vision segment was especially important early in learning given that participants received no specific instructions regarding how to produce the required pattern. Information gleaned during the 3 s vision segment was essential for 'getting an idea of the movement' (Gentile, 1972). Therefore, a central cost was incurred as a result of the increased cognitive effort needed on the part of the learner. Once again, these interpretations assume a relatively permanent change in behaviour when visual feedback is added or removed from the perceptual display. Adding or removing augmented feedback during a trial may induce only a momentary increase in attention demands. Once an individual adjusts to the new information source, central costs may subsequently decrease. However, considering that the final segment of the trial was 3 s, we could not address whether the central costs observed in our previous investigations were a temporary effect or a more permanent, long-term effect.  139  In this experiment, participants practiced a bimanual coordination task requiring a phase offset of 90° on two consecutive days, engaging in 100 trials per day. Participants were then required to return to the laboratory for a third session to perform the criterion task under four different feedback conditions. Each test trial lasted a total of 17.5 s, which was subdivided into 3 segments. The first segment was 1.5 s (also referred to as calibration), while the final two segments were equal in length, lasting 8 s each. Relative motion feedback was manipulated in each of the final two segments, whereby participants could receive one of four possible feedback combinations based on whether continuous on-line visual feedback was provided (vision) or removed (no-vision) from the computer screen. Attentional costs were also assessed in each condition using the probe-RT technique. First, we hypothesized superior performance in the presence of real-time relative motion feedback, while removal of concurrent visual feedback was predicted to permanently degrade performance for as long as visual feedback remained unavailable to direct on-going movement. We also postulated that performance of the 90° pattern would demand attention irrespective of whether visual feedback was available or unavailable in the environment. However, reaction time to the secondary probe was predicted to be slowest immediately following a change in information source (e.g., from a visual information source to a proprioceptive source, or from a proprioceptive source to a visual source). Method Participants Informed consent was received from ten self-professed right-handed individuals from a university population (n = 5 females; n = 5 males). The mean age of participants was 24.6 years (SD = 4.1, range = 19-31 years). None of the participants had prior experience with the bimanual coordination task and all individuals were naive to the purpose of the experiment. Upon completion of the final session, all participants received remuneration. The experiment was carried out according to the ethical guidelines set by the University Behavioural Science Screening Committee for research involving human participants.  140  Apparatus The apparatus was the same as described in Experiment 1. Primary and Secondary Tasks Participants were required to learn to coordinate the limbs with a phase offset of 90° RP. While performing the criterion task, participants were required to emit a vocal response as quickly as possible if at any time a computer-generated auditory signal occurred. Both tasks have been previously explained in detail (see Experiment 1). Procedure The present experiment consisted of three separate sessions. The first two sessions were held at the same time on consecutive days, while the last session occurred following a 1-day retention interval of no practice. The procedures and general instructions for each session were identical for all participants. For all bimanual coordination tasks, participants were told that they could move at their own pace, but their overall objective was to make continuous full movement cycles within each acquisition trial that matched as closely as possible to the criterion template displayed on the computer screen (i.e., try not to let the trace vary). The requirement for one movement cycle was explained to consist of the full moving distance between the "IN" and "OUT" markers so that each hand returned to the same starting position. On phases of the experiment where probe-RT was collected, participants were informed that there was the potential for a computer-generated auditory 'beep' to sound while performing the required bimanual coordination pattern. If the signal for the auditory stimulus occurred, they were instructed to say the word 'pop' as quickly as possible. Instructions also stressed that performance of the coordination pattern should be given maximal priority.  141  Session 1 Baseline Trials: Probe-RT.  Participants were comfortably seated in front of  the computer monitor. They were then instructed that they would receive a total of 15 trials. During each trial, reaction time to the auditory probe was collected in the absence of the primary task. These trials served as measures of baseline RT. The auditory probe was delivered at variable times, randomized across trials. Each trial was 13 s in length.  Baseline Trials: Intrinsic Coordination Patterns.  Following collection of  baseline RT, participants were permitted to place their forearms in the manipulanda and to familiarize themselves with the movement of the apparatus. Criterion templates representing the relative phase patterns of in-phase and anti-phase were then displayed on the monitor. Participants were instructed on how to coordinate the limbs to produce the pattern represented by each template. Following general instructions, participants were provided two practice trials, one at the in-phase pattern and one at the anti-phase pattern. Baseline measures of in- and anti-phase were then collected. Participants performed a set of 12 trials in which the goal was to flex and extend the limbs in an in-phase mode of coordination, and a set of 12 trials requiring anti-phase coordination. Half of the participants performed the in-phase set first, while the other half received the anti-phase set first. Criterion templates representing in-phase and anti-phase patterns of coordination were displayed on the monitor during the participants' performance. Participants were provided with concurrent visual feedback for the entire trial by orthogonally plotting the displacements of both limbs overlaying the template of the criterion task. All trials were 13 s in duration. For each pattern, auditory probes were presented at variable times on 8 of the 12 trials. Baseline measures of in-phase and anti-phase coordination were collected to assess whether any changes occurred in the dynamic stability of pre-existing behavioural tendencies as a function of practice (see Zanone & Kelso, 1992, 1997).  142  Practice Trials. The criterion template (i.e., a circle) was now displayed in the middle of the computer monitor. Participants were informed that the goal of the practice trials was to learn how to coordinate the limbs to produce the circle pattern. Participants were told that they could move their limbs with either a right- or left-lead, but after the first ten trials they were required to move the limbs in the same lead-lag relationship for the duration of the experiment. Participants were required to discover the phasing requirements on their own and therefore, received no specific instructions on how to produce the criterion task at any time. To facilitate acquisition of the 90° criterion task, participants received relative motion feedback (in the form of a Lissajous figure) for the entire 13 s of each trial. Relative motion feedback was superimposed in real-time onto the template of the criterion task and stayed on the screen throughout the trial such that a history plot emerged as the trial progressed. The augmented feedback remained there until the completion of the trial. In contrast to previous investigations (Experiments 1-4), no terminal feedback was provided at the end of the trial. This was due to the fact that participants received concurrent visual feedback for the duration of each trial. Participants received a total of 100 practice trials, subdivided into 10 blocks of 10 trials. Each participant determined the interval between blocks, in order to offset any potential effects of fatigue across the 100 trials. Session 2 Practice Trials. Participants performed a total of 100 practice trials in the same manner as described above. No baseline measures of probe-RT or intrinsic coordination patterns were collected on Day 2. Session 3 Test Trials. Participants were required to perform 90° RP under four different feedback conditions. Trials in each condition lasted 17.5 s and were divided into 3 segments. The first segment lasted 1.5 s, while the final two segments of the trial lasted 8 s each. For the first 1.5 s, participants received relative motion feedback in real-time, irrespective of condition. In the final two segments of the trial, participants  143 performed 90° RP either in the presence or in the absence of continuous visual feedback (see Table 5.1). In Condition 1, participants received visual feedback for Segment 2 only. For Segment 3, participants were required to perform the criterion task in the absence of feedback (Vision: No-Vision or V-NV). The opposite manipulation occurred in Condition 2. Visual feedback was provided in Segment 3, but not in Segment 2 (No-Vision: Vision or NV-V). In Condition 3, participants received relative motion feedback for the duration of the trial (Vision: Vision or V-V), while visual feedback was removed for the last two segments in Condition 4 (NoVision: No-Vision or NV-NV). Participants performed 24 trials of each condition, which were presented randomly across 96 trials. To prevent fatigue, the 96 trials were subdivided into 8 blocks of 12 trials. For each test condition, probes were presented randomly at eight different times. These occurred at: 2.5 s, 4.5 s, 6.5 s, 8.5 s, 10.5 s, 12.5 s, 14.5, and 16.5 s. Probes delivered at the first four times (i.e., 2.5 s to 8.5 s) occurred during the first 8 s segment; while, the last four times (i.e., 10.5 s to 16.5 s) took place during the final 8 s segment of the trial. For the purposes of data analysis, probes at the 2.5 s position and at the 10.5 s position were referred to as Time 1 (T1) because they were the first probes to occur within each segment. Probes at the 4.5 s position and the 12.5 s position were referred to as Time 2 (T2), while probes at 6.5 s and 14.5 s, as well as at 8.5 s and 16.5 s were identified as Time 3 (T3) and Time 4 (T4), respectively. For each condition, a total of 2 probes were presented at each time (n = 16 probes/condition). Baseline Trials: Intrinsic Coordination Patterns. Participants performed 12 trials of the in-phase pattern and 12 trials of the anti-phase pattern in the manner previously described. Once again, auditory probes were presented at variable times on 8 of the 12 trials for each coordination pattern. Baseline Trials: Probe-RT. Measures of baseline-RT were collected in an identical manner as pre-test measures described earlier.  144  Table 5.1. Summary of Feedback Conditions (V = Vision; NV = No-Vision) Trial (17.5 s) Feedback Condition  1 2 3 4  Condition Abbreviation*  V-NV NV-V V-V NV-NV  Segment 1  Segment 2  Segment 3  1.5 s  8s  8s  V V V V  V NV V NV  NV V V NV  * Determined by the presence or absence of concurrent relative motion feedback in Segments 2 & 3, respectively.  145  Data Analyses Statistical analysis was conducted on each of the following dependent measures: RMSE, SD of RP, cycle frequency, and probe-RT. These measures have been discussed previously (see Experiment 1). All post-hoc comparisons were performed using the Tukey HSD procedure (fi < .05). Results Baseline RT To determine baseline RT, a mean reaction time was calculated for pre- and post- acquisition control conditions and submitted to a one-way repeated measures analysis of variance. No significant difference (fi = .190) was found between preacquisition (M = 393 ms; SD = 80 ms) and post-acquisition (M = 390 ms; SD = 75 ms) baseline RT. Therefore, baseline RT was calculated using a mean of both preand post-acquisition measures (M_ = 392 ms; SD = 78 ms). Intrinsic Coordination Patterns To assess whether pre-existing coordination patterns were affected by practice on the 90° out-of-phase task, the dependent measures of RMSE, SD of RP, and cycle frequency were submitted to separate 2 x 2 (Pattern x Time) repeated measures ANOVAs. The pattern factor included performance for in-phase and antiphase, while the time factor included pre-practice and post-practice measures. Results revealed a significant difference between in-phase and anti-phase for RMSE, F(1, 9) = 12.05, fi = .007, and SD of RP, F(1, 9) = 15.10, fi = .004. The inphase pattern was performed with less error (M = 11 -2°, SD = 3.8°) and was more stable (M = 9.6°, SD = 3.2°) in comparison to performance of the anti-phase pattern (RMSE: M = 16.1°, SD = 5.2°; SD of RP: M = 15.0°, SD = 4.8°). For RMSE, inphase coordination was consistent between pre-test (M = 11.2°, SD = 3.9°) and post-test measures (M = 11 -2°, SD = 3.8°), while anti-phase coordination exhibited a  146  slight, but non-significant (p = .085), decrease in error from the pre-test (M = 17.9°, SD = 5.9°) to the post-test (M = 14.3°, SD = 4.0°) (see Figure 5.2, top panel). For SD of RP, the time factor approached conventional levels of significance (p = .057). Post-test performance (M = 10.9°, SD = 3.6°) tended to be more stable than pre-test performance (M = 13.5°, SD = 5.6°) for both in-phase (Pre-test: M = 10.1°, SD =3.7°; Post-test: M = 9.1°, SD = 2.7°) and anti-phase (Pre-test: M = 16.6°, SD = 5.2°; Post-test: M = 12.7°, SD = 3.7°) coordination (see Figure 5.2, bottom panel). Contrary to the conclusions of Zanone and Kelso (1997), no evidence was found for the destabilization of preferred coordination modes following learning of a new bimanual coordination pattern. Self-selected rate of cycle frequency increased for in-phase coordination from a pre-acquisition rate of 0.809 Hz to a frequency of 0.972 Hz post-acquisition. Antiphase coordination also increased in cycle frequency from 0.726 Hz to 0.865 Hz. This was confirmed by a significant main effect for time, F(1, 9) = 7.14, p = .024. Although preferred rate of cycle frequency tended to be faster for the in-phase pattern when compared to anti-phase coordination, this trend did not reach significance (p = .085). Finally, a Task Condition (probe-RT, baseline RT) x Pattern (in-phase, antiphase) x Time (pre-test, post-test) repeated measures ANOVA was administered to assess the attention demands of pre-existing coordination patterns. Results showed that probe-RT was significantly slower while performing naturally occurring coordination tendencies, in comparison to baseline RT, F(1, 9) = 81.64, p = .0000. Analysis revealed no significant differences for pattern (p = .517) or time (p = .687). Mean probe-RT was 603 ms for the in-phase pattern (SD = 144 ms) and 619 ms for the anti-phase pattern (SD = 169 ms) (see Figure 5.3). Practice Trials By the last block of acquisition on Day 2, all participants were able to produce a relative phase pattern within a ±10° bandwidth of the criterion pattern. A representative sample of participant data is displayed for Day 2 as Figure 5.7 in the Chapter Footnotes (see #1).  147  Figure 5.2. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for preferred coordination modes (0° and 180°) pre- and post-acquisition (Error bars = SEM).  0-deg RP  180-deg RP  Reouired Pattern of Coordination Pre-test Post-test  0-deg RP  180-deg RP  Required Pattern of Coordination  148  Figure 5.3. Mean probe-RT for 0° and 180° of relative phase compared to baseline RT (M = 392 ms). An asterisk indicates a statistically significant difference when baseline RT was compared with the corresponding measure of probe-RT (Error bars = SEM).  i = i 900 -j 800 700 -  0-deg RP  180-deg RP  Required Pattern of Coordination  Pre-test Post-test Baseline  149  Test Trials Performance of the 90° out-of-phase Pattern. Each dependent measure (RMSE, SD of RP, Cycle Frequency) was submitted to a condition (V-NV, NV-V, VV, NV-NV) x segment (2) repeated measures ANOVA. No significant differences were found for self-selected cycle frequency. Mean cycle frequency was 0.50 Hz across all conditions and trial segments. Figure 5.4 displays RMSE in the top panel, while SD of RP is presented in the bottom panel. Analysis revealed a significant condition effect for both RMSE, F(3, 27) = 12.29, p = .0000, and SD of RP, F(3, 27) = 13.40, p = .0000. Further post-hoc  analysis revealed that the 9 0 ° pattern was more stable and performed with significantly less error in the V-V condition  (M RMSE = 11.4°; M SD of RP = 1 0 . 5 ° ) ,  (M RMSE = 2 5 . 8 ° ; M_ SD of RP = 1 4 . 6 ° ) and the NV-NV condition (M RMSE = 3 0 . 1 ° ; M SD of RP = 16.1°). The goal pattern was as compared with the NV-V condition  also less accurate, as well as more variable, in the NV-NV condition versus the VNV condition  (M RMSE = 18.9°; M SD of RP = 12.8°).  Results showed a significant condition by segment interaction for both RMSE, F(3, 27) = 34.84, p = .0000, and SD of RP, F(3, 27) = 10.04, p = .0001. This analysis  revealed that RMSE and SD of RP were significantly lower in the presence of visual feedback. However, when a no-vision segment immediately preceded a vision segment (NV-V) the improvements observed in performance during the vision segment did not reach the levels displayed under vision for both the V-NV and V-V conditions. Figure 5.5 presents a series of typical relative motion plots obtained for a representative participant on each condition. Panel A represents the V-V condition, while panels B, C, and D correspond to the V-NV, NV-V, and NV-NV conditions, respectively. In panel A it is evident that the individual is coordinating the limbs with a 9 0 ° offset for the entire trial. Performance is clearly superior when comparing panel A to the remaining three panels. In Panel B, a circle shape is produced during the vision segment, but a rightward leaning elliptical shape emerges when vision is removed. Following calibration in Panel C, the individual performs one circle shape  150  Figure 5.4. RMSE of goal relative phase (top panel) and SD of RP (bottom panel) for condition presented by trial segment. An asterisk indicates that a statistically significant difference was found when comparing segments within a respective condition (Error bars = SEM). Segment 1 »  Segment 2  40  CD OJ CD  35  TD  £  30  CD  ro 2 5 CD >  20  £  15  CD  a  o  10  W CO  5  £  n  0  V-NV  NV-V  V-V  NV-NV  Condition Segment 1 Segment 2  V-NV  NV-V  V-V  Condition  NV-NV  151  Figure 5 . 5 . Individual trial data displayed as Lissajous figures for Participant 3. Panel through D represents the V - V , V - N V , N V - V , and N V - N V conditions, respectively. The solid line always represents a vision segment, while the dotted line always represents a no-vision segment (x-axis = right forearm displacement, y-axis = left forearm displacement). A  152  and then drifts towards in-phase. Performance is maintained at this preferred phasing tendency until vision is restored, whereupon the individual immediately returns to a phase offset of 90°. Panel D, displays a similar finding. Following calibration, the learner makes one circle shape and drifts away from the criterion pattern towards in-phase. However, this is maintained for the rest of the trial because feedback was not restored to the computer monitor. This series of Lissajous figures clearly demonstrates a permanent phase drift towards a more preferred coordination tendency whenever visual feedback is removed from the perceptual display. Most importantly, performance does not return to the same levels of accuracy and stability exhibited prior to removal of the feedback even when time is provided to recover from the perturbation. Probe-RT. A preliminary analysis was first conducted to verify that performance of the 90° pattern was maintained on trials where a probe occurred, in comparison to no-probe trials (see Experiment 1 for further discussion of rationale). RMSE data were submitted to a Trial Type (probe, no-probe) x Condition (V-NV, NV-V, V-V, NV-NV) x Segment (2) repeated measures ANOVA. No significant differences were found for trial type (p = .284). Mean RMSE was 21.5° when an auditory probe was administered while performing the 90° pattern and 22.0° in noprobe trials. The probe data were then submitted t o a 4 x 2 x 4 x 2 (Condition x Segment x Probe Time x Reaction Time) repeated measures ANOVA. Probe time referred to the position of a probe within a trial segment: T1 (2.5 s, 10.5 s), T2 (4.5 s, 12.5 s), T3 (6.5 s, 14.5 s), or T4 (8.5 s, 16.5 s); while the factor of reaction time included the measures of probe-RT and baseline RT. Findings revealed a significant four-way interaction, F(9, 81) = 2.23, p = .024, for all factors. Mean probe-RT for each condition is displayed as a function of segment and probe position in Figure 5.6. The analysis revealed that probe-RT was consistently faster in the presence of visual feedback, as compared to all no-vision segments. Importantly, when visual feedback was followed by no-vision (V-NV), RT to the secondary probe increased markedly. In comparison, when a no-vision  153  Figure 5.6. Mean probe-RT plotted as a function of segment and probe position for each condition compared with baseline RT (M = 391 ms)*.  