FIFTH EDITION

Richard A. Schmidt, PhD, Professor Emeritus

Psychology Department
University of California, Los Angeles
and
Human Performance Research
Marina del Rey, California

Timothy D. Lee, PhD

Department of Kinesiology
McMaster University, Ontario, Canada

HUMAN KINETICS
 Library of Congress Cataloging-in-Publication Data

Schmidt,RichardA., 1941-
Motor control and learning : a behavioral emphasis I Richard A. Schmidt and Timothy D. Lee. -- 5th ed.
p. ;cm.
Includes bibliographical references and indexes.
ISBN-13: 978-0-7360-7961-7 (hard cover)
ISBN-10: 0-7360-7961-0 (hard cover)
1. Movement, Psychology of. 2. Motor learning. I. Lee, Timothy Donald, 1955- II. Title.
[DNLM: 1. Motor Activity. 2. Motor Skills. 3. Learning. WE 103]
BF295.S248 2011
152.3--dc22
2010044051

ISBN-10: 0-7360-7961-0 (print)

ISBN-13: 978-0-7360-7961-7 (print)

Copyright© 2011, 2005, 1999 by Richard A. Schmidt and Timothy D. Lee

Copyright © 1988, 1982 by Richard A. Schmidt

All rights reserved. Except for use in a review, the reproduction or utilization of this work in any form or by any electronic, mechani-
cal, or other means, now known or hereafter invented, including xerography, photocopying, and recording, and in any information
storage and retrieval system, is forbidden without the written permission of the publisher.

The Web addresses cited in this text were current as of August 2010, unless otherwise noted.

Acquisitions Editor: Myles Schrag; Managing Editor: Melissa J. Zavala; Assistant Editors: Antoinette Pomata and Steve Calde-
rwood; Copyeditor: Joyce Sexton; Indexer: Andrea J. Hepner; Permission Manager: Dalene Reeder; Graphic Designer: Bob
Reuther; Graphic Artist: Denise Lowry; Cover Designer: Keith Blomberg; Photographer (cover): ©Yuri Kadobnov/AFP/Getty
Images; Photographer (interior): ©Human Kinetics, unless otherwise noted; Photo Production Manager: Jason Allen; Art
Manager: Kelly Hendren; Associate Art Manager: Alan L. Wilborn; Art Style Development: Joanne Brummett; Illustrations:
© Human Kinetics; Printer: Sheridan Books

Printed in the United States of America 10 9 8 7 6 5 4 3 2

The paper in this book is certified under a sustainable forestry program.

Human Kinetics
Web site: www.HumanKinetics.com

United States: Human Kinetics, P.O. Box 5076, Champaign, IL 61825-5076

800-747-4457
e-mail: [email protected]

Canada: Human Kinetics, 475 Devonshire Road Unit 100, Windsor, ON N8Y 2L5
800-465-7301 (in Canada only)
e-mail: [email protected]

Europe: Human Kinetics, 107 Bradford Road, Stanningley, Leeds LS28 6AT, United Kingdom
+44 (0) 113 255 5665
e-mail: [email protected]

Australia: Human Kinetics, 57 A Price Avenue, Lower Mitcham, South Australia 5062
08 8372 0999
e-mail: [email protected]

New Zealand: Human Kinetics, P.O. Box 80, Torrens Park, South Australia 5062
0800 222 062
e-mail: [email protected]

E4689
 <+j!~';i~c-'.>':/ /::<;;~;;;->-- '<'~-,.,,- _,---, -'
'>>:.:;::''s/:>;;-,:,;-/;{;:C:-.c'>":-"'->~ . ---' ,
-"" :;:,o_; Y'i",. \&:i·C:>"-"-, CONTENTS

Preface vii Acknowledgments ix

PART I INTRODUCTION TO MOTOR BEHAVIOR 1

CHAPTER l Evolution of a Field of Study . .......................... .3

Understanding Movement . . 4
Origins of the Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

CHAPTER 2 Methodology for Studying Motor Performance ......... .21

Classification of Motor Skills . . . . . . . 21
Basic Considerations in Measurement. . .24
Measuring Motor Behavior. . . . . . . . .26
Measuring and Evaluating Relationships .48
Summary . . . . . . . . . . . . . . . . . .54

CHAPTER 3 Human Information Processing ....................... .57

The Information-Processing Model . . . .57
Three Stages of Information Processing. .58
Anticipation . . . . . . . . 78
Signal-Detection Theory . . 85
Memo~ . . . . . . . . . .88
Summary . . . . . . . . . .93

CHAPTER 4 Attention and Performance ......................... .97

Types of Attention . . . . . .97
Theories of Attention . . . . 100
Competition for Attention. . 103
Attention During Movement 116
Focus of Attention . . 124
Attention and Anxiety 127
Summary . . . . . . . 130

iii
iv Contents

PART II MOTOR CONTROL 133

CHAPTER 5 Sensory Contributions to Motor Control. . . . . . . . . . . . . . . . . . 135

Closed-Loop Control Systems . 135
Vision . . . . . . 136
Audition . . . . . . . . . . . . 153
Proprioceptors . . . . . . . . . 153
Proprioception and Motor Control . 158
Feedforward Influences on Motor Control . 170
Summary . . . . . . . . . . . . . . . . . . 174

CHAPTER 6 Central Contributions to Motor Control . ................. 177

Open-Loop Processes . . . . . . . . 177
Central Control Mechanisms. . . . . 182
Central Control of Rapid Movements 188
Generalized Motor Programs . . . . 208
Summary . . . . . . . . . . . . . . . 221

CHAPTER 7 Principles of Speed and Accuracy ..................... 223

Fitts' Law: The Logarithmic Speed-Accuracy Trade-Off. . . . . 224
The Linear Speed-Accuracy Trade-Off . . . . . . . . . . . . . 231
The Temporal Speed-Accuracy Trade-Off. . . . . . . . . . . . 234
Central Contributions to the Spatial Speed-Accuracy Trade-Off 241
Correction Models of the Speed-Accuracy Trade-Off. . . . . . 257
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260

CHAPTER 8 Coordination . ....................................... 263

Discrete Tasks . . . 264
Continuous Tasks . . . . . . . . . . . . . . . 277
Summary . . . . . . . . . . . . . . . . . . . . 294

CHAPTER 9 Individual Differences and Capabilities .... _............. 297

Experimental Versus Differential Approaches . . 297
Abilities . . . 301
Taxonomies. . . . . . . . . . . 310
Prediction . . . . . . . . . . . 312
Individual-Difference Variables 315
Summary . . . . . . . . . . . . . 324

PART III MOTOR LEARNING 325

CHAPTER l 0 Motor Learning Concepts and Research Methods . ...... 327

Motor Learning Defined . . 327
Measuring Motor Learning . . . . . . . . . . . . . . . . . . . . . . . 329
 Contents v

Designing Experiments on Learning . . . . . 335

Alternative Methods for Measuring Learning 338
Issues About the "Amount" of Learning . . . 342
Understanding Learning and Performance Variables . 345
Summary . . . . . . . . . . . . . . . . . . . . . . . 345