1000 950 900 850 800 750 CO E 700 650 600 CD 550 £ 500 Q. 450  —  -  V-NV NV-V V-V NV-NV Baseline  -v  o o.  . ,o  ^ .  —V  -v  I  •  -~  '»—  /  i  ,  i  2.5  1  4.5  1  6.5  8.5  10.5  Segment 1  1  1  12.5  14.5  1  16.5  Segment 2 Time of Probe (s)  *Note: Error bars are not presented above to enhance visual presentation of the four-way interaction. Instead the same data is displayed with error bars in Figure 5.8, which is provided in the Chapter Footnotes (see #2). In Figure 5.8, the V-NV and NV-V conditions are displayed separately from the V-V and NVNV conditions.  154  segment was followed by visual feedback (NV-V) there was a spike in probe-RT (immediately after the transition to visual feedback) followed by a rapid decline in probe-RT to levels observed in the vision segment of the V-NV and V-V conditions. These data support the idea that attention demands increase when a switch from visual to proprioceptive feedback sources occurs within a trial. Moreover, the central costs associated with these changes appear to be relatively persistent in nature. In contrast, shifting from a proprioceptive information source to a visual source induces only a temporary increase in attention requirements. Following the switch, it appears individuals quickly adjust to the provision of continuous on-line visual feedback, and central costs subsequently decrease to pre-transition levels. Irrespective of whether visual feedback was available or unavailable to guide on-going movement, post-hoc analysis revealed that probe-RT was always significantly slower when performing the required pattern, as compared to baseline RT. This provides support for our previous findings (Experiments 1-4), which suggest that the execution of the 90° pattern is attention demanding. Central costs were significantly less in the V-V condition (M = 563 ms) when compared to all other conditions (V-NV = 642 ms; NV-V = 685 ms; NV-NV = 669 ms). Discussion In the present experiment, four different feedback conditions were examined based on whether concurrent visual feedback was available or unavailable to guide performance in one of two trial segments. We predicted superior performance of the 90° pattern in the presence of continuous visual feedback, while removal of visual feedback was hypothesized to permanently degrade performance for as long as the information source was unavailable to guide on-going movement. Our findings confirmed this prediction. The 90° pattern was more accurate in the presence of concurrent visual feedback. Importantly, these data also show that when visual feedback is removed from the perceptual display, a relatively permanent phase drift occurs towards a more preferred coordination tendency. That is, performance does not return to the same levels of accuracy and stability exhibited prior to removal of the feedback even when time is provided to recover from the environmental  155  disruption. These findings indicate that the 3 s segment in Experiment 1 was representative of coordination performance under conditions of no-vision. The data confirms that when vision is removed as an information source (while executing the 90° pattern) individuals are generally less effective at suppressing the influence of pre-practice coordination tendencies to maintain the phasing requirements of the required task. When a no-vision segment immediately preceded a vision segment, findings showed that the difference between segments was not as pronounced in comparison to the V-NV condition, nor did performance during the vision segment reach levels comparable with performance in the V-V condition. The former observation is also consistent with our previous results (Experiments 2 and 3). These data indicate that concurrent visual feedback is less effective for inhibiting the influence of pre-existing coordination biases if it is presented after the limbs are already moving. When vision is restored the individual attempts to correct the movement, but appears to experience difficulty in overcoming the physical properties of a moving limb (e.g., inertia) or in reducing the already-established coupling strength between the limbs. As a result, performance gains are not as large in comparison to performance when vision is presented first. It is even possible that the performer would experience greater performance benefits by stopping the movement and beginning the pattern over again. Of course, this was not an option given the procedures of the current experiment. With respect to probe-RT, we predicted performance of the 90° pattern would demand attention irrespective of whether visual feedback was available or unavailable in the environment. This prediction was also supported. Our data revealed that performance of the 90° pattern was attention demanding under all four feedback conditions. Interestingly, probe-RT dramatically increased in the V-NV condition directly following the switch between information sources. However, this increase was only temporary as probe-RT immediately returned to levels comparable with the V-V condition. This finding indicates that presenting visual feedback during a trial will momentarily increase attentional load. Once the individual adjusts to this new information source, central costs will subsequently decrease. For  156  example, when relative motion displacement plots are provided in the middle of a trial, individuals are suddenly required to make spatial (or visuomotor) transformations that relate limb movements made in the horizontal plane to information captured in a single percept (i.e., a circle) on a vertical screen. It is possible that spatial transformations require additional attention (Puttemans, Vangheluwe, Wenderoth, & Swinnen, 2004) until the individual can adapt to the feedback source. An increase in probe-RT following a switch from proprioceptive feedback to visual feedback is consistent with the findings of both Experiments 2 and 3 where probe-RT during the vision segment was significantly slower in comparison to probeRT in the absence of vision. However, the present findings also show that providing a longer final segment in Experiments 2 and 3 would have enhanced our interpretation of the data. Given a longer final segment, we would have likely seen only temporary increases in probe-RT. The findings of the present investigation suggest that performing the 90° pattern in the presence of augmented visual feedback is less attention demanding than performing in its absence. One possible reason for this decrease in central costs is that concurrent visual feedback (in the form of a Lissajous trace) is relatively simple, clearly specifies the task goal, and provides performers with crucial features of the movement, which alert the individual to the properties and relationships inherent within the task. In essence, real-time relative motion displacement plots provide the individual with information that they do not necessarily have to find on their own, thereby reducing cognitive load. This idea also follows when providing a criterion template, as well as Lissajous traces of performance superimposed onto the template. In this respect, participants are provided very salient information regarding performance, which reduces the cognitive effort required to detect that an error has occurred. Explanations for differences in attention demands between vision and novision feedback conditions may also have neural origins. For example, Debaere, Wenderoth, Sunaert, van Hecke, and Swinnen (2003) have recently provided evidence for a relative functional dissociation between several regions of the brain  157  when examining cyclical bimanual coordination movements in the presence or in the absence of continuous on-line visual feedback. In this investigation, participants were required to perform a newly learned 90° pattern under externally guided or internally guided conditions. In the former condition, on-line feedback was provided in the form of a Lissajous figure. In the latter, participants were required to close their eyes and produce the movement on the basis of a memory representation, without on-line visual feedback. Using functional magnetic resonance imaging (fMRI), results showed higher activation levels under externally guided coordination in the inferior temporal gyrus, the superior parietal cortex, the premotor cortex, the thalamus, and the cerebellum (lobule VI). In contrast, preferential involvement of the basil ganglia, the supplementary motor area, cingulated motor cortex, the inferior parietal, frontal operculum, and cerebellar lobule IV-V/dendate were shown during internally-guided coordination. The idea that performance discrepancies in the present investigation may be a result of the activation of differential pathways in the central nervous system is an important one. It may help explain the role of various brain regions, and the efficiency with which each processes information for the control and guidance of bimanual coordination movements. This has obvious implications for motor disordered groups in a rehabilitation setting. For example, providing Parkinson's patients with continuous visual feedback may facilitate acquisition of new bimanual coordination patterns. In a motor disorder like Parkinson's disease the basal ganglia is a deficient brain area, which results in disruption of internal generation of movement. However, since real-time relative motion feedback provokes higher levels of activation in the parietal-premotor area of the brain, bimanual coordination performance is predicted to improve because the basal ganglia can be largely bypassed. From this perspective, investigating the benefits of augmented feedback is clearly warranted for clinical applications in various populations, including motor disordered groups, as well as older adults (see discussion of Experiment 2).  158  Chapter Footnotes Figure 5.7. Individual trial data for Participants 1 (top panel) and 9 (bottom panel). Mean RP is plotted across 100 trials for Day 2 of practice. 180 170 160 H 150 -» 140 o £ 130 % 120 = 110 £ »  I  o  90 80 70 60 50 40 30 20 10 0  H  101  110  119  128  137  146  155  164  173  182  191  200  164  173  182  191  200  Trial Number 180 170 160 150 140  8  130 £ °> 120 •D  £ "g £  5  Io  1 1 0  100  80 4 80  0 c 60 ra  |  7  50 40 30 20 -\ 10  0  101  110  119  128  137  146  155  Trial Number  159  Figure 5.8. Mean probe-RT compared with baseline RT (M = 391 ms) for the V-NV and NV-V conditions (top panel) and the V-V and NV-NV conditions (bottom panel) plotted as a function of segment and probe position (Error bars = SEM). 1000  V-NV NV-V Baseline  900 800 to  E c  700  I-  600  °r cb  o  500 400 300  V  0i 2.5  4.5  6.5  8.5  10.5  Segment 1  12.5  14.5  16.5  Segment 2 Time of Probe  1000  V-V NV-NV Baseline  900 800 CO  E c  700  H  600 \  i o  500 H  *  400 300  . i 2.5  4.5  6.5  8.5  10.5  Segment 1  12.5  14.5  Segment 2 Time of Probe  16.5  160  CHAPTER VI Attention Demands across Extended Practice of a Bimanual Coordination Task: General Discussion Man's versatility in developing skills required to cope with modern living is extremely great. Depending on where he lives in the world, he will acquire one or more of the thousands of different languages and dialects. Depending on the kind of work he undertakes, he will develop hundreds of technical skills. Entirely without being aware of it, he will develop countless universal skills ranging from those required for standing up, sitting, walking, running and manipulating objects to those required in perceiving the world about him. Yet in spite of this tremendous versatility, man's capacity is not limitless. The rate at which he acquires skills and the level of performance he is able to attain are subject to limitations imposed by his musculature, his nervous system, and the characteristics of the activities themselves (Fitts & Posner, 1967, p. 1).  When observing everyday behaviour, it becomes readily apparent that human motor performance is not constrained to the execution of only one task at a time. Rather, individuals often display the capacity to effectively perform two tasks concurrently (e.g., walking and talking at the same time). However, the simultaneous execution of two tasks can also be very difficult, and performance is rarely of the same quality than if each task had been performed alone. Referred to as dual-task interference, the primary purpose of this thesis was to investigate the performance of dual-tasks and observe how attentional limitations in performance change across the time course of motor learning. Specifically, we investigated the attention demands of three bimanual coordination patterns (0°, 180°, and 90° of relative phase) across a series of five experiments. The bimanual coordination tasks were chosen because they are more representative of real-world activities (Hodges & Lee, 1999), when compared to traditional tasks normally employed in the laboratory setting. Furthermore, this approach allowed us to compare the attention demands of pre-existing coordination tendencies (i.e., 0° RP, 180° RP) with the attention demands of a relatively difficult pattern (i.e., 90° RP) that cannot generally be performed prior to practice. Within a controlled setting, however, this pattern can be acquired in a relatively short period  161  of time versus the acquisition of sport skills in general. Providing practice sessions for the learning of the 90° pattern allowed us to investigate changes in attention demands as a function of practice via a dual-task paradigm. Results showed that RT to a discrete secondary probe decreased as a function of practice, but never reached automatic levels (as defined by modal perspectives of attention) even after 200 (Experiment 5) and 1, 000 (Experiments 13) trials of task-specific practice. These findings suggest that skilled performance is not impervious to dual-task decrements; rather performers in the advanced stages of learning merely show less decrement than in earlier stages. This interpretation was further strengthened by the finding that intrinsic bimanual coordination patterns did not appear to operate automatically. In Experiment 4, dual-task decrements were demonstrated for both in-phase and anti-phase modes of coordination indicating that voluntary stabilization of even preferred coordination modes require at least a minimal amount of cognitive or attentional involvement. Findings of this nature have important practical applications for the concomitant execution of many everyday tasks. One such example is the use of cell phones while driving. Research using car simulators and laboratory-based simulated driving situations provide evidence that demonstrates an attention-related basis for driving accidents. For instance, Strayer & Johnson (2001) reported that drivers talking on a cell phone miss 2 x more simulated traffic signals when compared to driving without talking on a cell phone. More importantly, when signals were identified and reported, RT was significantly slower as compared to RT when only performing the task of driving. The results of our experiments would further suggest that attention-based deficits in performance will continue to persist even if the individual is well-practiced at the two tasks, further highlighting the dangerousness of cell phone use while driving.  Attention Demands: Theoretical Perspectives From a theoretical perspective, our findings do not support traditional modal views of automatic processing, which postulate that performance on a task will eventually become automatic (i.e., attention-free processing) as a function of  162  practice. From a practical perspective, it seems highly unlikely that humans would ever reach this level of processing given the complex nature of even the simplest of human movements. This is obvious in that research has shown that practice does not necessarily make perfect. For example, Crossman (1959) reported small performance improvements in the task of cigar-rolling by workers who made over 10 million cigars and possessed seven years of task experience. This is an important idea as dual-task interference may always persist as long as learning takes place and motor performance continues to improve (Blischke, 1998). Even Fitts (1962), who labeled his final stage of learning with the term 'autonomous', recognized that new learning occurs at this final stage albeit at a decreased rate. Rather than true performance plateaus or limits in the individual's capacity for improvement, Fitts suggested that physiological aging and/or a loss of motivation are the real determinants of an individual's failure for perpetual gains in performance. As long as learning takes place, performance will be attention-demanding. In contrast to the original controlled and automatic theory of Schneider and Shiffrin (1977, see Appendix A), more recent discussions on attention tend to focus on the qualitative changes that occur in task structure (e.g., strategy changes Cheng, 1985), the performance of individual task components (e.g., strengthening Anderson, 1982), or the performance of component sequences (e.g., composition Anderson, 1982). For example, Vallacher & Wegner (1987) suggest that attention to a task is devoted to one of many different levels of organization. In general, people attend to the highest level of organization possible, which allows the task to be completed without shifting to lower levels of attention. In this sense, the lower levels of organization are automatic, but attention is not withdrawn, merely shifted to a higher level of organization, which becomes progressively higher with more practice. In certain environmental conditions, however, attention may be forced down to a lower level. Logan (1992) provides an example to illustrate this idea by describing different attention levels associated with dialing a telephone number. When dialing we tend to think of the number we are dialing, not the buttons we are pressing. However, if one of the buttons was to stick, our focus of attention would shift to pressing the button. From this perspective, attention is always required for goal-  163  directed movement; however, the way in which the individual attends may change as a function of skill level. Cheng (1985) suggests that improvement in performance results from optimizing strategies and the restructuring of the task components. This idea becomes apparent when an individual is asked to find the sum of ten 5s. To derive a solution to the problem, alternative methods may be used. The individual may perform ten addition operations to derive the answer or may solve the problem in only one operation by using the multiplication table. However, use of either method is dependent on the skill level of the performer. An individual who is just learning mathematics (i.e., a beginner or novice), will not know the multiplication table. Therefore, the person will have to perform ten addition operations. It is only later in learning, when the individual knows the multiplication table, the same problem can be solved in only one step. Such gains in efficiency are not a result of the individual performing the addition operations in a capacity-free mode of processing. Instead, restructuring the task components and using a more efficient procedure achieved the gains. What changed was how the information was processed, not what information was processed. According to Haider and Frensch (1996), decreases in RT observed using the dual-task paradigm, may also reflect a reduction in amount of information being processed, and not necessarily a reduction in attention demands. This idea is based on the observation that skilled performers learn to discriminate between taskrelevant and task redundant information. This is advantageous to the performer because it reduces the amount of information to be processed to only the relevant aspects of the task. Therefore, information processed at early stages of acquisition may be qualitatively different than information that is processed at advanced stages. From this perspective, what changed was the information, not necessarily how the information was processed. In recent years, attention demands (in relation to automaticity) have been discussed most often from either a procedural account or a memory account of skill acquisition. A procedural account proposes that procedures are developed then refined and strengthened as a result of practice. In contrast, memory accounts  164  suggest that performance is dependent on memory retrieval (rather than resource limitations), which becomes faster and more efficient as a function of practice. Procedural accounts of skill acquisition generally refer to Anderson's Framework for Cognitive Skill Acquisition, while memory accounts reference Logan's (1988) Instance Theory of Automaticity, the latter of which, will be presented first. Logan's (1988) Instance Theory of Automaticity According to Instance Theory, the novice performer initially performs many tasks by following a mental algorithm that is general in nature. Eventually, this same task is solved by a single-step direct-access retrieval of the solutions from long-term memory. For example, when an adult is asked to report the sum of 9 plus 6, they can provide a response rather fast and effortlessly. For a child, calculating the sum of 9 plus 6 is arduous and slow. To explain this, Logan proposes that a 'memory trace' (i.e., instance) is left in long-term memory each time the individual solves the problem. The next time the individual performs the task, associated traces are activated, and a new trace is laid down. Furthermore, Logan maintains that the individual may respond with either solution, but it is generally a race between computational processes (i.e., the algorithm known to solve the problem) and memory processes (i.e., retrieval of an appropriate memory trace). Early in learning, the computational processes are faster. With practice, however, the memory trace becomes stronger and eventually the likelihood that an appropriate trace can be retrieved before the algorithm is finished increases. Accordingly, processing at later stages of learning does not imply processing without attention; rather, processing involves a different way of attending relative to early stages of acquisition.  Anderson's (1982) Cognitive Framework of Skill Acguisition Following Fitts and Posner's (1967, see Appendix A) phase approach to the acquisition of skill, Anderson (1982) developed a framework, which takes into consideration the underlying mechanisms of performance throughout each phase of acquisition, as well as the transition that occurs between one primary stage and another. In general terms, Anderson's theory is based on the notion of a production  165  system in which a distinction between declarative and procedural knowledge is fundamental. In the first stage, referred to as the declarative knowledge phase, it is postulated that the learner receives instruction and information regarding the to-belearned skill or task. This information is then encoded by the learner as a set of facts, which are used by general procedures already possessed by the individual in order to generate behaviour. Rehearsal is crucial at this stage because the facts must be retained in working memory so that they are available for use. Activities that occur within this stage of development are analogous to the activities described in Fitts' cognitive stage (see Appendix A), and essentially involve the acquisition of explicit knowledge in the form of declarative rules pertaining to a specific domain. As practice continues, specific procedures develop so that the goals of the task are more readily achieved. This process involves the conversion of knowledge from declarative form into procedural form. This process can be described as the acquisition of production rules that associates a particular stimulus condition with an appropriate response (e.g., if X-ray contains Feature A, Feature F, and Feature S, then consider Diagnosis Y). Conversion affords an individual the capability to directly apply procedures without having to actively process declarative knowledge regarding step-by-step information about how the task is executed. Transition between declarative and procedural knowledge is referred to as the knowledge compilation phase and is akin to Fitts' intermediate or associative phase. According to Anderson, attention demands are reduced when domain specific knowledge is compiled into a set of if-then-statements, which is referred to as proceduralization. Moreover, composition can also occur, whereby productions occurring in sequence and sharing the same overall goal are collapsed into a single production. A benefit to reducing the number of production rules required to perform a single task is that overall task goals may be achieved in fewer discrete steps. If each processing step is associated with a unit of time, composition allows the individual to save processing time and ultimately leads to greater efficiency in performance. Since fewer steps are required for task execution, it is also postulated  166  that there is an associated load reduction in working memory since it does not have to be updated as often during the performance of a task. After declarative knowledge has been compiled into a production system, 'tuning' of the procedure takes place. Improvement occurs as a result of refining the procedure through three processes: generalization, discrimination, and strengthening. Generalization refers to broadening the range of procedure applicability; discrimination denotes the narrowing of rules to fit only appropriate situations; while strengthening implies the weakening of poor rules and the strengthening of better ones (Proctor & Dutta, 1995), all of which enables the individual to execute productions in a fast and reliable manner. As learning progresses from the application of explicit knowledge to implicit knowledge, the contents or working memory are dramatically reduced so that it contains only initial condition and final products of performance. The production rules no longer resemble the original declarative knowledge. Moreover, when individuals are asked to verbally describe their performance strategy they are often unable to do so. This occurs because skilled performers no longer use working memory to describe how a task is executed. Instead, individuals must reconstruct what must have taken place during performance. Once again, according to this perspective, processing at later stages of learning does not imply processing without attention; rather, processing involves a different way of attending relative to early stages of acquisition. Is Bypassing Attention a Good Thing? Automatic processing has always been viewed as a highly desirable characteristic and is a general aim of motor learning - an idea that has become heavily entrenched in traditional theories of skill acquisition. However, discussion pertaining to the costs of automatic processing has been largely ignored within the literature. For example, Schmidt (1987) has stated that research has focused "too heavily on the benefits received from automatic processing in skilled performance without considering sufficiently their costs" (p. 100). Moreover, some theorists have challenged whether true automaticity (as defined by capacity-free processing)  167  should even be a desirable state of behaviour. This latter idea is briefly discussed with respect to error detection and correction capabilities and deliberate practice. Error Detection and Correction According to the original definitions proposed by Schneider and Shiffrin (1977; Shiffrin & Schneider, 1977), controlled processing is: under intentional control, limited by the availability of processing resources, effortful, and flexible in response to changes in the environment. In contrast, automatic processing was defined as unintentional behaviour, which does not use any processing resources, and is effortless. Importantly, their original theory posits that once an automatic process is initiated, all processing runs to completion in lieu of subject intervention. This makes control, modification, and/or suppression of the current action extremely difficult, if not impossible. From this perspective, 'less control' is equated with the execution of involuntary responses that occur outside of awareness. However, this conceptualization poses a problem for the observation that skilled performers appear to possess greater control and awareness of well-developed movement skills (Ivry, 1996). One such example is the capability to detect and correct movement errors during the execution of a task. According to closed-loop theories of movement control, an individual compares his or her current output to an internalised standard of correctness during the execution of a movement. It is further hypothesized that any difference between the output and input is then fed to the executive level of processing, which acts to reduce the discrepancy to zero. As such, discrepancies must be brought into conscious awareness by way of system monitoring if an action is to be altered during performance. In accordance with Schneider and Shiffrin's (1977; Shiffrin & Schneider, 1977) conceptualization, skilled individuals should exhibit a diminished capability to detect errors when compared to novice (or less skilled) performers. Such a prediction, however, is contrary to a number of skill acquisition theories (e.g., Fitts & Posner, 1967). Empirically, Schmidt and White (1972) have provided evidence suggesting that error detection capabilities are enhanced as a function of practice. Using a ballistic timing task, Schmidt and White (1972) asked participants  168  to verbally estimate their movement time immediately following the performance of the movement task. The difference between the participant's evaluation of movement time (subjective error) and actual movement time (objective error) was calculated, and within subject correlations were computed. Findings revealed that as practice progressed, there was an increasing correspondence between the two error scores suggesting that error detection capabilities were enhanced as a function of practice. Recently, Bredin and Bouffard (2000) and Bredin, Hodges, Sturrock, Franks, and Chua (2000) have demonstrated similar findings by examining the error detection and estimation capabilities of skilled and novice performers following the execution of complex motor tasks (i.e., the full golf swing and the golf chip, respectively). Although these studies did not use an on-line error detection task, this evidence still suggests that characteristics exhibited by skilled performance (e.g., the capability to detect error) may only manifest as a result of an increased capability for controlled processing, not its elimination. Other empirical evidence demonstrates that skilled typists are able to inhibit high speed typing within one or two keystokes of detecting an error (Rabbitt, 1978) or upon the occurrence of an overt signal to stop (Logan, 1982). As such, it appears that skilled typists exhibit very close control over his/her typing by monitoring each individual keystroke. Therefore, controlled processing (via attention) is an essential component for on-line detection and correction of error. At higher levels of skill acquisition, controlled processing is not eliminated, but appears to become highly specialized (Ivry, 1996).  Automatic Processing Arrests Development of Skilled Performance Recently, Ericsson (2003) has argued that automatic processing actually arrests the development of skilled performance. When acquiring a task such as skiing or tennis, individuals attempt to gain an acceptable level of mastery as quickly as possible. Ericsson suggests that for most recreational levels, individuals take approximately 50 hours to progress from the cognitive phase of learning through to the autonomous stage, whereby the individual has reached a fairly stable level of performance that does not demand high levels of attention. In contrast to Fitts and Posner's three-stage conceptualization (1967), Ericsson asserts that improvements  169  will not emerge with further experience on the task. Experts, in contrast, resist performing in an automatic mode by remaining in the cognitive/associative stages. It is only when the performer deliberately attends to the continual refinement of the mental representations associated with task performance that further improvement occurs. When the individual stops engaging in deliberate practice performance will subsequently fall into automated modes of control, arresting further development of the skill. In essence, deliberate practise prevents an individual from 'premature automation' (p. 65), which prevents the learner from achieving higher levels of performance. Although attractive, this theory is not without its problems (also see Abernethy, Farrow, & Berry, 2003). For example, the tenets of this theory would predict that the expert should experience greater dual-task interference in comparison to the non-expert. Instead, empirical research, including the results of the present investigations (Experiments 1-3, & 5), demonstrate that dual-task interference decreases as a function of practice. Nevertheless, this conceptualization warrants further investigation for a better understanding of attention demands and skilled motor performance.  Attention Demands. Augmented Feedback and Bimanual Coordination One important issue for the acquisition of bimanual coordination tasks has been the finding that pre-existing coordination tendencies appear to disrupt the learning process. Therefore, a major difficulty of learning is breaking away from prepractice behaviours, as well as resisting the influence of intrinsic patterns when task conditions change. To break away from pre-existing coordination biases, individuals may use a number of information sources. Therefore, a secondary purpose of this thesis was to examine the effects of two such sources for the acquisition of a new bimanual coordination pattern. Specifically, we examined the role of continuous online visual feedback and manual guidance. Visual feedback was provided by plotting relative motion displacement plots in real-time, while manual guidance was provided via a set of servo torque motors, which moved the participants' limbs in the required phase relation, as well as in a summary playback of their own performance. In all of our experiments, availability of concurrent visual feedback was manipulated so that  170  visual feedback was either removed from the perceptual display (while performing the 90° task) or became available during a trial to guide execution of the movement. By manipulating availability of real-time relative motion feedback we were able to assess changes in attention demands while performing the goal coordination task. Across experiments, findings revealed that the 90° pattern was more accurate in the presence of visual feedback regardless of whether vision was presented in the first segment or the second segment of the trial. Importantly, when visual feedback was removed from the perceptual display during a trial, a relatively permanent phase drift occurred towards the more preferred coordination tendency of anti-phase (Experiment 1). This observed phase drift was considered relatively permanent because performance did not return to the same levels of accuracy and stability exhibited prior to removal of the feedback, even when time was provided to recover from the environmental disruption (Experiment 5). When visual feedback was unavailable to guide on-going movement in segment 1 , participants were generally unable to inhibit the influence of pre-existing tendencies until much later in learning, or in some cases, not at all (Experiment 2). However, our findings showed that using servo torque motors to physically guide the learner's limbs provided some learning benefits when visual feedback was limited and individuals were left to their own devices to learn how to produce the task (Experiment 3). With respect to attention demands, the present series of investigations demonstrated that attention demands increase whenever information feedback sources are switched. Increases in attention demands may be temporary, or more permanent depending on the nature of the switch. Providing visual feedback in the middle of the trial (after performing in its absence), only temporarily increases attentional load, while switching from visual to proprioceptive sources produces a more permanent effect. This latter finding can be explained by the fact that performance of the 90° pattern was shown to be less attention demanding in the presence of real-time relative motion feedback than when performing in its absence. A possible explanation provided for the latter observation was the use of relative motion displacement plots. It was postulated that presenting spatiotemporal information in the form of a Lissajous figure may decrease attention demands and  171  facilitate learning of the new pattern because the trace clearly specifies the task goal, and provides performers with crucial features of the movement. This alerts the individual to the properties and relationships inherent within the task in lieu of specific task instructions. This finding may be particularly important when discussing instructional aids for older adults experiencing decreased learning as a result of normal aging. For example, Wishart, Lee, Cunningham, and Murdoch (2002) have shown that elderly individuals experience large benefits from the provision of concurrent visual feedback when learning a 90° out-of-phase task. One reason for this benefit may be that attention demands of the task are reduced when visual feedback is provided. The relative motion displacement plots presents the individual with information that they do not necessarily have to find on their own, thereby reducing cognitive load and facilitating acquisition of the required pattern. Clearly, further work is warranted for understanding the relationship between attention demands and the use of different feedback sources for the acquisition of a new bimanual coordination pattern in a wide variety of populations (e.g., children, the elderly, motor disordered patients). Moreover, the findings of the present series of investigations provide support for utilizing experimental designs that observe learning across extended periods of practice, in comparison to the number of practice trials typically employed in motor learning experiments. Such experimental designs provide new insights into the acquisition of bimanual coordination patterns and how attention demands evolve across the time course of learning.  172  REFERENCES Abernethy, B. (1988). Dual-task methodology and motor skills research: Some applications and methodological constraints. Journal of Human Movement Studies, 14,101-132. Abernethy, B. (1993). Attention. In R. N. Singer, M. Murphey, & L. K. Tennant (Eds.), Handbook of Research on Sport Psychology (pp. 127-170). New York: MacMillan Publishing. Abernethy, B., Burgess-Limerick, R., & Parks, S. (1994). Contrasting approaches to the study of motor expertise. Quest, 46,186-198. Abernethy, B., Farrow, D., & Berry, J. (2003). Constraints and issues in the development of a general theory of expert perceptual-motor performance. In J. L. Starkes, & K.A. Ericsson's (Eds.), Expert performance in sports: Advances in research on sport expertise (pp. 349-369). Champaign, IL: Human Kinetics. Abernethy, B., & Russell, D.G. (1987). Expert-novice differences in an applied selective attention task. Journal of Sport Psychology, 9, 326-345. Adams, J.A. (1971). A closed-loop theory of motor learning. Journal of Motor Behavior, 3, 111-149. Adams, J.A. (1987). Historical review and appraisal of research on the learning, retention, and transfer of human motor skill. Psychological Bulletin, 101, 41-74. Ahissar, M., Laiwand, R., & Hochstein, S. (2001). Attentional demands following perceptual skill training. Psychological Science, 12, 56-62. Allard, F., Graham, S., & Paarsalu, M.L. (1980). Perception in sport: Basketball. Journal of Sport Psychology, 2, 14-21. Allport, D.A. (1980). Attention and performance. In G. Claxton (Ed.), Cognitive psychology: New directions (pp. 112-153). London: Routledge & Kegan. Allport, D.A., Antonis, B., & Reynolds, P. (1972). On the division of attention: A disproof of the single channel hypothesis. Quarterly Journal of Experimental Psychology, 24, 225-235. Anderson, J.R. (1982). Acquisition of cognitive skill. Psychological Review, 89, 369406. Armstrong, T.R. (1970). Feedback and perceptual-motor skill learning: A review of information feedback and manual guidance training techniques (tech. Rep. No. 25). Ann Arbor, Ml: University of Michigan, Department of Psychology.  173  Bain, A. (1868). The senses and the intellect (3 Ed.). London: Longmans, Green. rd  Baldissera, F., Cavallari, P., &Tesio, L. (1994). Coordination of cyclical coupled movements of hand and foot in normal subjects and on healthy side of hemiplegic patients. In S.P. Swinnen, H. Heuer, J. Massion, & P. Casaer (Eds.), Interlimb coordination: Neural, dynamical, and cognitive constraints (pp. 229242). New York: Academic Press. Bardy, B.G., & Laurent, M. (1991). Visual cues and attention demands in locomotor positioning. Perceptual and Motor Skills, 72, 915-926. Beek, P.J., & van Santvoord, A.A.M. (1992). Learning the cascade juggle: A dynamical systems analysis. Journal of Motor Behavior, 24, 85-94. Beilock, S.L, Carr, T.H., MacMahon, C , & Starkes, J.L. (2002). When paying attention becomes counterproductive: Impact of divided versus skill-focused attention on novice and experienced performance of sensorimotor skills. Journal of Experimental Psychology: Applied, 8, 6-16. Bernstein, N.A. (1967). The co-ordination and regulation of movements. Oxford: Pergamon Press. Blandin, Y., Proteau, L., & Alain, C. (1994). On the cognitive processes underlying contextual interference and observational learning. Journal of Motor Behavior, 26, 18-26. Blischke, K. (1998). Automaticity and flexibility in motor performance. In P. Blaser (Ed.), Sport Kinetics '97: Theories of human performance and their reflections in practice. Fifth International Scientific Conference of the International Association of Sport Kinetics common with the Section "Sportmotorik" of the German Association for Sport Science (dvs) and the Institute of Sport Science of Ottovon-Guericke-University of Magdeburg (pp. 297-301). Hamburg, Germany: Czwalina Verlag. Bloom, B.S. (1985). Generalizations about talent development. In B.S. Bloom (Ed.), Developing talent in young people (pp. 507-549). New York: Ballantine Books. Bourdin, C , Nougier, V., & Teasdale, N. (1997). Attentional demands of the organization of grasping in rock climbing. In R. Lidor & M. Bar-Eli (Eds.), Innovations in sport psychology: Linking theory and practice (pp. 139-141). Proceedings World Congress of Sport Psychology IXth. Bourdin, C , Teasdale, N., & Nougier, V. (1998). Attentional demands and the organization of reaching movements in rock climbing. Research Quarterly for Exercise and Sport, 69, 406-410.  