CHAPTER 11 Conditions of Practice ............................... 347

The Law of Practice . . 347
Motivation for Learning. 351
Verbal Information . . . 353
Perceptual Learning .. 355
Observational Learning 356
Mental Practice. . . . . 359
Distribution of Practice. 361
Variability of Practice .. 367
Contextual Interference 371
Part Versus Whole Practice 382
Guidance . . . . . . . . . . 386
Principles of Practice Specificity. 388
Summary . . . . . . . . . . . . . 390

CHAPTER 12 Augmented Feedback .............................. 393

Classifications and Definitions . . . . . . . . . . 393
Research on Augmented Feedback. . . . . . . . 395
Evaluating the Effects of Augmented Feedback. . 398
Knowledge of Performance . . . . . . . . . . . . 400
Knowledge of Results . . . . . . . . . . . . . . 405
Theoretical Issues: How Does Augmented Feedback "Work"? . 424
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427

CHAPTER 13 The Learning Process ................................ 429

Characteristics of the Learning Process. 429
Closed-Loop Theory . . . . . . . . . . . . . . . . . . 438
Schema Theory . . . . . . . . . . . . . . . . . . . 441
Differing Theoretical Perspectives of Motor Learning . 446
Summary . . . . . . . . . . . . . . . . . . . . . . . . 456

CHAPTER 14 Retention and Transfer............................... 461

Fundamental Distinctions and Definitions . 461
Measuring Retention and Transfer . 463
Retention and Motor Memory 467
Retention Loss . . . 471
Transfer of Learning 481
Summary . . . . . . 488

Appendix 491 Glossary 493 References 503 Author Index 565

Subject Index 573 About the Authors 581
 PREFACE

ost of us have marveled at one time detection of errors, spinal cord processes that are
or another about how the most highly capable of generating patterns of skilled activi-
skilled performers in industry, sport, music, ties in locomotion, and various biomechanical
or dance seem to make their actions appear so factors that act to determine the nature of our
simple and easy-performed with incredible effi- movement behaviors are all important if we are
ciency, smoothness, style, and grace. Like the first to understand highly skilled performance. This
four editions of this text (Schmidt, 1982, 1988; blending of behavioral, neurophysiological, and
Schmidt & Lee, 1999, 2005), the fifth edition of biomechanical analyses reflects the fact that the
Motor Control and Learning: A Behavioral Emphasis fields of motor behavior and motor learning,
was written for those who would like to under- movement neurophysiology (or motor control),
stand how it is that these performers can achieve and biomechanics are rapidly moving together
such artistry while we, as beginners in a similar toward the shared understanding of complex
task, are clumsy, inept, and unskilled. This book movement behaviors.
was written particularly as a textbook for uni- This edition of the text retains the same goal
versity or college undergraduate and graduate of presenting an up-to-date review of the state of
students taking courses in human performance knowledge in movement control and learning,
or motor learning, primarily in fields such as and it does so with a format similar to that of
kinesiology or psychology. However, students in the previous editions. We have directed consid-
other fields, such as the neurosciences, physical erable effort toward including the most recent
therapy, occupational therapy, speech therapy, knowledge from a number of rapidly develop-
biomedical or industrial engineering, human ing subfields, and each chapter has been revised
factors or ergonomics, and sport, will also find extensively in light of these newer concepts. In
many of the concepts contained here to be of addition to including more than 280 new refer-
interest, as movement behavior is a part of all of ences to work published since the last edition (in
them. And for those who are, or who are becom- 2005), we have also endeavored to pay homage
ing, practitioners in these fields, the principles of to some of the importan~ early research develop-
motor behavior outlined here should provide a ments in the various areas. One of the features
solid basis for tasks such as designing human- introduced in the previous two editions was
machine systems, developing training programs sidebars highlighting specific research issues
in sport or industry, or teaching progressions in throughout the book. We have created more of
dance or music. these sidebars in this fifth
The emphasis of the text is behavioral. That edition with the idea that
is, the primary focus is on movement behavior certain material requires
that can be observed directly, as well as on the a more detailed treatment
many factors that affect the quality of these per- to ensure understanding
formances and the ease with which they can be than is typical of passages HumanKinetics.com
learned. In this sense, the book has strong ties in the text. Some of these
to the methods and traditions of experimental sidebars highlight quotations from selected his-
psychology. Yet, at the same time, we focus on torical papers; other sidebars highlight careers
the neurological and biomechanical processes of key researchers in motor control and learning;
out of which these complex movement behaviors and still others deal with various applications of
are crafted. Brain mechanisms that allow the specific concepts in motor control and learning.

vii
viii mi Preface

Some chapters from the previous edition have Part II deals with motor control. Chapter 5
been reduced in order to lessen complexities in views motor control from a closed-loop per-
the text without sacrificing the in-depth cover- spective, in which the sensory contributions to
age or richness of the concepts. And we have movement control are examined, with particular
expanded other chapters and sections to present emphasis on new research regarding the role of
new, exciting areas of research that have emerged vision. In chapter 6, the focus shifts to contribu-
since the previous edition. Many new figures tions of the central nervous system to movement
have also been included to help illustrate and control, with emphasis on preorganized actions
emphasize concepts and data that are difficult that involve motor programs and generalized
to communicate effectively in words. Practical motor programs. Some principles related to
examples from areas such as human factors, sport, speed and accuracy are presented in chapter 7,
therapy, and music are provided to illustrate together with a discussion of theoretical concepts
motor control and learning concepts and provide that integrate the central and sensory contribu-
suggestions for application. As before, the fifth tions to movement control. Chapter 8 presents a
edition reflects a logical progression so that later discussion of the factors involved in movement
chapters build upon concepts presented in earlier control that make coordination both easy and
chapters, with the final result being a consistent, difficult to achieve. The final chapter in this part
defensible framework of ideas about motor skills. contains a discussion of factors that determine
Having such a framework, or point of view, is skill differences between people and among
important for those who wish to use the informa- groups of people, with emphasis on important
tion presented here, both so that contributions to themes about abilities and the prediction of skills.
new applications may be made and so that the Part III deals with the acquisition of skill, or
design of new skills research is facilitated. motor learning. Chapter 10 concentrates on some
The book is divided into three parts. Part I pro- special methodological problems for studying
vides an introduction to research and fundamen- learning. The effects of practice, the structure
tal concepts that are important to understanding of the practice session, and the many variables
motor behavior. The first chapter, a brief history under the control of a teacher, coach, or therapist
of the field, is followed by a presentation of meth- are discussed in chapter 11, while feedback con-
ods in movement research in chapter 2, focusing tributions to learning are included in chapter 12.
on various paradigms and statistical techniques In both of these chapters, much new information
used in the study of movement behavior. In is covered that demands important changes in
chapter 3 the human is regarded as a processor our understanding of the processes involved in
of information, and we focus on the many ways practice and the ways in which these influence
that information is dealt with in motor behavior. learning. Chapter 13 presents various theoretical
The concept of attention is the focus of chapter 4, treatments of motor learning. And finally, chap-
with particular emphasis on the role of attention ter 14 deals with the factors associated with the
in motor behavior. retention and transfer of skills.
 hroughout the long process of this revision, Thanks also go to our editors, Melissa Zavala
a number of people provided very highly and Antoinette Pomata, for their excellent work
valued input. Judy Wright of Human Kinetics on the project. We called upon a number of col-
provided encouragement and many valuable leagues to read specific sections of the book. The
suggestions for not only this revision, but on many suggestions made by Ramesh Balasubra-
previous editions of the book as well. Her recent maniam, Ian Franks, Waclaw Petrynski, Gwen
retirement closes an important chapter not only Gordon, and Laurie Wishart were invaluable in
for the publisher, but also for the expression making these revisions. The final manuscript is
of ideas presented in this book. Judy, we wish much better as a result, and we are grateful for
you an energetic and successful next dance. their time and effort.

ix
CHAPTER 1 Evolution of a Field of Study
CHAPTER 2 Methodology for Studying Motor Performance
CHAPTER 3 Human Information Processing
CHAPTER 4 Attention and Performance

his first part introduces the field of motor control and learning. In chapter 1 the area
is described, and the important distinctions separating motor control and learning
from other, related fields of study are made. Then, a brief history of the field is given,
showing how knowledge about movements from psychology and kinesiology, as well
as from the neurosciences, has been combined. The second chapter deals with the vari-
ous scientific methods used for studying motor skills. Here, we explain the tools of
motor behavior research, focusing on the various ways in which motor behavior and
skill can be measured. Chapter 3 presents the information-processing approach, which
is fundamental to understanding how humans think and act. The last chapter in this
section describes attention and its role in motor behavior.

1
 ovement is a critical aspect of life. Without ing a typewriter, or performing a triple twist-
movement, we could not feed ourselves, ing somersault from the diving board. These
we could not reproduce, and we would not learned movements are often termed skills.
survive. Our capacity to move is more than just They are not inherited, and mastering them
a convenience that enables us to walk, play, or requires long periods of practice and experi-
manipulate objects; it is a critical aspect of our ence. Guthrie (1952) perhaps provided the best
evolutionary development, no less important definition of skills: "Skill consists in the ability
than the evolution of our intellectual and emo- to bring about some end result with maximum
tional capacities. Some assert that our highly certainty and minimum outlay of energy, or of
developed cognitive capacities evolved so that time and energy" (p. 136). Skills are especially
we could make the movements essential to critical to the study of human behavior, as they
survival-those involved in the construction of are involved in operating machines in industry,
shelter, the making of tools, and communication. controlling vehicles, preparing meals, playing
Surely the study of movement needs no further games, and so on. Skills .and genetically defined
justification than its significance in terms of the movements can range from very simple (e.g.,
evolution of humankind. snapping fingers or blinking eyes) to very com-
Movement takes many forms. Some forms plex (e.g., pole-vaulting).
can be regarded as genetically defined, such as This book is about all these kinds of move-
the way in which people control their limbs or ments, whether primarily genetically defined or
the ability of centipedes to coordinate their legs. learned through practice. In particular, we will
Other examples include the "scratch reflex" of be concerned with how these various movements
dogs or the rapid blink of the eye in response are controlled-how the central nervous system is
to an unexpected puff of air. Here, the patterns organized so that the many individual muscles
of action appear to be determined by genetic and joints become coordinated. We will also be
makeup, through growth and development, concerned with how sensory information from
or in both ways; and these actions appear to the environment, the body, or both is used in the
be quite stereotypical for members of the same control of movement. The scientific field of study
species. A second class of movements can be that addresses these issues is known as motor
thought of as "learned"-for example, those control-the study of the control of movements
involved in controlling an automobile, operat- in humans and animals.

ii 3
4 mi Motor Control and Learning

In this book, we add one important aspect to the historical and sociological effects of illnesses
motor control that is sometimes not included- on entire societies. Because these various ways
the study of how movements are learned, that is, of considering a single problem are so diverse, a
how movements are produced differently as a given scientist usually focuses on one, or at most
result of practice or experience. Indeed, much two, of these levels of analysis.
evidence suggests that many of the movements
already mentioned consist of a complex combina- A Behavioral Level of Analysis
tion of genetic determinants coupled with modi-
The focus of this text is primarily at the behavioral
fications made through practice or experience.
level of analysis, centering on cognitive, informa-
Understanding how movements are learned is
tion-processing concepts. The major goals will be
the major concern of a field of study called motor
to understand the variables that determine motor
learning. We see no good justification, however,
performance proficiency and to understand the
for separating the study of motor learning from
variables that are most important for the learn-
the study of movement or of motor control in
ing of movement behaviors. We also want to
general, as this artificial separation inhibits the
understand how such information can be used in
understanding of both issues. For these reasons,
the solution of certain practical problems such as
as the title reveals, the subject matter of the book
those involved in the design of equipment that
is motor control and learning.
humans must operate, in the selection of indi-
viduals for occupations, in the teaching of skills
Understanding Movement in sport and industry, and in the rehabilitation of
skills after injury or stroke.
This behavioral level of analysis, however, is
How can knowledge and information about
more interesting and complete when combined
movement be acquired? A logical way to proceed
with two other fields of study, each representing
would be to study some relevant aspect of the
a deeper level of analysis. The field of biomechan-
movement-control process using scientific meth-
ods. But which processes should be examined?
ics concerns the mechanical and physical bases of
biological systems. Certainly in order to under-
One possibility would be to focus on the nature of
stand movement we must understand something
biochemical interactions that occur within cells as
of the body itself, with all its joints, levers, and
individuals move. Or we could focus on the cell
associated mechanical characteristics. The field
itself, asking how cells interact with each other
of neurophysiology concerns the functioning of the
in the control of movement. In a similar way, we
brain and central nervous system and the ways
could consider groups of cells, such as a whole
in which they control the contractions of muscles
muscle, the spinal cord, or the nerves, and ask
that move the limbs. The study of movement will
how these relatively more complex structures are
be addressed at various levels of analysis-but as
involved in movement control. Another possibil-
the subtitle of the book suggests, the emphasis is
ity would be to focus on the movements of the
at the behavioral level.
freely moving animal or human, concentrating on
the factors that determine movement accuracy,
Emphasizing Movements
the choice of movement, or the patterns of action.
Along the same lines, we could study movement In considering movements, especially skills, it
in an even more global context, asking questions is often difficult to isolate a movement from its
about the role of movement in society, the choice environment. In driving a car, for example, there
of certain skilled occupations or sports, move- are the coordinated actions involved in changing
ment in groups or teams, and so on. gears (clutch, accelerator, shift lever, etc.) as well
Clearly, there are various ways to consider the as the movements involved in steering. These
same phenomenon. They involve the study of a parts of the skill are the means through which
phenomenon on different levels of analysis, and the driver affects his environment. But skills are
analogous levels are present in any area of scien- also affected by the environment. For example,
tific concern. Illnesses, for example, can be con- whether or not there are turns in the road or
sidered at levels that range from the biochemical whether snow is present influences the driver's
and neurological determinants of disease through interactions with the vehicle controls. Such recip-
 Evolution of a Field of Study 111 5