174  Bredin, S.S.D., & Bouffard, M. (2000). The influence of movement skill level on performer error detection capability. Manuscript submitted for publication. Bredin, S.S.D., Hodges, N.J., Sturrock, S., Franks, I.M., & Chua, R. (2000). The influence of skill level on performer ability to judge movement outcome. Journal of Exercise and Sport Psychology, 22, S23.  Broadbent, D.E. (1958). Perception and communication. Elmsford, NY: Pergammon Press. Brown, L.A., Shumway-Cook, A., & Woollacott, M.H. (1999). Attentional demands and postural recovery: The effects of aging. Journal of Gerontology Series A: Biological Sciences and Medical Sciences, 54, M165-M171.  Brown, S.W. & Bennett, E.D. (2002). The role of practice and automaticity in temporal and nontemporal dual-task performance. Psychological Research, 66, 80-89. Bryan, W.L., & Harter, N. (1897). Studies in the physiology and psychology of telegraphic language. Psychological Review, 4, 27-53. Bryan, W.L., & Harter, N. (1899). Studies on the telegraphic language. The acquisition of a hierarchy of habits. The Psychological Review, VI, 345-375. Camicioli, R., Howieson, D., Lehman, S., & Kaye, J. (1997). Talking while walking: The effect of dual-task in aging and Alzheimer's disease. Neurology, 48, 955958. Carson, R.G., Byblow, W.D., Abernethy, B., & Summers, J.J. (1996). The contribution of inherent and incidental constraints to intentional switching between patterns of bimanual coordination. Human Movement Science, 15, 565589. Carson, R.G., Byblow, W.D., & Goodman, D. (1994). The dynamical substructure of bimanual coordination. In S.P Swinnen, H. Heuer, J. Massion, & P. Casaer (Eds.), Interlimb coordination: Neural, dynamical, and cognitive constraints.  Orlando, FL: Academic Press. Carson, R.G., Goodman, D., Kelso, J.A.S., & Elliot, D. (1994). Intentional switching between patterns of interlimb coordination. Journal of Movement Studies, 27, 201-218. Castiello, U., & Umiita, C. (1988). Temporal dimensions of mental effort in different sports. International Journal of Sport Psychology, 19,199-210.  175  Cheng, P.W. (1985). Restructuring versus automaticity: Alternative accounts of skill acquisition. Psychological Review, 92, 414-423. Chua, R. & Weeks, D.J. (1996). Dynamical explorations of compatibility in perception-action coupling. In B. Hommel & W. Prinz (Eds.), Theoretical issues in stimulus-response compatibility (pp. 373-398). Amsterdam: Elsevier Science. Cohen, L. (1971). Synchronous bimanual movements performed by the homologous and nonhomologous muscles. Perceptual and Motor Skills, 32, 639-644. Crossman, E.R.F.W. (1956). A theory of acquisition of speed skill. Ergonomics, 2, 153-166. Debaere, F., Wenderoth, N., Sunaert, S., van Hecke, P., & Swinnen, S.P. (2003). Internal vs. external generation of movements: Differential neural pathways involved in bimanual coordination performed in the presence or absence of augmented feedback. Neurolmage, 19, 764-776. Delignieres, D., Nouritt, D., Sioud, R., Leroyer, P., Zattara, M., & Micaleff, J-P. (1998). Preferred coordination modes in the first steps of the learning of a complex gymnastics skill. Human Movement Science, 17, 221-241. Deutsch, J.A., & Deutsch, D. (1963). Attention: Some theoretical considerations. Psychological Review, 70, 80-90.  Ells, J.G. (1973). Analysis of temporal and attentional aspects of movement control. Journal of Experimental Psychology, 99, 10-21.  Ericsson, K.A. (1996). The acquisition of expert performance: An introduction to some of the issues. In K.A. Ericsson (Ed.), The road to excellence: The acquisition of expert performance in the arts, sciences, sports and games (pp.1-  50). Mahwah, NJ: Lawrence Erlbaum. Ericsson, K.A. (2003). Development of elite performance and deliberate practice: An update from the perspective of the expert performance approach. In J.L. Starkes & K.A. Ericsson (Eds.), Expert performance in sports: Advances in research on  sport expertise (pp.49-84). Champaign, IL: Human Kinetics. Ericsson, K.A., Krampe, R. Th., & Tesch-Romer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100, 363-406. Ericsson, K.A., & Lehmann, A.C. (1996). Expert and exceptional performance: Evidence of maximal adaptation to task constraints. Annual Review of Psychology, 47, 273-305.  176  Evbersbach, G., Dimitrijevic, M.R., & Poewe, W. (1995). Influence of concurrent tasks on gait: A dual-task approach. Perceptual and Motor Skills, 81,107-113. Fitts, P.M. (1962). Factors in complex training. In R. Glaser(Ed.), Training research and education. Pittsburgh: Pittsburgh University Press. Fitts, P.M. (1964). Perceptual-motor skill learning. In A. W. Melton (Ed.), Categories of human learning (pp. 243-285). New York: Academic Press. Fitts, P. M., & Posner, M. I. (1967). Human performance. Belmont, CA: Brooks/Cole. Fontaine, R.J., Lee, T.D., & Swinnen, S.P. (1997). Learning a new bimanual coordination pattern: Reciprocal influences of intrinsic and to-be-leamed patterns. Canadian Journal of Experimental Psychology, 51, 1-9.  Gentile, A. M. (1972). A working model of skill acquisition with application to teaching. Quest, 17, 3-23. Gentile, A. M. (1998). Implicit and explicit processes during acquisition of functional skills. Scandinavian Journal of Occupational Therapy, 5, 7-16.  Goodman, D. (1985). Synergies and functional constraints in a theory of action. In D. Goodman, R.B. Wilberg, & I.M. Franks (Eds.), Differing perspectives in motor learning, memory, and control (pp. 319-340). Amsterdam: Elsevier Science. Guthrie, E.R. (1952). The psychology of learning. New York: Harper & Row. Hagman, J.D. (1983). Presentation- and test-trial effects on acquisition and retention of distance and location. Journal of Experimental Psychology: Learning, Memory,  and Cognition, 9, 334-345. Haider, H., & Frensch, P.A. (1996). The role of information reduction in skill acquisition. Cognitive Psychology, 30, 304-337. Haken, H., Kelso, J.A.S., & Bunz, H. (1985). A theoretical model of phase transition in human hand movements. Biological Cybernetics, 51, 347-356. Helsen, W., Starkes, J.L., & Hodges, N.J. (1998). Team sports and the theory of deliberate practice. Journal of Sport and Exercise Psychology, 20, 13-25.  Hirst, W. (1986). The psychology of attention. In J.E. Ledoux & W. Hirst (Eds.), Mind and brain: Dialogues in cognitive neuroscience (pp. 105-141). Cambridge:  Cambridge University Press.  177  Hodge, T., & Deakin, J. (1998). Deliberate practice and expertise in the martial arts: The role and context in motor recall. Journal of Sport and Exercise Psychology, 20, 260-279. Hodges, N.J. (2001). Role of instructions and demonstrations in learning a coordination task. Unpublished doctoral dissertation. University of British Columbia. Hodges, N.J., & Franks, I.M. (2000). Attention focusing instructions and coordination bias: Implications for learning a new bimanual task. Human Movement Science, 19, 843-867. Hodges, N.J., & Franks, I.M. (2001). Learning a coordination skill: Interactive effects of instruction and feedback. Research Quarterly for Exercise and Sport, 72, 132-  142.  Hodges, N.J., & Franks, I.M. (2002). Learning as a function of coordination bias: Building upon pre-practice behaviours. Human Movement Science, 21, 231-258. Hodges, N.J., & Lee, T.D. (1999). The role of augmented information prior to learning a bimanual visual-motor coordination task: Do instructions of the movement pattern facilitate learning relative to discovery learning? British Journal of Psychology, 90, 389-403.  Howe, M.J.A., Davidson, J.W., & Sloboda, J.A. (1998). Innate talents: Reality or myth? Behavioral and Brain Sciences, 21, 399-442.  Hoyt, D.F., & Taylor, C R . (1981). Gait and the energetics of locomotion in horses. Science, 292, 239-240. Ivry, R. (1996). Representational issues in motor learning: Phenomena and theory. In H. Heur & S.W. Keele (Eds.), Handbook of perception and action. Volume 2:  Motor skills (pp. 263-330). London: Academic Press. James, W. (1890). The principles of psychology (Vol. 1). New York: Holt. Janelle, CM., & Hillman, C H . (2003). Expert performance in sport: Current perspectives and critical issues. In J.L. Starkes & KA. Ericsson (Eds.), Expert performance in sports: Advances in research on sport expertise (pp.19-48).  Champaign, IL: Human Kinetics. Jeka, J.J., & Kelso. J.A.S. (1989). The dynamic pattern approach to coordinated behavior: A tutorial review. In S. Wallace (Ed.), Perspectives on the coordination of movement (pp. 3-45). Amsterdam: North-Holland. Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.  178  Keele, S.W., & Hawkins, H.L. (1982). Explorations of individual differences relevant to high level skill. Journal of Motor Behavior, 14, 3-23. Kelso, J.A.S. (1981). On the oscillatory basis of movement. Bulletin of the Psychonomic Society, 18, 63.  Kelso, J.A.S. (1984). Phase transitions and critical behavior in human bimanual coordination. American Journal of Physiology: Regulatory, Integrative and Comparative Physiology, 15, R1000-R1004. Kelso, J.A.S. (1995). Dynamic patterns: The self-organization of brain and behavior.  Cambridge, MA: MIT Press.  Kelso, J.A.S., & Jeka, J.J. (1992). Symmetry breaking dynamics of human multilimb coordination. Journal of Experimental Psychology: Human Perception and  Performance, 18, 645-668.  Kelso, J.A.S., Scholz, J.P., & Schoner, G.S. (1986). Non-equilibrium phase transitions in coordinated biological motion: Critical fluctuations. Physics Letters A, 118, 279-284. Kelso, J.A.S., Southard, D.L., & Goodman, D. (1979). On the coordination of twohanded movements. Journal of Experimental Psychology: Human Perception and Performance, 5, 229-238.  Klein, R.M. (1978). Automatic and strategic processes in skilled performance. In G.C. Roberts & K.M. Newell (Eds.), Psychology of motor behavior and sport (pp.  270-287). Champaign, IL: Human Kinetics. Kohl, R.M., Ellis, S.D., & Roenker, D.L. (1992). Alternating actual and imagery practice: Preliminary theoretical considerations. Research Quarterly for Exercise and Sport, 63, 162-170. Laberge, D., & Samuels, S.J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 3, 293-323. Lajoie, Y., Teasdale, N., Bard, C , & Fleury, M. (1993). Attentional demands for static and dynamic equilibrium. Experimental Brain Research, 97,139-144. Lee, T.D. (1998). On the dynamics of motor learning research. Research Quarterly for Exercise and Sport, 69, 334-337.  Lee, T.D., Blandin, Y., & Proteau, Y. (1996). Effects of task instructions and oscillation frequency on bimanual coordination. Psychological Research, 59, 100-106.  179  Lee, T.D., & Swinnen, S.P. (1993). Three legacies of Bryan and Harter: Automaticity, variability, and change in skilled performance. In J.L. Starkes & F. Allard (Eds.), Cognitive issues in motor expertise (pp. 295-315). Amsterdam: Elsevier Science. Lee, T.D., Swinnen, S.P., & Verschueren, S. (1995). Relative phase alterations during bimanual skill acquisition. Journal of Motor Behavior, 27, 263-274. Lee, T.D., Wishart, L.R., & Murdoch, J.E. (2002). Canadian Journal on Aging, 21, 549-557. Logan, G.D. (1982). On the ability to inhibit complex movements: A stop-signal study of typewriting. Human Perception and Performance, 3, 778-792.  Logan, G.D. (1985). Skill and automaticity: Relations, implications, and future directions. Canadian Journal of Psychology, 39, 367-386.  Logan, G.D. (1988). Toward an instance theory of automatization. Psychological Review, 95, 492-527. Logan, G.D. (1990). Automaticity and memory. In W.E. Hockley & S. Lewandowski (Eds.), Relating theory and data: Essays on human memory in honor ofBennet  B. Murdoch. Hillsdale, NJ: Erlbaum.  Logan, G.D. (1992). Attention and preattention in theories of automaticity. American Journal of Psychology, 105, 317-339.  Logan, G.D., Taylor, S.E. & Etherton, J.L. (1999). Attention and automaticity: Toward a theoretical integration. Psychological Research, 62, 165-181. Magill, R. A. (1993). Motor learning: Concepts and applications (5 Ed). Dubuque, th  IA: Wm. C. Brown.  Magill, R.A. (2001). Motor learning: Concepts and applications (6 Ed). Dubuque, th  IA: McGraw-Hill.  Magill, R.A. (2004). Motor learning: Concepts and applications (7 Ed). Dubuque, th  IA: McGraw-Hill.  Marsh, A.P., & Geel, S.E. (2000). The effect of age on the attentional demands of postural control. Gait and Posture, 12, 105-113. Marteniuk, R.G. (1976). Information processing in motor skills. New York: Holt, Rinehart, & Winston.  180  Maslovat, D. (2002). Contextual Interference: Single-Task Versus Multi-Task Learning and Influence of Concurrent Temporal Interference. Unpublished Master of Science Thesis. University of British Columbia. Mayville, J.M., Bressler, S.L., Fuchs, A., & Kelso, J.A.S. (1999). Spatiotemporal reorganization of electrical activity in the human brain associated with a timing transition in rhythmic auditory-motor coordination. Experimental Brain Research, 127, 371-381. McCullagh, P. (1993). Modeling: Learning, developmental, and social psychological considerations. In R. N. Singer, M. Murphey, & L.K. Tennant (Eds.), Handbook of research on sport Psychology {pp. 106-126). New York: MacMillan McGarry, T., Hodges, N.J., Bredin, S.S.D., Franks, I.M., & Chua, R. (2000). Stability of the scanning procedure for mapping dynamical landscapes. Journal of Sport & Exercise Psychology, 22, S78-S78.  Miller, G.A., Galanter, E., & Pribram, K.H. (1960). Plans and structure of behavior. New York: Holt, Rinehart & Winston. Monno, A., Chardenon, A., Temprado, J.J., Zanone, P.G., & Laurent, M. (2000). Effects of attention on phase transitions between bimanual coordination patterns: A behavioural and cost analysis in humans. Neuroscience Letters, 283, 93-96. Morris, M.E., lansek, R., Summers, J.J., Matyas, T.A. (2000). Motor control considerations for the rehabilitation of gait in Parkinson's disease. In D.J. Glencross & J. Piek (Eds.), Motor control and sensory-motor integration: Issues  and directions (pp. 61-93). Amsterdam: Elsevier Science. Navon, D., & Gopher, D. (1979). On the economy of the human-processing system. Psychological Review, 86, 214-255.  Nettleton, B. (1979). Attention demands and ball-tracking skills. Perceptual and Motor Skills, 49, 531 -534.  Norman, D.A., & Bobrow, D.G. (1975). On the analysis of performance operating characteristics. Psychological Review, 83, 508-519. Ogden, G.D., Levine, J.M., & Eisner, E.J. (1979). Measurement of workload by secondary tasks. Human Factors, 21, 529-548. Pashler, H. (1994). Dual-task interference in simple tasks: Data and theory. Psychological Bulletin, 116, 220-244.  Pashler, H.E. (1998). The psychology of attention. Cambridge, MA: MIT Press.  181  Patrick, J. (1992). Training: Research and practice. San Diego, CA: Academic Press. Peck, A. J., & Turvey, M.T. (1997). Coordination dynamics of the bipedal galloping pattern. Journal of Motor Behavior, 29, 311-325.  Populin, L, Rose, D.J., & Heath, K. (1990). The role of attention in one-handed catching. Journal of Motor Behavior, 22, 149-158.  Posner, M.I., & Keele, S.W. (1969). Attention demands of movement. Proceedings of the 16  th  International Congress of Applied Psychology. Amsterdam: Swets and  Zeitlinger. Prezuhy, A. M., & Etnier, J. L. (2001). Attentional demands of horseshoe pitchers at two levels of task difficulty. Research Quarterly for Exercise and Sport, 72, 293-  298.  Proctor, R.W., & Dutta, A. (1995). Skill acquisition and human performance.  Thousand Oaks, CA: Sage. Proteau, L. (1992). On the specificity of learning and the role of visual information for movement control. In L. Proteau & D. Elliott (Eds.), Vision and motor control (pp. 67-103). Amsterdam: Elsevier Science. Proteau, L, & Isabelle, G. (2002). On the role of visual afferent information for the control of aiming movements toward targets of different sizes. Journal of Motor Behavior, 34, 367-384. Proteau, L, Marteniuk, R.G., Girouard, Y., & Dugas, C. (1987). On the type of information used to control and learn an aiming movement after moderate and extensive training. Human Movement Science, 6,181-199. Puttemans, V., Vangheluwe, S., Wenderoth, N., & Swinnen, S.P. (2004). Bimanual directional interference: The effect of normal versus augmented visual information feedback on learning and transfer. Motor Control, 8, 33-50. Rabbitt, P. (1978). Detection of errors by skilled typists. Ergonomics, 21, 945-958. Rose, D.J., & Christina, R.W. (1990). Attentional demands of precision pistolshooting as a function of skill level. Research Quarterly for Exercise and Sport, 61, 111-113. Salmoni, A.W., Schmidt, R.A., & Walter, C B . (1984). Knowledge of results and motor learning: A review and critical reappraisal. Psychological Bulletin, 95, 355386.  182  Salmoni, A.W., Sullivan, S.J., & Starkes, J.L. (1976). The attention demands of movements: A critique of the probe technique. Journal of Motor Behavior, 8, 161169. Salthouse, T.A. (1991). Expertise as the circumvention of human processing limitations. In K.A. Ericsson & J. Smith (Eds.), Toward a general theory of expertise: Prospects and limits (pp. 286-300). New York: Cambridge Press. Savelsbergh, G.J.P., Williams, A.M., van der Kamp, J., & Ward, P. (2002). Visual search, anticipation and expertise in soccer goalkeepers. Journal of Sports Sciences, 20, 279-287. Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.  Schmidt, R.A. (1980). Past and future issues in motor programming. Research Quarterly for Exercise and Sport, 51, 122-140.  Schmidt, R.A. (1987). The acquisition of skill: Some modifications to the perceptionaction relationship through practice. In H. Heuer and A.F. Sanders (Eds.), Perspectives on perception and action (pp. 77-103). Hillsdale, NJ: Erlbaum. Schmidt, R.A. (1988). Motor control and learning: A behavioral emphasis (2  nd  Ed).  Champaign, IL: Human Kinetics. Schmidt, R.A., & Lee, T.D. (1999). Motor control and learning: A behavioral  emphasis (3 Ed). Champaign, IL: Human Kinetics. rd  Schmidt, R. A., & White, J. L. (1972). Evidence for an error detection mechanism in motor skills: A test of Adams' closed-loop theory. Journal of Motor Behavior, 4, 143-153. Schmidt, R.A., & Wulf, G. (1997). Continuous concurrent feedback degrades skill learning: Implications for training and simulation. Human Factors, 39, 509-525. Schmidt, R.C., & Fitzpatrick, P. (1996). Dynamical perspective on motor learning. In H.N. Zelaznik (Ed.), Advances in motor learning and control (195-223).  Champaign, IL: Human Kinetics. Schneider, W. (1985). Training high-performance skills: Fallacies and guidelines. Human Factors, 27, 285-300.  Schneider, W., & Shiffrin, R.M. (1977). Controlled and automatic human information processing: I. Detection, search, and attention. Psychological Review, 84,1-66.  183  Scholz, J.P. (1990). Dynamic pattern theory - some implications for therapeutics. Physical Therapy, 70, 827-843.  Scholz, J.P., & Kelso, J.A.S. (1989). A quantitative approach to understanding the formation and change of coordinated movement patterns. Journal of Motor Behavior, 21, 122-144. Scholz, J.P., & Kelso, J.A.S. (1990). Intentional switching between pattens of bimanual coordination depends on intrinsic dynamics of the patterns. Journal of Motor Behaviour, 22, 98-125.  Scholz, J.P., Kelso, J.A.S., & Schoner, G.S. (1987). Non-equilibrium transitions in coordinated biological motion: Critical slowing down and switching time. Physics Letters A123, 390-394. Schoner, G.S. (1989). Learning and recall in dynamic pattern theory of coordination patterns. Biological Cybernetics, 63, 39-54. Schoner, G.S., & Kelso, J.A.S. (1988). A synergetic theory of environmentallyspecified and learned patterns of movement coordination. Biological Cybernetics, 58, 71-80. Schoner, G.S., Zanone, P.G., & Kelso, J.A.S. (1992). Learning as change of coordination dynamics: Theory and experiment. Journal of Motor Behavior, 24, 29-48. Shadmehr, R., & Holcomb, H.H. (1999). Inhibitory control of competing motor memories. Experimental Brain Research, 126, 235-251.  Shea, J.B., Wulf, G., & Whitacre, C. (1999). Enhancing training efficiency and effectiveness through the use of dyad training. Journal of Motor Behavour, 31, 119-125. Sherwood, D.E. (2000). Motor control and learning. In T.J. Housh & D.J. Housh (Eds.), Introduction to Exercise Science (pp. 252-276). Boston: MA: Allyn & Bacon. Shiffrin, R.M., & Schneider, W. (1977). Controlled and automatic human information processing: II. Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127-190.  Simon, HA. & Chase, W.G. (1973). Skill in chess. American Scientist, 61, 394-403. Slife, B. D., & Williams, R. N. (1995). What's behind the research? Discovering hidden assumptions in the behavioural sciences. Thousand Oaks, CA: Sage  Publishing.  