rocal relations between the environment and the such as using a lathe, typing, woodcarving, and
individual make it very difficult to pinpoint the handwriting, are of critical importance to this
various determinants of motor behavior, because field of study. Artistic performances, such as
the interaction of the many motor control and the playing of musical instruments, the creation
environmental factors is extremely complex and of a painting, or the production of a dance, are
difficult to study with experimental procedures. certainly under the heading of motor behavior
The approach taken in this text is to focus on as treated here. The use of voice, whether by the
the mutual interactions between the environment vocalist in an opera or by the student learning a
and the motor system. A large portion of this new language,1 is also a motor task, as the sounds
approach deals with the behavior and capabili- are controlled by muscular activity of the vocal
ties of the motor system to produce movements, apparatus in ways analogous to the control of
studied more or less independently of the role the hands and fingers of the skilled typist. The
of sensory or environmental information. But at potential applications for the principles discov-
the same time, the role of environmental infor- ered in the field of motor control are present in
mation such as vision, and the ways in which it nearly every aspect of our lives.
is processed and used to guide movements, is
important. In any case, we are deliberately not
concerned with skills in which the quality of the Origins of the Field
movement components per se is almost irrelevant
to the outcome (as in playing chess). In an examination of the early research on move-
In deciding which skills to include in our field ment and learning, it will be evident that the field,
of study, it is helpful to consider the probable as we know it today, emerged from two isolated
limiting factors in the performance. In the chess bodies of knowledge. These two areas are (a) the
example, intellectual decision making seems branch of neurophysiology primarily concerned
to be the important factor and should not be with the neural processes that are associated with
included in this treatment. In a marathon, or in (or are causes of) movements, with only slight
weightlifting, the factors seem to be more closely reference to the movements themselves; and (b)
related to cardiovascular fitness and strength, the branch of psychology and related fields pri-
respectively-also not within the confines of the marily concerned with high-level skills with very
present field of study. We will emphasize skills little reference to the neurological mechanisms
in which the focus is on the capabilities to use involved. For nearly a century, these two fields
environmental information in the complex con- developed knowledge at different levels of analy-
trol of the limbs. sis but with little mutual influence. Only toward
the end of the 1970s did the two fields begin to
Potential Applications come together. For the reader interested in more
detail on these historical.developments, see Irion
Given an understanding of some of the processes
(1966), Adams (1987), and Summers (1992, 2004).
underlying the control of movements, where can
these principles be applied? High-level sports,
Early Research
games, and athletic events come to mind as areas
for application, as these activities often involve A fascinating account of some of the early insights
the same kinds of processes that are studied in the regarding actions and movement appears in a
area of motor control and learning. But potential review by Meijer (2001). In this historical paper,
generalizations should not be limited to these Meijer traces the origins of a number of ideas
kinds of activities. Many apparently genetically within current thinking to philosophers such as
defined actions such as walking and maintain- Plato, Aristotle, and Galen. Some of the earliest
ing posture are under consideration here. How empirical investigations of motor skills were
these movement capabilities, when disrupted by performed around 1820 by the astronomer Bessel
injuries or disease, can be improved by treatments (cited by Welford, 1968), who tried to understand
emphasizing the learning of new movement pat- the differences among his colleagues in recording
terns-the subject matter of physical therapy-is the transit times of the movements of stars. This
also an application area. Many industrial skills, skill involved estimating the time required for the
6 Ill! Motor Control and Learning

image of a star to move through the crosshairs of in German and French publications during the
a telescope. Bessel was interested in the processes middle and late 1800s. Some of this research is
underlying this complex skill, as well as in the summarized by Worringham (1992).
reasons some of his colleagues estimated accu- A major influence of the time was Thorndike
rately and others could not. Considerably later, (1914), who was concerned with processes under-
studies addressed the visual contributions to lying the learning of skills and other behaviors.
hand movements in localizing targets (Bowditch His Law of Effect, which continues to have its
& Southard, 1882). Leuba and Chamberlain (1909) influences in psychology, states that responses
studied the accuracy of limb-positioning move- that are rewarded tend to be repeated. Responses
ments; Fullerton and Cattell (1892) examined that are not followed by a reward tend not to be
force reproducibility; Stevens (1886) studied repeated. This idea formed the cornerstone for
timing; and Judd (1908) studied transfer oflearn- much of the theorizing about learning that was to
ing with dart-throwing tasks. Some researchers follow in the 20th century (Adams, 1978). Thorn-
used experimental methods to study expertise dike was also a pioneer in the area of individual
in sport performance (Scripture, 1894; see also differences, in which the focus is on the differ-
Fuchs, 1998). An important trend was established ences among individuals surrounding practice
by Bryan and Harter's (1897, 1899) work on (see chapter 9).
receiving and sending Morse code; periods of Most of the work mentioned here originated
no improvement (plateaus) between segments of from the field of psychology, and much of the
improvement were identified, and considerable field of motor behavior today is the legacy of
debate about the existence and interpretation this early thinking and research. But the early
of these plateaus continued for some decades research, which is similar in method to at least
(e.g., Book, 1908/1925; Keller, 1958). Galton (see some of today's work, marked a severe break in
Boring, 1950) studied the relationships among tradition from the pre-1900 views of behavior.
strength, steadiness, and body configuration The pre-1900 research often involved introspec-
in over 9,000 British males and females; Book tion, including subjective self-reports of feelings
(1908I1925) examined typing skills for very large that were unobservable. Skills were studied only
samples of subjects ranging widely in ability and because they were thought to provide "access
age. Retention of skills over long intervals of no to the mind." As the 19th century ended, there
practice was an important theme, and typing was was a shift to more systematic and objective
a convenient way to study it (e.g., Bean, 1912; approaches to the study of skills. And, of equal
Swift & Schuyler, 1907). A remarkable series of importance, skills were beginning to be studied
studies on the retention of typing skill, initiated because investigators wanted to know about the
by Hill, Rejall, and Thorndike (1913), showed skills themselves.
"savings," in terms of practice time or the amount Toward the end of this period, the number of
of practice, involved in the relearning of typing studies involving skills increased slightly. Some
skill after two consecutive 25-year periods of no of these concerned handwriting proficiency, ways
practice (Hill, 1934, 1957). in which practice sessions could be structured to
One of the earliest systematic approaches to maximize motor learning, and whether or not
the understanding of motor skills was used by skills should be "broken down" into their com-
Woodworth (1899), who sought to identify some ponents for practice. Skills research placed greater
of the fundamental principles of rapid arm and emphasis on industrial applications (Gilbreth,
hand movements. This work, together with that 1909; Stimpel, 1933). So-called time-and-motion
of Hollingworth (1909), uncovered principles studies analyzed production-line assembly move-
about visual-motor performance that remain a ments; such research became the target of criti-
topic of current debate (e.g., Elliott, Helsen, & cism by workers because of the strict standards
Chua, 2001; Newell & Vaillancourt, 2001b). Some of performance it imposed on them. There was
other research, published in German and French, rising interest in the most efficient ways to per-
went unnoticed in the English literature for many form tasks such as carrying mortar and shoveling
years. Work on such topics as memory for move- coal and in methods of improving the conduct of
ments, speed-accuracy trade-offs, and phase work in extremely hot environments; these stud-
transitions in bimanual movements appeared ies became the early contributions to the emerg-
 Evolution of a Field of Study 7