184  Smethurst, C.J., & Carson, R.G. (2001). The acquisition of movement skills: Practice enhances the dynamic stability of bimanual coordination. Human Movement Science, 20, 499-529. Smethurst, C.J., & Carson, R.G. (2003). The effect of volition of the stability of bimanual coordination. Journal of Motor Behavior, 35, 309-319. Smith, L.B., & Thelan, E. (1993). A dynamic systems approach to development: Applications. Cambridge, MA: MIT Press. Snoddy, G.S. (1926). Learning and stability: A psychophysical analysis of a case of motor learning with clinical applications. Journal of Applied Psychology, 10, 1-36. Sparrow, W.A., Bradshaw, E.J., Lamoureux, E., Tirosh, O. (2002). Ageing effects on the attention demands of walking. Human Movement Science, 21, 961-972. Spelke, E., Hirst, W., & Neisser, U. (1976). Skills of divided attention. Cognition, 4, 215-230. Spencer, H. (1881). The principles of psychology (Vol 1). London: Williams and Norgate. Starkes, J.L. (1987). Attention demands of spatially locating position of a ball in flight. Perceptual and Motor Skills, 64, 127-135. Starkes, J.L. (1993). Motor experts: Opening thoughts. In J.L. Starkes and F. Allard (Eds.), Cognitive Issues in Motor Expertise (pp. 3-16). Amsterdam: Elsevier. Starkes, J.L., Deakin, J.M., Allard, F., Hodges, N.J., & Hayes, A. (1996). Deliberate practice in sports: What is it anyway? In K.A. Ericsson (Ed.), The road to excellence: The acquisition of expert performance in the arts and sciences, sports, and games (pp. 81-106). Mahwah, NJ: Erlbaum. Starkes, J.L. (2000). The road to expertise: Is practice the only determinant? International Journal of Sport Psychology, 31, 431 -451. Stelmach, G.E., & Hughes, B.G. (1983). Does motor skill automation require a theory of attention? In R.A. Magill (Ed.), Memory and control of action (pp. 6792). Amsterdam: North-Holland. Stelmach, G.E., & Larish, D.D. (1980). A new perspective on motor skill automation. Research Quarterly for Exercise and Sport, 51, 141 -157.  185  Strayer, D.L., & Johnston, W.A. (2001). Driven to distraction: Dual-task studies of simulated driving and conversing on a cellular telephone. Psychological Science, 12, 462-466. Suinn, R. (1993). Imagery. In R. N. Singer, M. Murphey, & L.K. Tennant (Eds.), Handbook of research on sport Psychology (pp. 492-510). New York: MacMillan.  Swinnen, S.P. (1996). Information feedback for motor skill learning: A review. In H.N. Zelaznik (Ed.), /Advances in motor learning and control (pp. 37-66). Champaign, IL: Human Kinetics. Swinnen, S.P., & Carson, R.G. (2002). The control and learning of patterns of interlimb coordination: Past and present issues in normal and disordered control. Acta Psychologica, 110, 129-137.  Swinnen, S.P., Lee, T.D., Verschueren, S., Serrien, D.J., & Bogaerds, H. (1997). Interlimb coordination: Learning and transfer under different feedback conditions. Human Movement Science, 16, 749-785.  Temprado, J.J., & Laurent, M. (1999). Perceptuo-motor coordination in sport: Current trends and controversies. International Journal of Sport Psychology, 30,  417-436. Temprado, J.J., Monno, A., Zanone, P.G., & Kelso, J.A.S. (2002). Attentional demands reflect learning-induced alterations of bimanual coordination dynamics. European Journal of Neuroscience, 16,1390-1394.  Temprado, J.J., Zanone, P.G., Monno, A., Laurent, M. (1999). Intentional stabilization of bimanual coordination: A study trhough attentional load measure. Journal of Experimental Psychology: Human Perception and Performance, 25,  1576-1594. Temprado, J.J., Zanone, P.G., Monno, A., & Laurent, M. (2001). A dynamical framework to understand performance trade-offs and interference in dual tasks. Journal of Experimental Psychology: Human Perception and Performance, 27,  1303-1313. Thelen, E., & Smith, L. B. (1994). A dynamics systems approach to the development  of cognition and action. Cambridge, MA: Bradford. Thorndike, E.L. (1914). Educational psychology. New York: Columbia University. Tremblay, L., & Proteau, L. (1998). Specificity of practice: The case of powerlifting. Research Quarterly for Exercise and Sport, 69, 284-289.  186  Tsutsui, S., & Imanaka, K. (2003). Effect of manual guidance on acquiring a new bimanual coordination pattern. Research Quarterly for Exercise and Sport, 74,  104-109. Tsutsui, S., Lee, T.D., & Hodges, N.J. (1998). Contextual interference in learning new patterns of in bimanual coordination, Journal of Motor Behavior, 30, 151157. Tuller, B., & Kelso, J.A.S. (1989). Environmentally-specified patterns of movement coordination in normal and split-brain subjects. Experimental Brain Research, 75, 306-316. Turvey, M.T., Fitch, H.L., & Tuller, B. (1982). The Bernstein perspective: I. The problems of the degrees of freedom and context-conditioned variability. In J.A.S. Kelso (Ed.), Human motor behavior: An introduction (pp. 239-252). Hillsdale, NJ: Lawrence Erlbaum. Vallacher, R.R., & Wegner, D.M. (1987). What do people think they're doing? Action identification and human behavior. Psychological Review, 94, 3-15. van der Linden, D.W., Cauraugh, J.H.,6* Greene, T.A. (1993). The effect of frequency of kinetic feedback on learning an isometric force production task in nondisabled subjects. Physical Therapy, 73, 79-87. Vereijken, B. (1991). The dynamics of skill acquisition. Meppel: Kripps Repro. Vereijken, B., Whiting, H.T.A., & Beek, W.J. (1992). A dynamical systems approach to skill acquisition. The Quarterly Journal of Experimental Psychology, 45A, 323-  344.  Verschueren, S.M.P., Swinnen, S.P., Dom, R., De Weerdt, W. (1997). Interlimb coordination in patients with Parkinson's disease: motor learning deficits and the importance of augmented information feedback. Experimental Brain Research, 113, 497-508. Wallace, S.A. (1996). Dynamic pattern perspective of rhythmic movement: An introduction. In H.N. Zelaznik (Ed.), Advances in motor learning and control {pp.  155-194). Champaign, IL: Human Kinetics. Wenderoth, N., & Bock, O. (2001). Learning a new bimanual coordination pattern is governed by three distinct processes. Motor Control, 1, 23-33. Whitall, J., & Getchell, N. (1995). From walking to running: Applying a dynamical systems approach to the development of locomotor skills. Child Development, 66, 1541-1553.  187  Whiting, H.T.A. (1978). Input and perceptual processes in sports skills. In D.J. Glencross (Ed.), Psychology in Sport (pp. 22-47). Sydney: McGraw-Hill. Whiting, H.T.A., & Brinker, B. P. L. M. den (1982). Image of an act. In J. P. Das, R. F. Mulcahy and A. E Wall (Eds.), Theory and research in learning disabilities (pp.  217-235). New York: Plenum. Wickens, C D . (1980). The structure of attentional resources. In R.S. Nickerson (Ed.), Attention and performance VIII (pp. 239-257). Hillsdale, NJ: Erlbaum. Wickens, C D . (1984). Processing resources in attention. In. R. Parasuraman & D.R. Davies (Eds.), Varieties of attention (pp. 63-101). Orlando, FL: Academic Press. Winstein, C.J., Pohl, P.S., Cardinale, C , Green, A., Scholtz, L., & Waters, C S . (1996). Learning a partial-weight-bearing skill: Effectiveness of two forms of feedback. Physical Therapy, 76, 985-993. Winstein, C.J., Pohl, P.S., & Lewthwaite, R. (1994). Effects of physical guidance and knowledge of results on motor learning: Support for the guidance hypothesis. Research Quarterly for Exercise and Sport, 65, 316-323.  Wishart, L.R., Lee, T.D., Cunningham, S.J., & Murdoch, J.E. (2002). Age-related differences and the role of augmented visual feedback in learning a bimanual coordination pattern. Acta Psychologies, 110, 247-263. Wrisberg, CA. (1993). Levels of performance skill. In R.N. Singer, M. Murphey, & L.K. Tennant (Eds.), Handbook of research on sport psychology (pp. 61-68).  New York: MacMillan Publishing. Wulf, G., McNevin, N., Shea, C.H., & Wright, D.L. (1999). Learning phenomena: Future challenges for the dynamical systems approach to understanding the learning of complex motor skills. International Journal of Sport Psychology, 30, 531- 557. Wulf, G., Shea, C.H., & Whitacre, CA. (1998). Physical-guidance benefits in learning a complex motor skill. Journal of Motor Behavior, 30, 367-380. Yamanishi, T., Kawato, M., & Suzuki, R. (1980). Two coupled oscillators as model for the coordinated finger tapping by both hands. Biological Cybernetics, 37, 219225. Zanone, P.G., & Kelso, J.A.S. (1992). Evolution of behavioral attractors with learning: Nonequilibrium phase transitions. Journal of Experimental Psychology: Human Perception and Performance, 18, 403-421.  188  Zanone, P.G., & Kelso, J.A.S. (1994). The coordination dynamics of learning: Theoretical structure and experimental agenda. In S.P. Swinnen, H. Heuer, J. Massion, & P. Casaer (Eds.), Interlimb coordination: Neural, dynamical, and  cognitive constraints (pp. 461-490). San Diego: Academic Press. Zanone, P. G., & Kelso, J.A.S. (1997). Coordination dynamics of learning and transfer: Collective and component levels. Journal of Experimental Psychology: Human Perception and Performance, 23, 1454-1480.  Zanone, P.G., Monno, A., Temprado, J.J., & Laurent, M. (2001). Shared dynamics of attention cost and pattern stability. Human Movement Science, 20, 765-789. Zanone, P.G., Temprado, J.J., & Monno, A. (1999). Attentional load covaries with the stability of a coordination pattern. In M.A. Grealy & J.A. Thomson (Eds.), Studies in Perception and Action V(pp. 297-300). Hillsdale, NJ: Lawrence Erlbaum.  189  APPENDIX A REVIEW OF LITERATURE UNDERSTANDING LEARNING PHENOMENA: A FOUNDATION The literature presented and discussed within the main body of this thesis has been constrained to content directly related to the empirical investigations contained within. As such, the purpose of the following appendix chapter is to provide a more detailed description of central concepts not permitted within the selected format, as well as present concepts that are closely linked and lay the foundation (or rationale) for the preceding line of empirical investigations. The intent here is to provide the reader with an enhanced description of concepts relating to skill acquisition from two distinct theoretical approaches to motor learning. Specifically, the learning process will be discussed from both an information-processing framework, as well as from a dynamical systems view of motor learning and control. Rationale will also be provided in support of an integrative experimental approach whereby qualities of both perspectives are combined to examine the learning of complex, coordination patterns. The content of this chapter is presented in four sections. First, discussion will focus on understanding the complex nature of skilled behaviour. The second section will then examine skill acquisition from a human information processing perspective. Emphasis will be given to discussion on representational-based accounts of skill acquisition, the role of attention within these frameworks, as well as the relationship between attention and automatic processing for human motor performance. The third section will shift theoretical approaches to discuss motor learning from a dynamical systems perspective. Emphasis here will be on understanding the basic tenets and concepts of dynamic pattern theory and how the tools afforded by nonlinear dynamics can explain the acquisition of complex coordination patterns. In the fourth, and final section of this review, both theoretical positions will be brought together, and discussion will focus on the potential role of each framework for future exploration of motor learning phenomena. Specifically, an integrative approach for empirical investigation will be advocated with a particular emphasis on experimental designs that incorporate features central to both of these theoretical positions.  190  SECTION ONE The Nature of Skilled Behaviour: An Introduction "Skill consists in the ability to bring about some end result with maximum certainty and minimum outlay of energy, or of time and energy" (Guthrie, 1952).  The term 'skill' is used to communicate a quality of performance or degree of proficiency, which is often subjectively established by characteristics of performance (e.g., consistency) or how well the goal of the task was accomplished (e.g., outcome, degree of productivity) (Magill, 1993) . As this operational definition 1  suggests, skill cannot be directly observed, rather it is inferred through performance. Furthermore, skilled action does not manifest as an absolute, all-or-none phenomenon. Instead, skill is a relative term whereby individuals vary in their degree of proficiency (Patrick, 1992) and the line between what does or does not constitute skilled performance is tenuous at best (Logan, 1985). This latter statement is particularly accurate when trying to determine what constitutes motor expertise. For example, the normal individual achieves a high degree of competence and proficiency in many everyday tasks. From most perspectives, however, the cultural label of 'motor expert' is bestowed only upon those individuals who achieve excellence on tasks that are not part of the movement repertoire of the normal population (Starkes, 1993). Stated differently, "who we class as an expert depends on the relative exclusivity of the skill" (Starkes, 1993, p. 3). Moreover, this problem is further complicated in that skilled performance is considered open-ended because humans continually surpass current standards of motor performance to achieve new heights. In figure skating, for example, the inclusion of one or two triple jumps was once considered a highly competitive and demanding program at the international level. Today, a male skater does not perform a jump smaller than a triple and within his repertoire must possess 'quad' jumps if he is to challenge on the world stage.  In general, the word 'skill' may take on different meaning depending on the context of its usage. For example, in one context, one might say, "that individual is a highly skilled driver". However, in a different context, the sentence, "tracking is an essential skill for driving a car", may be the appropriate word connotation (Magill, 1993). To eliminate any misunderstanding in this thesis, the term 'skill' is used to indicate one's quality of performance; whereas, the term 'action' or 'task' will generally be used to refer to a motor skill, which is defined as voluntary physical goal-directed movement of the body and/or limb in a highly specific response sequence (Magill, 1993). 1  191  This implies that performances garnered by today's expert level sport performers may eventually be surpassed by the endeavors of future performers. Vast strides in sport performance are usually attributed to a range of factors. Some of which relate to the human element (e.g., improved levels of fitness, knowledge, and prerequisite techniques), while others are associated with advances in material technology (e.g., improvements in sports equipment design and the materials used in its construction). Consequently, high performance standards have led to an increased trend towards specialization (Ericsson, 1996). Moreover, the capabilities rendered by the learning process are specific to the task practiced (Keele & Hawkins, 1982), whereby transfer across elite domains in which the performer has no experience is limited. Only in rare instances are sport performers able to obtain an international level of performance in another field - occurring only when the activities are highly related and consistent with the individual's current and past training (Ericsson, 2003). Given these types of problems, a universal definition of what constitutes expert performance is unavailable (Salthouse, 1991). However, it is generally accepted that expert performance refers to consistently superior performance on a specified set of representative tasks for a particular domain (Ericsson & Lehmann, 1996). Perspectives of Expert Performance: Innate Talent or Practice? Outstanding performance has been a popular topic of discussion within the literature, and has been debated from two perspectives. The first view suggests that individual differences reflect innate capacities or attributes, also referred to as 'talents' or 'gifts', which cannot be modified by practice. Although instruction and practice are necessary to reach expert levels of performance, a talent-based approach postulates that these factors are not the primary antecedents for attaining outstanding performance. Instead, stable, genetically determined factors limit the level of performance achieved by the individual. A central tenet of this perspective is also the notion that innate talent in a particular domain can be detected at an early stage, which provides a basis for predicting future success. As a result, proponents of this perspective have traditionally focused empirical research on the early  192  identification of talent in various types of activities (see Howe, Davidson, & Sloboda, 1998, for a review). Although the talent account of expert performance has been at the centre of much discussion, little empirical evidence has surfaced to support the 'nature' argument. In contrast, a second, more contemporary perspective has emerged at the opposite end of the continuum, which suggests that skilled performance is primarily mediated by the accumulated duration of systematic training, and arises irrespective of genetic factors or hereditary predispositions (Janelle & Hillman, 2003). According to Ericsson and colleagues (see 1993, 1996, 2003, for examples of perspective), empirical evidence clearly demonstrates that a wide range of human characteristics, including both anatomical and physiological constraints, as well as cognitive and perceptual-motor skills, are modifiable as a result of extended and appropriately designed practice. It is well documented, for example, that heart size, number of blood-supplying capillaries, and metabolic properties of critical muscles change because of long-term, highly intense practice. Findings also reveal that anatomical and physiological adaptations are especially dramatic at young ages when the body is still at the early stages of its development. Intense training leads to specific stimulation of the body far beyond the normal range encountered in everyday living. For example, exposure to early and appropriate training increases the turnout of young classical ballet students (Ericsson & Lehmann, 1996). In fact, the only genetic exception to the 'nurture argument' has been height and body size, which appears to constrain an individual from reaching high levels of performance in such activities as basketball and gymnastics. According to a strict practice approach, individuals do not require innate talent to achieve expert levels of performance. Rather, differences in early experiences, preferences, opportunities, habits, training and practice are the real determinants of excellence (Howe et al., 1998). Given these factors, a central tenet to this perspective is the idea that everyone has the potential to achieve expertise. Evidence that provides support for the causal role of practice in attaining expert performance shows that individuals cannot attain exceptional performance on a task without extended appropriate practice. Furthermore, when randomly selected  193  individuals engage in training, improvements in performance are a direct result of the experimental protocol (Ericsson, 2003). In response to the emergence of an extreme 'nurturist' perspective, some researchers advocate a more 'interactionist' approach to expert performance. From this perspective, it is postulated that expert performance is mediated by a combination of both genetic and environmental factors. Although practice plays a critical role in the development of expertise, they do not deny the influence of innate hereditary factors on level of acquirable expertise (Janelle & Hillman, 2003). The preceding line of investigations sought to explore changes in attention demands as a function of extended practice. Therefore, the rest of this review of literature will focus on the role of practice in skill acquisition; rather than on innate talents of a performer. Acquiring Task Proficiency To achieve excellence, individuals must expend considerable time and effort in an attempt to develop an action behaviour that is proficient, effective, and remarkably consistent in nature (Schneider, 1985). For example, Anderson (1982) asserts that the significant acquisition of cognitive skill requires at least 100 hours of training if an individual is to perform with a reasonable degree of proficiency. Moreover, many complex skills require years of intense practice before a high level of proficiency is ever attained. Using chess play as one such example, Simon and Chase (1973) have reported that achieving the highest level or rating of skill (i.e., Grandmaster) requires no fewer than nine or ten years of intense and extensive preparation. Work by Bloom (1985) and Ericsson, Krampe, and Tech-Romer (1993) have provided support for the generalizability of this '10-year rule', which appears to hold for a wide variety of other skills in both the cognitive (e.g. mathematics) and motor domains (e.g. swimming). It should be noted, however, that many individuals spend considerably longer than 10 years in attaining international levels of performance. In addition to length of time, a critical factor in achieving a level of expertise is the type of practice an individual engages in. To achieve superior levels of  194  performance, Ericsson and colleagues maintain that the learner must undertake a form of practice called deliberate practice (see Ericsson, Krampe, & Tech-Romer, 1993 for original discussion). In contrast to play, deliberate practice is a highly structured activity, requiring a large commitment of time and energy, as well as other resources (e.g., access to instructors, materials, or training facilities). Strictly motivated by its instrumental value for improving performance (rather than immediate extrinsic reward or financial gain), deliberate practice is not viewed as an inherently enjoyable activity in general. Interestingly, research in the sport domain has shown evidence suggesting that athletes tend to deem a variety of practice activities as enjoyable (e.g., mat work in wrestling, on-ice training in skating) (see Helsen, Starkes, & Hodges, 1998; Hodge & Deakin, 1998; Starkes et al., 1996). However, sport performers will still classify practice activities as less enjoyable when comparing them with such everyday activities as sleep and leisure (Starkes, 2000). According to Ericsson (1996), athletes who express a positive assessment of enjoyment, are usually referring to the social nature of sport, and not necessarily responding to the structured elements of the practice environment. Another characteristic of deliberate practice is that large amounts of concentration and effort are required on the part of the individual to maximize improvement gains. Therefore, an effective level of daily practice is only sustainable if the individual is limited to a time-frame that allows the performer to adequately recover between sessions. As such, reaching expert levels of performance is not entirely dependent on the number of practice hours. Instead, it is the structure and content of the practice sessions that is of paramount importance. Moreover, Ericsson et al. (1993) propose a monotonic relationship between expert performance and deliberate practice, whereby increases in hours of deliberate practice directly translates into associated levels of improvement in performance. Therefore, cumulative hours of deliberate practice should be a good predictor of the level of performance attained by the individual.  