ing fields of human factors and ergonomics. Some of commands delivered to the muscles. Indeed,
early theories of learning were published (e.g., Sherrington's early writings (e.g., Sherrington,
Snoddy, 1935), and work by physical educators 1906) remain relevant today (see tributes to his
interested in sports and athletic performances work in Gallistel, 1980; Stuart, Pierce, Callister,
emerged (e.g., McCloy, 1934, 1937). An interest Brichta, & McDonagh, 2001).
in factors associated with growth, maturation, Sherrington was one of those involved in
and motor performance began to surface; and research on the perception of movement. Various
studies by Bayley (1935), Espenschade (1940), sensory receptors were identified, such as the
McGraw (1935, 1939), and Shirley (1931) led the Golgi tendon organ, thought to signal changes in
way to the formation of the subarea that we now muscle tension, and the muscle spindle, thought
call motor development (see Thomas, 1997, for a to be involved in the perception of muscle length
historical review). and hence joint position. Sherrington coined the
The evolution of the study of the physiological now-common term proprioception, which refers
or neural bases of movement paralleled work in to the sense of body position and orientation
the motor behavior area during this period, but thought to be signaled by the various muscle and
without much formal contact between the fields. joint receptors together with receptors located in
The characteristics and contraction properties the inner ear.
of muscle tissue were a topic of early study by Somewhat later, scientists conducted research
Blix (1892-1895) and Weber (1846; see Partridge, on various brain structures. Herrick (1924) pro-
1983), who identified "spring-like" properties of posed numerous hypotheses about the functions
muscle that were later "rediscovered." Jackson of the cerebellum, many of which seem at least
conducted early investigations of the neural reasonable today. Also, patients with accidental
control of movement in the 1870s, well before the cerebellar damage were studied (e.g., by Holmes
advent of electrophysiological techniques that [1939]) in an attempt to pinpoint some of the
were to revolutionize the field. But what led to movement-control deficits associated with this
the development of various electrophysiological structure. Other brain structures, studied in
methods was the discovery by Fritsch and Hitzig patients with various kinds of brain damage,
(1870) that the brain is electrically excitable. These became subjects of interest (Adrian & Buytendijk,
methods gave rise to studies by Ferrier (1888) on 1931).
the responses in the brain's cortex to movements, Early neural control research mainly involved
as well as to the work by Beevor and Horsely very simple movements. Indeed, experimenters
(1887, 1890) on sensory and motor areas of the sometimes isolated nerve-muscle preparations
brain. or used animals with various degrees of experi-
One of the more important influences in the mentally induced spinal cord damage; here the
neural control area was the work on reflexes at concern about movement was usually secondary
about the end of the 19th century by Sherrington to interest in the neurological processes. When
and his coworkers. Sherrington studied and clas- movements were studied, the movement was
sified the major responses to stimuli presented often not considered in much detail; and mea-
to the extremities, and he believed that most of sures of the speed, accuracy, or patterns of move-
our voluntary movements resulted from these ment were usually missing from these reports.
fundamental reflexes. Sherrington is credited The motor behavior work, on the other hand,
with the creation of a number of classical con- typically involved very complex actions (e.g.,
cepts of motor control, most of which influence typing, telegraphy) but with very little empha-
thinking today. For example, he first talked of sis on the underlying neural or biomechanical
reciprocal innervation, the idea that when the mechanisms that controlled these actions.
flexors of a joint are activated, the extensors tend We can see an exception to this general sepa-
to be automatically deactivated, and vice versa. ration of the neural control and motor behavior
Also, Sherrington coined the term final common areas in the research of two important physiolo-
path, which referred to the notion that influences gists in the 1930s and 1940s. During this period,
from reflexes and sensory sources, as well as Nikolai Bernstein and Erich von Holst published
from" command" sources in the brain, eventually a number of seminal papers that have had a
converge at spinal levels to produce the final set significant impact on motor control theorizing
8 111 Motor Control and Learning

today (for more on Bernstein see "Nikolai Bern- until translations of their work appeared in Eng-
stein"). Unfortunately, many scientists involved lish-Bernstein's work had been published in
in the study of movement, from both behavioral Russian, von Holst's in German. Their early papers
and neural control areas, were unaware of the reappeared in English in the late 1960s and early
contributions made by Bernstein and von Holst 1970s (see Bernstein, 1967, 1996; Whiting, 1984;

Nikolai A. Bernstein

B!!Pri~ted; ~y permission, .from Fizkultura Sports

\~ai~i, .• · · · PUblishers. ·

* ~.In'; Befl'lsteil1 used the degrees of

e.~~ly•stages Of skilf acquisition, tearn-
.t~e;nufober o1 degrees of freedom so
·· · · nee' of movement is seen when
ttledegrees of .freedom,
i;lrrent motor control theorizing is
j<~~~~li\<~\Qglish t~anstations of Bernstein's
· · .i~cy got:)tim1esJo this day.

tJ~forcJ: Perg.a.mon Press.

;~~T.Ilj~.e}f(Eds.), Dexterity

,<_;'-,;:':,-!'

· ~V.M. Zatsiorsky (Eds.),

tics.

.,l(\)sciMce:JnM.L. Latash &

.,Bh~ll'tpaign,.tL: Murnan .Kinetics.
 Evolution of a Field of Study m 9

vonHolst, 1937/1973; Gallistel, 1980). Thus, while In addition to the formal laboratories that were
the two areas were being blended in Russia and supported by defense funds, research relevant to
Germany, these trends were not seen in the United the military was given increased federal funding.
States or England, where most of the work on This funding, in the form of contracts, grants,
movement was being conducted. Ironically, it was and training programs, was responsible for a
the translation of this work many years later, and shift of attention among psychologists toward
the attention that it received (e.g., Turvey, 1977), motor behavior research. The directions imposed
that served as a significant catalyst to the merging by federal funding agencies had, and continue
of the neural control and motor behavior areas. to have, a profound influence on the behaviors
studied and the research questions asked. The
Postwar Research area of motor behavior was important at the time,
World War II had profound effects on the world, and a great deal of funding was directed toward
and it is not surprising that it also had major it, convincing a large number of psychologists to
effects on movement research. One of the earliest become interested in research in this area.
and most direct effects can be traced to the need to A second major influence in the creation of the
select the most suitable people for pilot training, boom in motor behavior research in the postwar
which resulted in the creation of the U.S. Army period was the emergence of various theories
Air Force's Psycho-Motor Testing Program, initi- of learning, most notably that of Hull (1943).
ated by Arthur Melton in the early stages of the In scientific inquiry, theories generally provide
war (see Melton, 1947, for a description of some an organization of the conceptual issues and
of this work). Important studies were conducted findings as well as strong suggestions for future
on underlying motor, perceptual, and intellectual research. Theories stimulate and provide focus
abilities as they related to the selection of pilots for the research of others, and Hull's theory was
and other military personnel (see chapter 9). no exception. His was a general learning theory,
Similar studies were conducted in England. In applying to animals and humans and to verbal
addition, scientists studied gunnery, physical and motor behavior, and it was often tested with
training in the heat and cold, vehicle control, and motor tasks. A major emphasis of the theory was
many other issues related to combat performance. the fatigue-like process associated with long
When the war ended in 1945, the prevailing practice periods. The theory attempted to explain
attitude in the United States was that the efforts how fatigue and recovery processes combined to
related to selection and training of military per- determine the learning of motor skills, and many
sonnel should not be abandoned. Consequently, scientists worked with motor tasks to test Hull's
this research continued for many years (e.g., predictions. Most of this work has relevance to the
Druckman & Bjork, 1991, 1994). The military distribution of practice (see chapter 11) or to the
research effort was sustained when Arthur effects of fatigue on performance and learning.
Melton created (in 1949) the U.S. Air Force Hull's theory later proyed to be an inadequate
Human Resources Research Center, which car- account of the processes and variables that deter-
ried on many of the wartime programs but also mine motor learning and performance. However,
expanded to include studies of more general theories like Hull's provide strong directions for
interest. A major contribution of this program was research and contribute experimental data for use
Fleishman's work on individual differences and by future generations, even though the original
abilities (e.g., Fleishman, 1965, 2004). The war- theory may be shown to be inadequate.
time programs, devoted to personnel selection As the complexity of machines increased in
and motor abilities, had not resulted in the suc- this period and industrial methods became more
cess in pilot selection that had been anticipated. complicated, it became obvious that the capabili-
Researchers began to realize that training-not ties of humans to operate machinery effectively
selection-was perhaps more important to the were being exceeded. For example, a number
development of proficient pilots. Hence, much of serious airplane accidents that were initially
attention was directed toward procedures for attributed to "pilot error" were eventually traced
teaching motor skills, the transfer of motor skills to the way in which the instruments and controls
from one activity to another, and the retention of in the cockpit were arranged (Chapanis, 1965;
skills (chapter 14). Fitts & Jones, 1947; Schlager, 1994). Thus, shortly
1O Iii! Motor Control and Learning

after the war, a study of man-machine interac- movements-their movement time, their move-
tions, variously termed human factors, ergonomics, ment extent, and their accuracy (see chapter 7 and
or engineering psychology (a subarea of industrial "Paul Fitts"). The discovery of these two laws of
psychology), emerged. The guiding concepts behavior was an important advance in research,
were that humans were an important component tied together by information theory (Seow, 2005).
in most of the machinery involved in industry, In the middle of this postwar period, a great
and that such machinery must be designed with deal of motor behavior research was being con-
humans in mind. Although this thinking began in ducted-enough that Robert and Carol Ammons,
the military, it is now seen in automobile design themselves researchers in this area, created a
(Lee, 2008), the organization of assembly lines journal in 1949 titled Perceptual and Motor Skills
and work spaces, the design of home appliances Research Exchange. 2 The journal now publishes
and computer workstations, and many other both motor and nonmotor research, but during its
areas (Chapanis, 1999; Jagacinski & Flach, 2003; early years it served as a major outlet for motor
Karwowski, 2001; Wickens & Hollands, 2000). behavior work. In addition, Research Quarterly,3 a
Professional societies and journals were founded physical education research journal, and the Jour-
in the mid-20th century and continue to flourish nal of Experimental Psychology4 published a great
today (Cooke, 2008; Waterson & Sell, 2006). deal of motor behavior research during this period.
This period also saw a great deal of experimen- Toward the end of the postwar period, the
tal effort in England. One of the most important number of psychologists interested in motor
contributions was by Craik (1948), who proposed behavior research gradually declined, while
that we consider the brain as a kind of computer the number of physical educators interested in
in which information is received, processed, and the study of motor skills greatly increased. The
then output to the environment in the form of psychologists' lack of interest may be attributed
overt actions of the limbs. An important part of to decreased federal support for motor behavior
this general idea is the notion of central intermit- research, disillusionment with Hull's theory, and
tency, by which the human movement is seen as a increasing interest in other types of human behav-
series of discrete bursts rather than as continuous ior such as verbal learning and memory. This
(as it might appear). Craik's idea paved the way trend away from motor behavior research reached
for other English psychologists such as Welford, its peak in the mid-1960s when an "academic
who in 1952 proposed the still-relevant single- funeral" sponsored by Ina and Edward Bilodeau
channel hypothesis of attention (see chapter 4). was held at Tulane University. Renowned motor
Also, a great deal of work was done in ergonom- behavior psychologists gathered to hear the
ics, on training and conditions of practice, and on "last rites" and to bid each other farewell as each
hand movement control, particularly with respect moved on to different research topics in psychol-
to anticipation and timing (Poulton, 1950). ogy. The eulogies were recorded in a volume titled
The ideas about central intermittency and Acquisition of Skill (~ilodeau, 1966), which well
the analogies of the brain to the computer were describes the attitude of the times.
accompanied by similar new directions in psy- Motor behavior research was dead, or so the
chology and related fields. One of the new ideas psychologists thought; but they did not consider
was represented by Wiener's (1948) book Cyber- a man named Franklin Henry, trained in psy-
netics, which outlined an information-processing chology and working in the Physical Education
basis for human behavior. Also, Shannon and Department at Berkeley, who had a continuing
Weaver's (1949) The Mathematical Theory of Com- interest in motor behavior research (see "Frank-
munication established important principles of lin Henry" on p. 12). Together with A.T. Slater-
information processing that later led to system- Hammel and other leaders in physical education,
atic attempts to study the motor system in terms these new motor behavior scientists organized
of its capabilities and limitations in processing the North American Society for the Psychology
information (see Hick's law in chapter 3). In of Sport and Physical Activity (NASPSPA) and
keeping with the information-processing basis the Canadian Society for Psychomotor Learning
for behavior suggested by Craik and others, and Sport Psychology. These groups flourished in
Fitts (1954) established some now-famous funda- the 1970s. During this period, two books devoted
mental relations among characteristics of aiming strictly to motor behavior and motor learning
 Evolution of a Field of Study 11

Paul Fitts
Probably no empirical discovery in the area of motor control is better
known than Fitts' law, which states that the time observed to complete
an aimed movement depends on a simple mathematical relationship
between the distance to move and the size of the intended target (see
chapter 7). Called "Fitts' law" out of respect for its originator, this for-
mulation was an early attempt to apply mathematical and information-
processing principles to the understanding of human movements, and
it suggested that more complex limb control could be understood by
future application of such methods and thinking. The mathematical
equation described in Fitts' law characterizes the trade-off between
speed and error during simple aiming movements in a way that has
remarkable generalizability.
But Fitts' law was just one of many legacies of a psychologist whose
research had many implications for activities of daily living, especially
FIGURE 1.2 Paul M. Fitts (1912- for equipment design. Fitts' early research on the effects of the spatial
1965). compatibility between work-space displays and the controls used in
responding to these displays had a profound influence on the then-
emerging field of ergonomics and human factors (see chapter 3). In later years, Fitts also wrote about
perceptual-motor learning, suggesting that learning involves a progression through various stages,
each with distinctive characteristics regarding the capabilities of the human to process information
(see chapter 13). Paul Fitts was widely regarded as a leader in his area of research when he died
unexpectedly in the mid-1960s at the age of 53, well before his full potential could be realized.

Selected Bibliography
Fitts, P.M. (1954). The information capacity of the human motor system in controlling the amplitude of
movement. Journal of Experimental Psychology, 47, 381-391.
Fitts, P.M. (1964). Perceptual-motor skills learning. In A.W. Melton (Ed.), Categories of human learning
(pp. 243-285). New York: Academic Press.
Fitts, P.M., & Seeger, C.M. (1953). S-R compatibility: Spatial characteristics of stimulus and response
codes. Journal of Experimental Psychology, 46, 199-210.