195 Investigating Practice In essence, transition from inexperienced to skilled states of behaviour largely manifests as a function of practice or experience. By way of practice, individuals acquire a set of processes that leads to a persistent or permanent change in their capability for performance. Known as the field of motor learning, researchers interested in examining the learning process, are concerned with understanding the nature of this change. Therefore, the questions that give rise to empirical investigation typically ask: what has the performer learned; what types of practice enable and facilitate the acquisition process; and what are the time-periods over which change occurs (Hodges, 2001)? Interestingly, these questions are common to two different theoretical perspectives. The first discusses motor behaviour from an information-processing framework of human performance, while the second is known as the dynamical systems approach. A principal feature of theories derived from an information-processing premise (e.g., feedback-based learning theories and motor programming theories) is the existence of centrally stored internal representations, acquired as a function of practice, which are used as a basis for regulating and controlling desired actions of the performer. From this perspective, traditional research paradigms from cognitive psychology are applied to the motor domain in an effort to understand the acquisition of skilled behaviour. In contrast, theories based on the more radical dynamical systems approach, reject the assumption of central representations. Dynamical systems theorists argue that behaviour results from the interaction of multiple system components (e.g., the interaction between body parts, and between body parts and the environment) whereby no one system component prescribes the action of the performer. Because of this interplay of forces and mutual influence between system components, the motor system spontaneously self-organizes towards preferred states or patterns of behaviour that arise under certain conditions. Using mathematical tools from nonlinear dynamics, this approach to motor learning seeks to understand how both coordination and control of the body are achieved in providing solutions to a new movement problem (Vereijken, 1991).  196  SECTION TWO Skill Acquisition from an Information-processing Perspective Prior to the 1950's, the prevailing paradigm that guided scientific research in psychology was the behaviourist perspective, whereby scientists directed their line of inquiry to investigations pertaining to the relation between environmental information (the stimuli) and resultant action (the response). With the emergence of cognitive psychology in the 1950's, a shift in research focus towards an informationprocessing framework occurred, where scientists now concerned themselves with processing events that occurred in the "black box" of the mind. That is, the processing events that occur between the presentation of a stimuli and the execution of an individual's motor response. According to proponents of the information-processing model of human behaviour, mental functioning is considered analogous to the operations of a computer. Individuals take in information from the environment, this information is stored in various memory systems to be processed, and finally, after a complex chain of processing activities, a response or output is made. Such an approach suggests that a performer's behaviour is mediated by transactions of information between the individual and the environment (Whiting, 1978). Therefore, humans are not only independent of their environment, but act as active processors of information (Slife & Willliams, 1995). Emphasizing "mental operations that precede a motor act, rather than the act itself (Goodman, 1985, p. 319), theoretical consideration is directed toward the cognitive activities that underlie motor behaviour. Inherent within the information-processing framework are several key assumptions. First, it is assumed that a number of processing stages exist between the presentation of a stimulus and the production of a response, and each stage can be identified. Second, each stage represents an operation, and therefore, requires time for execution. Third, operation at each stage can only be conducted on information that has been made available for processing. Finally, information is transformed at each stage and is then passed on to the next processing stage (Goodman, 1985). Based on these assumptions, it is suggested that cognitive  197  processes sub-serve action by modifying and selecting the information that gets into the system, as well as the information that passes through each processing stage. As such, this approach assumes a top-down 'executive level' of functioning, whereby control of movement is carried out by what has traditionally been referred to as a 'homunculus' or "little man inside the head" (Turvey, Fitch, & Tuller, 1982, p. 239). Such a perspective gives rise to models of movement that are both prescriptive and hierarchical in nature. A typical information-processing model of human performance (presented in Figure A1) postulates that there are at least three intervening stages mediating information, which are based upon perceptual, cognitive, and motor processes, respectively. The perceptual mechanism is responsible for identifying and classifying incoming information from the environment, as well as transmitting a description of the environment to the decision mechanism. The function of the decision mechanism is to then formulate a course of action based upon the description of the environment, the individual's past experiences, as well as his or her current objectives. Once a course of action has been chosen, the effector mechanism receives a sequence of commands. A suitable muscle response is then organised and muscle commands are transmitted to the musculature for execution of the response. Both intrinsic and extrinsic forms of feedback play important roles within the information-processing model in that it allows for the detection and correction of movement error since feedback information may be fed back into the system by way of the effector and perceptual mechanisms (Marteniuk, 1976). Any model that seeks to describe human information processing must take into consideration the mediating influences of attention and memory in cognitive processing. For example, there is a tendency to view attention as the executive mechanism or "unitary source of control" (Stelmach & Hughes, 1983, p. 80) modifying and selecting what information gets into the system, as well as what passes through each processing stage for the accomplishment of perceptual, cognitive, and motor aspects of the task (Klein, 1978). As a consequence of selective stage processing, individuals often respond to the presentation of identical stimuli in a rather diverse manner. The performer not only has to consider the  MD  •3  •a CD  o ccj O  s <D O  OH  0^  60  1  199  current circumstances for the proficient execution of a movement, but he or she must also retrieve information from long-term memory to plan and execute the movement. Control of motor performance is achieved by accessing an existing motor plan (or program) from memory, or by synthesizing a new plan from various subroutines in memory (Goodman, 1985). As such, cognitive theory for human performance mandates that (a) "some kind of symbolic representation of the intended movement is necessary to allow the mind, body, and environment to interact", and (b) "the body is essentially subservient to the mind" (Abernethy, Burgess-Limerick, & Parks, 1994, p. 188). Representational-Based Accounts of Skill Acquisition The term 'representation' (see italics at the end of the preceding section) refers to the psychological structures and processes that allow an individual to utilize past experience for the improvement of motor behaviour (Ivry, 1996). Therefore, the information processing perspective posits that motor learning results from comparisons between an individual's memory representation of the movement and the actual result produced by the performer's action. Such comparisons serve to develop, refine, and strengthen a representation, which ultimately leads to a persistent or permanent change in resultant behaviour over time. Classic Theories of Skill Acquisition A number of representational-based accounts of skill acquisition have appeared within the motor control and learning literature (e.g., Gentile, 1972,1998; Whiting & den Brinker, 1982). The most commonly cited framework is Schmidt's (1975,1980) schema theory. Specifically, the term schema refers to an abstract memory representation of an event or action, which can also be thought of as a rule, concept, or generalization. Using the basic idea of the schema, Schmidt proposed a theory of skill acquisition, which has played a central role in theoretical development of the field (Schmidt & Lee, 1999).  200  Schmidt's (1975, 1980) Schema Theory. Upon the execution of a movement,  Schmidt (1975,1980) posits an individual can store four types of information: 1) the initial conditions of the movement; 2) the specifications or parameters of the response; 3 ) the outcome of the response; and 4) the sensory consequences of the action. From this information, two movement representations, or abstractions, develop in memory. Referred to as schemata, these abstractions are responsible for initiating the movement, as well as evaluating its success during and/or after its completion. The first schema, identified as the recall schema, enables the performer to select and initiate the required parameters of the movement (i.e., the motor program). Motor programs (acquired by learning) control movement execution, much like software programs in computers. The second relationship, the recognition schema, is responsible for evaluating response-produced feedback and informing the performer about the direction and amount of response error. Using the information stored in memory, the performer estimates and compares expected sensory consequences of an action to actual response-feedback. Any discrepancy between the expected and actual sensory consequence produces an error in movement response giving rise to a sensory signal that is fed back to the schema so that adjustments can be made by the performer to reduce the error to zero. As such, the recognition schema acts as a reference mechanism, enabling on-line and off-line error detection and correction, which is largely determined by movement duration. To account for rapid ballistic movements, schema theory suggests that a movement can be executed without the use of feedback control processes. In this manner, it is postulated that the performer executes the task using open-loop processes or motor-program control, whereby movements are programmed in advance since there is not enough time to modify the movement on-line. In tasks of ample duration, processes operating in a closed-loop fashion may control movements. Closed-loop control allows the individual to use sensory feedback to modify the movement during its execution. However, it is also postulated that movements very long in duration can also be controlled by open-loop control after an extended period of practice. At early stages of learning, the individual lacks an adequate motor program for carrying out the action. With practice, the individual is  201  able to establish an abstract representation of the action using feedback-based control processes. Eventually, the system acquires a motor program capable of running off long sequences of action with little cognitive effort (Schmidt, 1987). In essence, a highly skilled performer, equipped with a strong motor program, gains the capability to shift from a feedback-based mode of control to a programmed mode, whereby movement is produced without regard to sensory information when activated (Schmidt & Lee, 1999). According to Schmidt, control flexibility is advantageous in that the performer can direct needed attention towards higher-order aspects of the movement since the action itself can be performed without interfering with other cognitive information-processing tasks (Schmidt, 1980, 1987). In addition, the individual can save effort when accuracy requirements of a task are not critical, or the costs of committing an error are minimal. This statement is based on the assumption that feedback-back based modes of control require attention, and attention facilitates accuracy on a task. Therefore, due to an inherent trade-off between accuracy and attentional costs, it is postulated that skilled performers can shift from one mode of control to the other depending on the accuracy demands of the task (Schmidt, 1980).  Phases of Skill Acquisition Inherent to many representational-based accounts of learning is the notion that skill acquisition proceeds in phases, whereby performance characteristics of the individual are observed to change in a qualitative way. Acceptance of a 'phase approach' to skill acquisition has laid the foundation for much of the empirical work and theoretical development that has occurred within a variety of domains (e.g., instructional pedagogy). Generally, theoretical discussion has focused on two (e.g., Adams, 1971; Gentile, 1972) or three (e.g., Anderson, 1982; Fitts & Posner, 1967) distinct stages of acquisition, although empirical investigation typically focuses on the early, initial acquisition stages when a learning paradigm is employed (Adams, 1987). Common to each of these perspectives, is the idea that during the initial phase of acquisition, learners are first preoccupied with getting the idea of the  202  movement. After realizing the general requirements of the task, practice then advances to where the individual is concerned with skill refinement. Finally, a few individuals will advance to an even higher level of practice, characterized by performance that appears to be executed with little or no effort on the part of the performer (Wrisberg, 1993). Embedded within these frameworks is the assumption that one mode of processing dominates early phases of learning and another mode of processing governs movement behaviour late in skill development. To demonstrate this latter notion, both a two- and three-phase framework of skill acquisition will now be presented. These include Adams (1971) closed-loop theory of motor control and Fitts' (1962,1964; Fitts & Posner, 1967) three-phase theory of acquisition, respectively. Adams' (1971) closed-loop theory of motor control is a framework derived from the influences of cybernetics and servo-control mechanisms. Although its primary emphasis is on the role of feedback in skill acquisition and the development of error detection capabilities, it is also a theory of verbal-motor learning. Central to this perspective is the idea that humans regulate their movement at initial stages of learning through the covert use of verbal responses. This theory of motor control and learning was formulated in criticism of Thorndike's (1914) Law of Effect , a major influence of the time in psychology, which 2  implied "an automatic, non-cognitive nature of learning" (Adams, 1971, p. 115). In contrast to Adams, Fitts (1962, 1964; Fitts & Posner, 1967) adopted a three-phase theory of skill acquisition. Proposed over thirty years ago, this theory has played a highly influential role within the motor learning literature. Although this framework was strictly based on observations and interviews with instructors from a wide variety of motor domains, it has a common sense appeal, which has led to its ubiquitous use. Thorndike's (1914) Law of Effect, often restated as the principle of reinforcement, is an empirical statement, which claims that learning is related to the presence of rewards or punishments following a response. If a response is closely followed by a rewarding event, the response will be repeated. If a response is closely followed by a punishing event, elimination of that response will follow. Although Thorndike began his work with animals, he went on to extend the use of the Law of Effect to human learning in the classroom and laboratory setting. However, extending the Law of Effect in this manner assumes that laws governing animals also govern learning in humans (see Adams, 1971 for arguments in criticism of this assumption). 2  203  Adams' (1971) Verbal-Motor and Motor phases of Learning. According to  Adams (1971), movements are made by comparing ongoing feedback from the limbs during the execution of a movement to an internal representation (or reference of correctness) that is developed as a result of practice. The formulation of the internal reference of correctness is based on the notion that sensory information available during the execution of a movement leaves a trace or "image" in the central nervous system (CNS) that is stored by the learner from each movement response. As a learner becomes more proficient at a skill, discrepancies between the 'perceptual trace' and criterion will decrease and the distribution of accumulated traces will gradually represent the feedback qualities of the criterion response. Movement is subsequently guided or directed by making comparisons between the sensory information arising from the movement response (i.e., response-produced feedback) and the perceptual trace of the desired outcome. If current responseproduced feedback does not match the perceptual trace, a discrepancy or error is detected, and a discrete correction is executed on the part of the performer. This error analysis continues until no discrepancies are detected between the responseproduced feedback and the perceptual trace. In this manner, movement is regulated by an error nulling process, and successful performance centers on the quality of the perceptual trace. Adams suggests that knowledge of results (KR) plays a fundamental role in the development of the perceptual trace. Information about the outcome of each trial strengthens the trace because it is used to drive the movement closer to the criterion. That is, the distribution of traces laid out early in learning if KR is provided should represent a closer approximation of the correction response then if KR was not present. Using KR implies that the learner is an active participant in the learning process, whereby the information content of the KR (and verbal transforms of it) is covertly used to regulate movement at initial stages of acquisition. Wanting to emphasize the cognitive nature of learning, Adams labelled this developmental phase as the 'verbal-motor stage' of skill acquisition.  204 When the perceptual trace is adequately developed (i.e., the error reported in KR is minimal over a long period of time), Adams suggests that learners possess the capability for recognizing their own performance errors without a reliance on external sources of information. In fact, at this phase of acquisition, it is postulated that performers can ignore KR and learning can continue to occur even in its absence. As Adams has stated, "the specifications for the correct response are within him [the performer] in the form of a strong perceptual trace and, in relying on it for movement, he can make the correct response over and over and strengthen the perceptual trace even more" (p. 125). This form of learning has been referred to as subjective reinforcement and is illustrative of Adams final 'motor stage' of learning. Most simply, the learner passes from a 'verbal-motor' stage of movement production to a purely 'motor' stage when the learner no longer relies on mediation by thought or augmented feedback. Knowledge of results can be withdrawn from use since the capability to detect and correct errors is based on an internalized reference of correctness. Moreover, Adams supports the idea that passage from the verbal-motor phase to motor phase of learning implies that conscious behaviour has become automatic. At initial phases of learning, movement is regulated by the covert use of verbal responses (or cognitive activity), while late phases of skill acquisition are characterized by an increased adeptness in serial regulation of movement (Lee & Swinnen, 1993). Fitts' Three-Phase Theory of Skill Acquisition. Fitts and colleague (1962,  1964; Fitts & Posner, 1967) proposed that the process of acquiring skill involves a progression through three phases: the cognitive phase, the intermediate or associative phase, and the autonomous phase. These phases overlap and the transition from one to the other is continuous, marked by gradual, rather than abrupt, transitional shifts. In fact, any distinctions made between these phases are made somewhat arbitrarily and merely represent different aspects along the learning continuum. In the initial phase of acquisition, the learner is presented with a to-be learned task. According to Fitts, the learner's primary focus of attention is directed towards  205  understanding the nature or expectations of the task, identifying how the task should be performed, as well as involved in the formulation of performance strategies based on strategies developed from previously learned tasks. As the learner determines how best to attempt the first few preliminary trials, it is necessary for the individual to attend to cues, events, and responses that go unnoticed in later stages of learning. It is also at this phase that the learner becomes familiar with the formal procedures (e.g., rules) governing the task. As such, a considerable amount of cognitive activity (or effort) is required on the part of the learner, hence the designation 'cognitive phase' of learning. Inconsistency and gross errors in performance are evident within this acquisition phase as the learner attempts to generate responses that crudely approximate the desired form of the behaviour by trying a variety of ways to solve the movement problem. At this point, Fitts and Posner (1967) state that, "behaviour is truly a patchwork of old habits ready to be put together into new patterns and supplemented by a few new habits" (p. 12). Performance gains are very dramatic, and are generally larger than the performance gain(s) at any other single period of time within the acquisition process. Once the learner is able to identify the perceptual and response components of a task, the learner begins to link common sets of stimuli to common sets of responses in a probabilistic manner. This behaviour marks a transition to the intermediate or associative phase, which is characterized by the learning of cognitive "sets". According to Fitts, a cognitive set signifies advanced preparation on the part of the learner for the contingencies, which distinguish a particular situation. Procedures learned in the early stage of acquisition have been attempted, and movement patterns begin to emerge as individual components of the skill become pieced together. The learner's focus becomes directed towards 'smoothing out the skill' by making adjustments to the pattern. These small changes result in a more effective movement pattern evidenced by fewer errors and increased consistency in performance. Errors that do occur manifest as a result of grossly inappropriate subroutines, wrong sequences of acts, or as responses to incorrect cues. Duration of this stage is dependent on the complexity of the skill and the extent to which it requires integration and the formulation of new subroutines. For highly complex  206  skills, the learner is usually engaged in this phase for a longer period of time when compared to the overall time duration of the cognitive phase. An example of the time frame that Fitts (1962) suggested for this phase of learning is demonstrated in the following statement: "In the case of a typist it would extend from the point at which the student has learned the position of the different keys and how the fingers are used in striking them to the point where he/she has perhaps graduated from his/her first typing course and reduced his/her errors to less than 1 per cent and has acquired a fair degree of typing speed" (p. 188). Passage into the final phase of acquisition, is marked by relatively few errors in performance as well as a high degree of performance consistency. At this stage, large chunks of behaviour have been 'programmed' and appear to be executed without direct monitoring or sensory control. That is, component processes have become increasingly autonomous, hence the label 'autonomous stage', whereby attention requirements are diminished or lessened and the individual executes most of the skill without thinking about it. Due to the lack of conscious awareness, automatized skills are executed without much verbalization; and, in fact, Fitts and Posner (1967) go on to suggest that overt verbalization may actually interfere with the performance of highly developed skills. Within this framework, the acquisition of automated behaviour is fundamental in that it affords the individual a number of advantages. According to Fitts, automatized skill components require less cognitive processing on the part of the learner. Therefore, the performer can execute the skill while new learning is in progress. This means that the speed and efficiency with which some skills are performed continues to improve even during the final stages of skill acquisition, although improvement occurs at a continually decreasing rate. If performance should level off, Fitts postulates that this may very well be attributed to the effects of physiological aging and/or a result of a loss of motivation rather than a true learning plateau or limit in the individual's capacity for further improvement. Furthermore, automatization reduces the information-processing load of the performer and  207  permits the individual to focus on other perceptual and cognitive aspects of the task for optimal performance (e.g., the planning of strategy). Therefore, as automaticity is acquired, the learner also appears to gain a capability to concurrently perform the automated task at the same time as other ongoing activities or environmental distractions. For example, while executing a well-practised task such as driving, many individuals attempt to drive, change the radio channel, and talk to a passenger, all at the same time. At initial phases of learning, a considerable amount of cognitive effort is required on the part of the learner. In contrast, late stages of skill acquisition are characterized by an increased adeptness for parallel processing (Lee & Swinnen, 1993). Summary An information-processing approach for human performance mandates the existence of a memory representation, which allows the mind, body, and environment to interact for the successful execution of a motor task. As a consequence, two- and three-phase representational-based frameworks of skill acquisition have emerged within the literature. These frameworks serve to describe how the memory representation of an event or action is developed, refined, and strengthened over time. Embedded within these models is the assumption that one mode of processing dominates early phases of learning and another mode of processing governs movement behaviour late in skill development. Specifically, these frameworks assume that learning progresses in phases that can be differentiated on the basis of the cognitive activity associated with the production of the response. Early in learning, a considerable amount of cognitive activity is required on the part of the learner, which diminishes as a function of practice. According to this perspective, diminishing demands on attention is advantageous to the performer in that processing load is reduced, thereby permitting the individual to focus on higher-order aspects of performance.  208  Investigating the Role of Attention in Motor Performance At high levels of acquisition, motor behaviour is thought to become automatic whereby a shift from the conscious and deliberate allocation of attention to the noncognitive regulation of an ongoing movement occurs (e.g., Adams, 1971). At the advanced phase, performers also demonstrate a capability to perform two tasks simultaneously without a decrement in performance (e.g., Fitts & Posner, 1967). As such, understanding the role of attention and its regulation is paramount to understanding the nature of human skill and the development of competence and expertise. In fact, most operational definitions make the assumption that attention is a necessary prerequisite for the performance of a voluntary act (Stelmach & Larish, 3  1980). Present day interest in attention arose during the Second World War as a result of the practical problems of skilled performance and its breakdown under conditions of 'cognitive load'. This interest was largely enabled by a number of coinciding events of the 1950s, which included the development of computers, the concomitant introduction of information transmission theory, and the introduction of experimental methodologies and techniques for analysing directly observable mental processes (Allport, 1980). As a result of these events, cognitive scientists endeavoured to examine the nature of human performance with a direct emphasis on discovering the attentional limitations of the processing system. In general, the study of attention can be broken up into two areas of study. The first area pertains to the study of the mechanisms, which allow an organism to select one object or message over other messages and is referred to as selective attention. The second area of study is concerned with examining the limitations on the amount of information an organism can select for processing at any given period of time (Hirst, 1986). Referred to as divided attention, discussion will be constrained to this area of study due to the nature of this dissertation.  The phrase 'voluntary act' is an important criteria in that our discussion is limited to only tasks requiring voluntary physical movement of the body and/or limb in a highly specific response sequence to accomplish the goal of an action or task (Magill, 1993). 3  209  The Study of Divided Attention When observing the passing of everyday life, it becomes apparent that human behaviour is not constrained to the performance of only one task at a time. Rather, individuals often display the capacity to effectively perform two tasks simultaneously. For example, we are able to walk and talk at the same time, as well as listen to music while reading. However, performing two tasks concurrently can also prove difficult and the quality of performance is often less than if each task had been performed solitarily. While we can talk while walking, we cannot talk while reading, nor can we problem solve while listening (example from Hirst, 1986). Often interference between two tasks arises because the tasks are physically incompatible (e.g. typing this page and eating a sandwich with your right hand) or intellectually demanding (e.g., studying for an exam while trying to keep track of the number of votes at tribal council on Survivor). Unfortunately, the reason(s) for interference between many other task pairs remains less obvious (Pashler, 1994). Within the laboratory, investigation of dual-task performance has utilised a wide range of tasks, differing greatly in degree of complexity. For example, task complexity has ranged from examination of simple reaction time (SRT) to real-world activities like dictation (Pashler, 1994). Explanations of dual-task interference have varied, with three of the most influential theoretical approaches from psychology being: bottleneck (or task-switching) models, cross-talk models, and capacity sharing (or resource) models. Bottleneck models suggest that the human information-processing system is limited in that selection of two independent tasks is impossible when the tasks require the same mechanism for processing. When the system is dedicated to the operations of one task, a bottleneck is produced, and the processing of the other task must be postponed or otherwise impaired. Bottleneck models have been widely investigated in the psychological literature; however, they are not a central theoretical approach to this dissertation. Therefore, further discussion falls beyond the scope of the present review. Rather than concentrating on the type of operation to be carried out, crosstalk models emphasize degree of similarity between tasks, or the similarity of the  210  information being processed in each task. In principle, it is postulated that the parallel execution of two tasks comprised of similar content can facilitate dual-task performance if the same mechanism is required for processing. Once one processor is active, it should be more efficient to use the same processor than activating two different processors for concurrent performance. Unfortunately, theorists generally oppose this view; favouring the idea that similarity of content actually intensifies interference between concurrent tasks (Pashler, 1994,1998). The third and most widely accepted theoretical explanation of dual-task interference is the one proposed by resource or capacity-sharing models. Since the empirical investigations in this dissertation followed the theoretical tenets of resource theory, an expanded discussion regarding this framework is necessary. Capacity-sharing (or resource) models. Resource theories assume that people share a limited pool (or pools) of processing capacity (e.g., Kahneman, 1973; Navon & Gopher, 1979; Wickens, 1980, 1984). When two tasks operate in parallel, this limited resource capacity may be partitioned in a flexible and voluntary manner. Efficiency then is dependent on the amount of resources allocated to each task. As the demands of a task increase, fewer resources become available for other tasks. When two concurrent tasks tap into the same resources and/or joint demands exceed the total capacity of available resources, a dual-task decrement incurs in one, or in both of the simultaneously performed tasks - a behavioural outcome referred to as interference. For example, given tasks A and B, interference may be demonstrated by: a) a deterioration in the performance of B, while A remains relatively unaffected; b) a decline in A, while B remains unaffected; c) worsening of performance in both Tasks A and B; or even more drastic, d) complete prevention of B during the execution of A (Schmidt & Lee, 1999). If the performance of one task deteriorates when combined with another task, theory posits that both tasks are attention demanding.  211  Some capacity theorists argue there is a single, general-purpose capacity for attention (Kahneman, 1973). Resources reside within this single undifferentiated 4  reservoir, which are voluntarily allocated in graded quantity to any act or stage of processing dictated by a higher-level executive program. One important aspect of Kahneman's (1973) model is his contention that resource allocation depends on the state of the organism (e.g., level of physiological arousal). Since Kahneman views attention as a 'pool of effort' placing demand on mental resources, he suggests that an organism's level of excitability (i.e., arousal level) has a distinct impact on the availability of resources. If, for example, the arousal level of an individual is low, there will be a decrease in available attentional capacity when compared to resource availability at higher states of arousal. As a result, Kahneman often uses the terms capacity, attention, and effort interchangeably. Other capacity theorists have argued that attention should not be conceptualized as a single resource. Rather, attention should be viewed as a multiple pool of resources, each with its own capacity and capability to handle certain types of information (i.e., structure specific resources). According to this perspective, task interference occurs because two tasks call upon the same reservoir of resources. If there is no overlap in the resources demanded by either task, then the individual should be able to perform both tasks in parallel. As such, attention can be simultaneously devoted to separate stages of processing (e.g., Allport, Antonis, & Reynolds, 1972; Navon & Gopher, 1979; Wickens, 1980, 1984). To lend viability to multiple resource theory, an important direction has been to establish the exact nature of these proposed resource pools. The most extensive work in this area has been conducted by Wickens (1980, 1984) who suggests that the structural composition of resource reservoirs can be characterized on at least four dimensions: (1) stimulus characteristics (visual and auditory), (2) internal codes (visual and verbal), (3) response characteristics (manual and speech), and (4) levels of processing (shallow and deep). Multiple resource theory then, holds important  Within the literature, the term 'resource' has been defined in a number of ways. For example, Norman and Bobrow (1975) have defined resources to include "processing effort, the various forms of memory capacity, and communication channels" (p. 45), while others have identified resources as "units, channels, and facilities" (Navon & Gopher, 1979, p. 233). 4  212  practical implications in such fields as human factors. For example, when designing a system, it suggests that a human factors engineer should contemplate a design that minimizes the overlap of demands on common resources for tasks that will be performed concurrently. Such consideration serves to not only increase the information processing capabilities of the human operator, but also reduces the potential for human error by distributing demands across resources (Wickens, 1984). From this perspective, attention is defined in terms of the interference between concomitant performance of two tasks. Using experimental protocols designed to overload the system, resultant patterns of interference provides investigators insight into understanding the nature of underlying control mechanisms of different tasks, as well as a general indication of the attentional capacity available to different individuals for processing information (Abernethy, 1993). Assessing Attention Demands Following the logic presented in the preceding section, one way to assess attention demands of a given task, is to engage participants in the simultaneous performance of two tasks - a basic task, which is referred to as the primary task, and a secondary task. The primary task is the task for which attention demands are assessed, while performance on the secondary task is used to derive attention demands required by the primary task. An important methodological constraint of this approach, referred to as the secondary-task paradigm , is to employ an 5  instructional set, which requires participants to give attentional priority to the primary task so that a given level of performance can be maintained during concurrent performance of the subsidiary task (Ogden, Levine, & Eisner, 1979). Performance measures obtained in the dual-task condition are then compared to levels of performance obtained when the secondary task is performed in isolation (i.e., under The terms secondary-task and dual-task paradigms have often been used interchangeably within the majority of published literature (Abernethy, 1988; 1993). However, Ogden et al. (1979) have made a distinction between the terms based on the use of different instructional sets. Participants are instructed to either afford processing priority to the primary task or instructed to perform both tasks to the best of their capability. As such, it is recommended that these instructional sets be referred to as secondary-task and dual-task paradigms, respectively.  213  control conditions). The difference between the two levels of performance is taken as an index of the attention demands imposed by the primary task. Implicit within this theoretical approach is the notion that the secondary task does not interfere in the performance of the primary task. If one assumes (1) the existence of limited pool (or pools) of processing resources, and (2) the use of an instructional set that allocates processing priority to the primary task, it is postulated that secondary-task performance is a direct reflection of the amount of "free attentional space" or "residual processing capacity" that remains for secondary task execution after the demands of the primary task have been met. If poor performance (relative to control levels) is displayed on secondary activity, it is hypothesized that a large proportion of informationprocessing capacity has been allocated to the primary task so that an acceptable level of performance can be maintained. In contrast, if secondary task performance remains relatively unchanged from control levels, it is proposed that only a small proportion of information-processing capacity has been allocated to the primary task (Figure A2). As such, poor performance is expected to accompany relatively difficult primary tasks, which demand a substantial proportion of attentional capacity for task execution; while, good secondary task performance is expected to be associated with more simple primary tasks, which place little demand on the performer's limited processing capacity. To determine if attentional capacity is required throughout the performance of a motor skill, researchers typically employ a secondary task that is continuous in nature. For example, participants could be required to count backwards in threes while trying to produce a criterion bimanual coordination pattern (see Hodges & Lee, 1999). In contrast, a discrete probe technique may be used if the researcher is interested in examining the attention demands required at a specific point in time or by a specific component of the motor task. The probe technique is illustrated in Experiments 1 - 5 of this dissertation. While performing a bimanual coordination task, a computer generated "beep" sounded. As soon as the individual detected the beep, (s)he was required to respond as quickly as possible by providing a vocal response into a microphone. Further discussion of this technique is now presented.  