Further Reading
Pew, R.W. (1994). Paul Morris Fitts, 1912-1965. In H.L. Taylor (Ed.), Division 21 members who made
distinguished contributions to engineering psychology (pp. 23-44). Washington, DC: APA.

were published, one in England (Knapp, 1963) ing; and Keele (1968) wrote an often quoted
and one in the United States (Cratty, 1964). Many review of motor control (see "Steve Keele" on p.
more followed (including the first edition of this 13). But these were the exceptions. As the 1970s
book in 1982). approached, the cluster of scientists in physical
Not all psychologists of the period were bored education and (to a limited extent) psychology
with motor behavior research. Fitts and Peterson began to evolve in new directions. Posner and
(1964) presented influential experiments on limb Konick (1966) and Adams and Dijkstra (1966) pre-
movement accuracy; Bilodeau and Bilodeau sented seminal articles dealing with short-term
(1961), Adams (1964), and Noble (1968) wrote memory for movements; Henry and his students
needed reviews of motor behavior research; (e.g., Henry & Rogers, 1960) were interested in
Adams (1968) wrote a theoretical treatment of motor programs; Posner (1969) studied attention
the role of sensory feedback in movement learn- and movement control; Pew (1966) examined
12 111 Motor Control and Learning

6~P~\~edormattceson two or more tasks, even

:i1as~~rn~l'Jiiu~derlying ability (e.g., dynamic vs. static
~ 1 :~~#0Qtbtal 1 StLJclents who subscribed to his general
· p€>$iUo~§lff Physical.education .departments during
;~f th~se j!>cholars created Ph.D programs and trained
:'~.resulfthc:ll Henry's influence became pervasive by

1
:n~~i~lea~hi~grnotdrskilL In R.C.. Brown·& G.S. Kenyon
i~~p.3lJ.1~340);.!Englewood Cliffs, NJ: Prentice Hall.