214  co  CO  o  CD  c  CD <  E  2  M—  "D CD  -*—<  Q. CO  T3  D) O O "O O  T3 O O  O  CD  E  CO CO -*—' I  CO TJ  c o o  o a U o  CD CO  O)  c >> l_  CD TJ C CO  c o  Q.  E  o o is h S » ci o « S< 3  Pi  3 u «  215  Probe Reaction Time Technique (Probe-RT). Although a diverse range of  secondary tasks has been employed (see Ogden et al., 1979, for a review), probe reaction time (Probe-RT) measurements are prevalently used to determine attentional fluctuations during the execution of a particular primary task. The premise of this technique is to present a stimulus (probe) while carrying out the required primary movement. At the onset of a probe, the participant is instructed to respond as quickly as possible to the stimulus (e.g., by pressing or depressing a telegraph key) while simultaneously maintaining primary task performance. If primary performance is the same on trials with and without the probe, then it is assumed that any delay in probe reaction time (relative to control levels) reflects the processing requirements of the primary task. The attention demands of component processes of the primary task are inferred by having the onset of the probe occur at different times within the movement (e.g., probe onset is simultaneous with the start of the movement). When a particular component of the task elicits a delay in RT, that process is said to be attention demanding. If there is no delay in RT it is assumed that attention is a minimal requirement or may not be needed at all for the execution of that particular task component. This line of investigation has sought to answer such questions as: Is attention necessary for response initiation and/or movement termination? Does the precision requirement of a movement task affect its attention demand? (Stelmach & Larish, 1980). Overall, empirical work using probe-RT measures suggests that movement control requires attention. Moreover, the relative degree of attentional involvement changes systematically throughout the execution of the movement. For example, in an early study examining the execution of a simple reaching movement, Posner & Keele (1969) found that probe-RT was always larger than baseline RT. This suggests that the linear positioning movement demanded attention. Furthermore, the shape of the curve relating RT to probe position was U-shaped. Attention demands were markedly elevated at the initiation of the movement, reduced in the middle portion, and then increased towards movement termination in an attempt to achieve precise positioning on the target. Posner and Keele support this latter interpretation by demonstrating that probe-RT increased when precision of the task  216  increased (also see Ells, 1973; Salmoni, Sullivan, & Starkes, 1976, for further investigations examining simple linear positioning movements). More recently, the probe technique has been used to assess the time course of attentional fluctuations in a number of real-world sport tasks, as well as laboratory simulations of these tasks. Examples include: ball-tracking (Nettleton, 1979), simple catching (Populin, Rose, & Heath, 1990; Starkes, 1987); service reception in tennis and volleyball; as well as sprinting and hurdle running (Castiello & Umilta, 1988), horseshoe pitching (Prezuhy & Etnier, 2001); rock climbing (Bourdin, Nougier, & Teasdale, 1997; Bourdin, Teasdale, & Nougier, 1998); and pistol shooting (Rose & Christina, 1990). In a number of these activities, the observed pattern of attentional demands were similar to those revealed in the control of simple linear aiming movements (e.g., Posner & Keele, 1969). Using a laboratory coincident-anticipation timing task, for example, Nettleton (1979) found the greatest allocation of attention to the early and final stages of an approaching stimulus, rather than to the middle 'less attention demanding' portion of its approach. In this investigation, a coincidentanticipation task was employed to replicate attentional demands of tasks requiring ball-tracking skills (e.g., soccer, field hockey). In consideration of more 'real-world' assessments, Prezuhy and Etnier (2001) asked individuals to pitch horseshoes to target stakes of varying difficulty. Results revealed that horseshoe pitchers allocated a greater portion of attentional resources to the initiation of the pitch and the point just prior to the release of the horseshoe, than to the full extension on the backswing. When examining sprinters executing a 100 m dash, Castiello and Umilta (1988) demonstrated delayed response latencies of probe-RT at the beginning and at the end of the race. An interpretation of the results suggests that the beginning portion of the race involves the start, whereby attention must be devoted towards detection of the start signal and preparation of the response. However, the performer must also allocate attentional resources to the end portion of the race to monitor distance to the finish line, the relative positions of the opponents, as well as the final thrust to the finish line. In contrast, a number of studies have revealed attentional patterns unique to the aforementioned findings. For hurdlers, Castiello and Umilta (1988) found elevated  217  probe-RTs throughout the middle of the race, particularly when the athlete was about to negotiate the first hurdle. In the same series of investigations, Castiello and Umilta also revealed task-specific attentional fluctuations for service reception in tennis and volleyball. In both activities, attentional demands were greatest when the ball hit the court (just prior to performing the return in tennis) or after crossing the net (just prior to hitting the ball in volleyball). Contrarily, the least demanding conditions for service reception in tennis and volleyball occurred at the opponent's serve and when the ball was above the net. Work by Rose and Christina (1990) has also revealed a unique attentional time course for shooting. When examining the distribution of attention across the aiming phase of a discrete pistol shot, it was found that probe-RT increased when the auditory probe was presented at a time more proximal to the time of the shot. Given that probe-RT increased in a near linear fashion, Rose and Christina concluded that the demands on attentional capacity increased as the actual time of the shot approached. Empirical work using the probe-RT technique not only demonstrates that attentional involvement changes systematically throughout the execution of a movement, but also reveals a unique array of attentional patterns. However, further examination is needed to establish a more extensive database regarding the time course of attention demands within the execution of a motor task (Abernethy, 1993). Although limited, current empirical work suggests that certain tasks may display common patterns of attention. In particular, are those tasks, which interact with the environment in a comparable manner. Prezuhy and Etnier (2001) argue that a general attention pattern may exist for different classifications of tasks. For example, projection-type tasks (horseshoe pitching, bowling, throwing) may display one kind of attentional demand profile, while reception-type tasks (hitting a pitched ball, receiving a serve in tennis or volleyball) may exhibit another. This line of inquiry is clearly warranted from a motor learning perspective in that it has the potential to enhance instructional techniques and ultimately improve performance.  218  Summary Any model that seeks to describe human skill acquisition must take into consideration the mediating influences of attention. One approach has been to examine the nature of human performance with a direct emphasis on discovering the attentional limitations of the processing system. Specifically, a prominent line of investigation has been to examine the limitations on the amount of information an organism can select for processing at any given period of time. Coined as the study of divided attention, researchers seek out an explanation for why the human system is seemingly able to perform two tasks concurrently, or contrarily, why the performance of one task interferes with the simultaneous performance of another task in certain movement situations. One of the most widely accepted theoretical explanations for dual-task interference is the one proposed by resource or capacitysharing models. These models postulate the existence of a limited pool (or pools) of processing capacity, which can be partitioned between two tasks. If the concurrent tasks tap into the same resources and joint demands exceed the total capacity of available resources, a dual-task decrement incurs in one, or both of the simultaneously performed tasks. Moreover, to examine processing capacity demanded by a task (i.e., attention demands), a basic secondary task paradigm is employed, which requires the simultaneous performance of two tasks - a primary task, for which attention demands are assessed, and a secondary task, from which attention demands of the primary task are derived. Empirical work using probe-RT measures suggests that attentional demands change systematically throughout the execution of a movement and the resulting pattern is specific to the particular task of interest.  The Relation between Attention and Automaticity The link between automaticity and skilled behaviour is not a new concept and is well entrenched within academic psychology. For example, Adams (1987) has traced the notion of automaticity back to the writings of Bain (1868) and Spencer (1881). Bain (1868) is of particular interest in that he discussed the principle of contiguous association whereby movements could be linked together in trains and  219  made to succeed each other. Using such examples as the sequence of acts in eating, the articulation of speech, the playing of the pianoforte, and the acquisition of the diagonal step in a military walk, Bain suggested that succession of regulated acts occur with mechanical coherence and certainty because they have been made to succeed each other a great many times. In the acquisition of articulate speech, Bain stated that, "every letter stands in need of an adjustment of tongue, jaws, and lips, difficult at first, but at last so easy that we do not know that we are performing a complicated act" (p. 331). Furthermore, he suggested that the effect (or 'feeling') produced at the end of each movement is a link in the transition to the next action. Early in learning, this notion of 'feeling' guides an individual's mechanical effort until the movements are fixed in a sequence, operating independently of any guidance. This last stage is epitomized to represent the highest perfection of mechanical acquirement. Inherent within Bain's writings, was an anticipated preamble of what was yet to come. In James' (1890) Principles of Psychology, an entire chapter was dedicated to the notion of 'habit', where automaticity was discussed in terms of the responsechaining hypothesis. According to this hypothesis, each segment of a movement produces response-produced feedback. The feedback from one movement segment becomes associated to the next movement sequence by a habit connection, and acts as a stimulus for the next movement in the sequence. With respect to acquisition, James suggested that a voluntary act has to be guided by "idea, perception, and volition, throughout its whole course" (p. 115). However, once a habit connection has been formed, movements can be run off as smooth serial actions. In a habitual action, "mere sensation is a sufficient guide and the upper regions of the brain and mind are set comparatively free .... habit diminishes the conscious attention with which our acts are performed" (p. 115-116 & p. 114, respectively). Although other psychologists had acknowledged the concept of automaticity in their writings (e.g., Bain, 1868), James popularised the responsechaining hypothesis. As such, the idea that a shift from controlled to automatic processing occurs over extensive practice has usually been attributed to the 1890 writings of James (see Adams, 1984,1987).  220 At the end of the 19 Century, Bryan and Harter published two seminal tn  papers, which examined the acquisition of skill in telegraphy (Bryan & Harter, 1897, 1899). The work of Bryan and Harter is considered 'seminal' in that many of the ideas presented within these two papers sparked an interest in the study of motor expertise, serving as an impetus for theorizing and research within the area of motor learning throughout the latter half of this century. According to Lee and Swinnen (1993), Bryan and Harter's work opened the door for research in three distinct areas: the examination of expertise in terms of systematic differences in performance variation; the nature of change in acquisition, as well as the development of automaticity in expert performance. The latter concept of automaticity was perhaps, the most important legacy of Bryan and Harter and has remained a consistent theme within present-day theories and research pertaining to the acquisition of skilled behaviour (see Lee & Swinnen, 1993). According to Bryan and Harter, acquiring expert levels of motor performance depends on the attainment of higher-order habits, and therefore, automaticity is hierarchical in nature. Higher-order elements of the task cannot be attended to until lower aspects of the task are performed relatively attention free. In their words, "only when all the necessary habits, high and low, have become automatic, does one rise into the freedom and the speed of the expert" (1899, p. 357). Stated differently, "there is no freedom except through automatism" (1899, p. 369). Unfortunately, interest in automaticity waned with the onslaught of behaviourism, and did not reappear until Miller, Galanter, and Pribram's (1960) publication of Plans and the Structure of Behavior (Lee & Swinnen, 1993). Within this piece of work, Miller et al. viewed skilled behaviour as a function of a plan that becomes habitual in nature through the process of overlearning. More specifically, Miller et al. suggested that early in acquisition, the beginner formulates a plan to guide gross actions for the control of a sequence of skilled actions. In order to execute the plan as a smooth and controlled behavioural unit, the individual must identify interposed elements that produce the skill. Eventually, the plan becomes almost innate (or instinctive) as it develops the qualities of automatic behaviour (e.g., inflexibility, resistance to change, involuntary responses). Simply stated, Miller et al.  221  suggested that automaticity provides the individual a method of manufacturing his own instincts whereby habits and skills are viewed as plans that have become automatic. Emergence of the cognitive revolution in the 1950s and the 1960s facilitated debate on the relation between attention and automaticity. At the outset, research focused on selective attention and the fate of unattended material (Logan, Taylor, & Etherton, 1999), where the purpose of empirical investigations was to determine at what level a parallel system, unlimited in capacity, narrowed to a serial system, limited in capacity. The former implying the use of preattentive or automatic processes, while the latter involving processes that require attention. Early selection theorists argued that information entered the cognitive system in parallel, but only low-level perceptual features were processed preattentively. Any further analysis at the semantic level was postulated to demand attention (e.g., Broadbent, 1958). In contrast, late selection theorists argued that limitations in the system occurred later on in processing, after a message has been processed to the extent that its relationship to other messages is determined. These models proposed the notion of a 'smart filter' where selectivity is based on highly complex discriminations occurring beyond the stimulus level and after at least some degree of preattentive semantic analysis. From this perspective, attention is required at the level of response selection since all incoming signals have been processed prior to this stage (e.g., Deutsch & Deutsch, 1963). Inherent within both early- and lateselection theories is the idea that automatic processes are independent of attention (Logan et al., 1999). In the 1970s, interest shifted from the investigation of selective attention to the study of divided attention. Embedded within this new dominant paradigm was the notion that automaticity was a learned phenomenon. Therefore, an interest in the study of behaviour under conditions of prolonged practice emerged within the psychological literature.  222  Automaticity as a Learned Phenomena One of the most significant contributions to the area during this time period was the work of Schneider and Shiffrin (Schneider & Shiffrin, 1977; Shiffrin and Schneider, 1977), who studied the nature of automatic processing by asking such questions as: Is automatic processing qualitatively different from effortful processing? What kinds of processes become automatic? (Hirst, 1986). In their work, Schneider and Shiffrin employed a basic visual search and detection paradigm, in which the participant was asked to search displays for target items. The individual was first shown a memory set of one (or more) alphanumeric characters), and then asked to identify whether the item (or any member of the memory set) appeared in a subsequent frame, referred to as the visual display set. A presented memory set item was called a target, while the other presented stimuli were always referred to as distractors. The task could be made extremely easy (i.e., the participant was required to memorize only one item and received only one item in the visual display); or the task could be made difficult (i.e., the participant was given four items to memorize as well as four items in the visual display). Participants were also examined under two different mapping conditions referred to as consistent mapping (CM) and varied mapping (VM). Consistent mapping (CM) was used to denote those conditions where items in both the memory and visual display sets never changed across trials. That is, only numbers were used as items in the memory set and only consonants were used as distractors in the visual display set (or vice-versa). The latter condition of varied mapping (VM) involved the intermixing of memory-set and distractor items, as well as a random change in roles across trials. For example, in the VM conditions both numbers and consonants were used as items in a manner such that a target on one trial could be a distractor on another trial, and vice-versa. This method of investigation led Schneider and Shiffrin to a number of important findings. After extensive training under conditions of CM, performance was found to be largely unaffected by load (defined as a product of memory-set size and visual display size). With prolonged practice, participants displayed fast reaction times in all conditions of load, and target detection accuracy was high even when frames were shown to the individual at a rapid rate of display.  223 Furthermore, participants were able to carry on a conversation while performing the task, which suggests that the CM condition became easy with practice and demanded little attention or effort from the performer. However, in conditions of VM, performance did not improve and remained relatively unchanged even after months of practice. In the VM conditions, participants reported that the task was rather arduous and commented that it demanded a conscious allocation of attention on their part. Since effects on both the memory-set size and display-set size were twice as large on trials when there was no match between any of the memory items and any of the visual display items (i.e., negative trials) Schneider and Shiffrin concluded that participants were using serial comparison to elicit a response. That is, participants were comparing each item in the visual display against each item in the memory set to find a match. Such a comparison can only be made serially and at a limited rate. According to Schneider and Shiffrin, the process of serial comparison is an example of a controlled mode of processing. In contrast, controlled comparison is not needed to locate the target after prolonged practice in CM conditions. Rather, it was proposed that target detection was attributed to the use of automatic processes operating independently and in parallel as a result of training. Interestingly, Schneider and Shiffrin also examined performance when task demands were reversed after substantial practice using consistent mapping. In the first 2100 trials of practice, participants received a memory set, which could only consist of consonants from B to L, and a visual display set consisting only of consonants from Q to Z (or vice-versa). After 2100 trials under consistent mapping, participants' accuracy for target detection increased from an initial 55% to a detection rate greater than 80%. However, this initial set of trials was then followed by an additional 2400 trials of consistent mapping where the composition of each set was reversed (i.e., memory set = consonants Q to Z; visual display set = consonants B to L,