'<~s~~1afet)by•fot complicated movements and a

~S:'e~t~h <f71ul1tterty, $1,448-458.

~~~~~~riffien~'6rioPeraonal perapectives
tp~1Q'rd~ye!()prh~nt, and sport psychology. Champaign,

practice and automaticity; and Adams (1971) initi- were being made in neural control that would
ated a return to theorizing about motor learning. later be influential in joining the two areas. One
These emphases provided strong leadership for of the more important contributions was the
the motor behavior area in the 1970s. work on muscle spindle mechanisms by Merton
As in the early period, the neural control and (1953; Marsden, Merton, & Morton, 1972), to
motor behavior scientists were nearly oblivi- be discussed in chapter 5. While the specific
ous to each other; but important contributions mechanisms proposed by Merton now appear
 Evolution of a Field of Study fill 13

Steve Keele
The 1960s marked a critical time in the history of motor qehavior research,
as many psychologists shifted their focus to issues of cognition\ such as
memory and attention. Psychology appeared to have abandoned the
study of motor behavior, with a few notable exceptions. The appearance
of Keele's motor control literature review in 1968 created renewed inter~
est in important issues such as the motor program and the role. of vision
in movement regulation. His paper (with Michael Posner) on the period
during a movement that is required to process and act on visual information
remains one of our favorites for its simple, elegant methods of answering
a difficult question. Keele's book, Attention and Human Performance,
published several years later, situated movement prominently within a
late-filter theory of attention, reminding psychologists that movement is
not just a simple "output" in the human-computer metaphor (Rosenbaum,
FIGURE 1.4 Steve Keele
(1940-2005). 2005). Updated and expanded reviews of motor control ih later years
Courtesy of Betty Jean Keele. were considered mandatory reading for students in this area. Keele's later
research interests included important contributions to our understanding
of individual differences, timing, and sequence learning.

Selected Bibliography
Keele, S. W. (1968). Movement control in skilled motor performance. Psychological Bulletin, 70, 387-403.
Keele, S.W. (1973). Attention and human performance. Pacific Palisades, CA: Goodyear.
Keele, S.W. (1986). Motor control. In K.R. Boff, L. Kaufman, & J.P. Thomas (Eds.), Handbook of per~
ception and performance (pp. 30.1-30.60). New York: Wiley.
Keele, S.W., Jennings, P., Jones, S., Caulton, D., & Cohen, A. (1995). On the modularity of sequence
representation. Journal of Motor Behavior, 27, 17-30.
Keele, S.W., & Posner, M.I. (1968). Processing of visual feedback in rapid movements. Journal of
Experimental Psychology, 77, 155-158.

Further Reading
lvry, R.B., Mayr, U., Corcos, D.M., & Posner, M.I. (2006). Psychological processes and neural mecha-
nisms for action: The legacy of Steven W. Keele. Journal of Motor Behavior, 38, 3-6.
Keele, S.W. (1983). This week's citation classic. Current Contents, 37 (Sept. 12), 20.

to be incorrect (Houk, 1979; Smith, 1977), Mer- joint angles, suggesting that these receptors have
ton's original ideas about automatic regulation a large role in the perception of joint position.
of movement are reasonable in very general Numerous studies on the nature of muscle and
terms. Merton was one of the first to measure its contractile and mechanical (e.g., spring-like)
movements and neurophysiological processes in properties were also completed during these
the same investigation, creating a beginning for postwar years, and these studies attracted the
a blend of behavior and neurological emphases attention of contemporary researchers in motor
that was to follow. At about the same time, a great behavior and motor control (Rack & Westbury,
deal of research was devoted to the sensory recep- 1969). These mechanical characteristics of muscle
tors associated with movement perception and and of the motor apparatus were utilized by scien-
kinesthesis. Skoglund (1956) published a classic tists in the Moscow laboratories who were follow-
paper showing that the various receptors in a joint ing the earlier tradition of Bernstein. The exten-
capsule appear to be activated at certain specific sive work on movement control by this group,
14 Motor Control and Learning

originally published in Russian and thus generally Led by such researchers as Adams and Dijkstra
unknown to American and British researchers, (1966) and Posner and Konick (1966), the process
attracted a great deal of attention through vari- orientation helped to create the area of short-
ous translations (e.g., Gelfand, Gurfinkel, Tomin, term motor memory-the study of the processes
& Tsetlin, 1971; Kots, 1977). This research has underlying memory loss in simple movements
special relevance for the control of locomotion over short periods of time. Many studies were
and provides important links between the neural conducted in this area during the late 1960s and
control mechanisms and behavioral principles. early 1970s (see chapter 14). Studies were also
But despite these efforts, by 1970 almost no asso- completed on information-processing activities
ciation existed between the behavioral scientists during the learning of simple motor tasks (see
interested in more global and complex skills and chapter 12).
the neurophysiological scientists interested in More importantly, theorizing returned to motor
simple movements and neural control. behavior and learning, a style of inquiry that
had been relatively dormant since the failure of
The End of the Century Hull's (1943) theory. Adams sparked the interest
in theory when he presented a feedback-based
The 1970s brought massive changes in the field theory of verbal learning (Adams & Bray, 1970),
of movement control and learning. The strict followed the next year by a similar theory devoted
stimulus-response (S-R) orientation that had had to motor learning (Adams, 1971) (see "Jack
such a strong foothold during most of the century Adams"). Pew (1974a) returned to the old idea of
was overshadowed by the cognitive, information- a movement schema (Bartlett, 1932)-the abstract
processing approach. The publication of two hypothetical structures responsible for movement
books during the 1960s-Miller, Galanter, and control and evaluation, to be discussed in chapter
Pribram's (1960) Plans and the Structure of Behavior 13. And, one year later, the schema theory for the
and Neisser's (1967) Cognitive Psychology-had a learning of simple motor skills was presented
large impact on the field of experimental psychol- (Schmidt, 1975b). Together, these theoretical ideas
ogy in general and, later, on motor behavior too. generated a great deal of interest in motor skills,
The move toward cognitive psychology was a as this text makes evident later.
reaction to S-R theories of behavior. Ideas about The motor behavior field not only changed its
mental and motor processes, together with many direction, but also grew rapidly. Formal courses of
methods and paradigms for understanding them, study in universities flourished, and new journals
took the place of S-R theories. Perhaps more than appeared. In 1969, Schmidt founded the Journal
anything else, the books by Miller and colleagues of Motor Behavior, which was closely followed in
and Neisser popularized the study of mental pro- 1975 by the Journal of Human Movement Studies,
cesses such as response selection and movement created by the English motor behavior scientist
programming, whose existence must be inferred John Whiting (see "H.T.A. (John) Whiting" on
from the behaving individual rather than directly p. 16). A review journal titled Exercise and Sport
observed (see also Baars, 1986; Miller, 2003). Sciences Reviews was created in this period, and
Influenced by cognitive psychology, the motor it devoted a major portion of its space to motor
behavior field seemed to undergo a transition behavior research. Two more journals devoted to
from a task orientation, which focuses primarily the study of motor control also appeared before
on the effects of variables on the performance or the turn of the century; Human Movement Science
learning of certain motor tasks (or both), to a pro- appeared in 1982 and Motor Control in 1997. And
cess orientation, which focuses on the underlying throughout this time, the psychological journals
mental or neural events that support or produce (e.g., Journal of Experimental Psychology: Human
movements (Pew, 1970, 1974b; Schmidt, 1975b, Perception and Performance, Psychological Research,
1989a). Humans were considered processors of Psychological Bulletin, British Journal of Psychology,
information, and this approach was an attempt Psychological Review, Human Factors, and Ergonom-
to understand how movement information is ics) continued to publish motor behavior research.
coded and stored, how actions are represented As the field grew, motor behavior textbooks
in memory, and how information about errors is proliferated. More than 30 textbooks written
processed so that learning can occur. subsequent to Knapp's (1963) and Cratty's (1964)
 Evolution of a Field of Study 15

Jack Adams
One of the landmark advances in the developing field of motor
learning was the publication of '7\ Closed-Loop Theory of Motor
Learning" in the Journal of Motor Behavior in 1971. This paper
described one of the very first theories directed specifically at
motor learning and presented numerous testable hypotheses that
became the focus of a considerable number of research studies in
the 1970s. Although many of the tenets of closed-loop theory were
later found to need revision (see chapter 13), which is the case
for most theories that are scrutinized carefully, Adams' theory was
clearly a catalyst that moved the research "bar" in motor learning
much higher. But the 1971 theory was only one of many contribu-
tions to motor learning research that Adams made in a long and
fruitful career. His early work on psychological warm-up effects
as a source of memory decrement in practice (e.g., Adams, 1961)
led to one of the very first of many investigations on short-term
FIGURE 1.5 Jack A. Adams (1922-
2010). memory for movement information (Adams & Dijkstra, 1966).
Courtesy of Jack Adams. Throughout his research career, Adams maintained an interest in
the application of motor learning research in the area of human
factors and ergonomics, beginning with his work at Lackland Air Force Base in Texas. Later he turned
his attention to the effects of modeling and observational learning (e.g., Adams, 1986), again spear-
heading a rise in research activity in that area. Perhaps his most lasting legacy, however, will be as a
research historian-Jack Adams' ability to accurately and concisely synthesize volumes of research
history into manageable chunks of understandable knowledge will serve students of motor control
and learning for many years to come. His 1964 paper in the Annual Review of Psychology and the
1987 paper in the Psychological Bulletin remain as landmark reference works in motor behavior.

Selected Bibliography
Adams, J.A. (1961). The second facet of forgetting: A review of warm-up decrement. Psychological
Bulletin, 58, 257-273.
Adams, J.A. (1964). Motor skills. Annual Review of Psychology, 15, 181-202.
Adams, J.A. (1971). A closed-loop theory of motor learning. Journal of Motor Behavior, 3, 111-150.
Adams, J.A. (1987). Historical review and appraisal of research on the learning, retention, and transfer
of human motor skills. Psychological Bulletin, 101, 41-74.

further Reading
Schmidt, R.A. (1990). Distinguished scholar award to Jack Ashton Adams. NASPSPA Newsletter, 15, 4-5.
Schmidt, R.A. (2011 ). Jack Adams, a giant of motor behavior, has died. Journal of Motor Behavior,
43, 83-84.

work were published, as were a large number of about movement control and made increased
edited volumes on more specific topics. use of various electrophysiological and biome-
The 1970s were the beginning of a long-needed chanical techniques to understand the functions
merger between the neural control and the motor of the central nervous system in movement. The
behavior scientists. Many people were trained for- neural control scientists were shifting from stud-
mally in both motor behavior and neural control, ies that examined only the neural mechanisms to
and these people completed the bridge between studies investigating these mechanisms during
the two levels of analysis. More and more behav- complex movements. Much of this latter work
ior-oriented scientists began to ask questions was done with animals, principally monkeys
16 Motor Control and Learning

H.T.A. (John) Whiting

Another pioneer who had tremendous influence on motor
behavior research was H.T.A. (John) Whiting. Like Franklin
Henry in the United States, discussed earlier in the chapter,
Whiting introduced traditional psychological techniques to
the study of human motor control in the performance of sport
tasks. His research on catching performance was the first to
demonstrate that continuous vision of a ball in flight is not
necessary in order to support accurate performance (see
chapter 5). Whiting's book, Acquiring Ball Skill (1969), was
one of the first psychologically oriented texts to be devoted to
the study .of motor skills. Later Whiting turned his attention to
the study of learning in the Bernstein tradition (see chapter
13). During this time Whiting al.so founded and served as the
editor of the Journal of Human Movement Studies and, later,
Human Movement Science. Another important legacy was
the many graduate students Whiting supervised at Leeds
during the 1960s and 1970s, and then at the Free Univer-
sity of Amsterdam in the Netherlands during the 1980s and
FIGURE 1.6 H.T.A. (John) Whiting (1929-
1990s. Many of these scientists continue to carry on Whiting's 2001).
legacy today. Courtesy of Vrije Universiteit.

Selected Bibliography
Vereijken, B., Whiting, H.T.A., & Beek, W.J. (1992). A dynamical systems approach to skill acquisition.
Quarterly Journal of Experimental Psychology, 45A, 323-344.
Whiting, H.T.A. {1969). Acquiring ball skill: A psychological interpretation. London: Bell.
Whiting, H.T.A., Gill, E.B., & Stephenson, J.M. (1970). Critical time intervals for taking in flight informa-
tion in a ball-catching task. Ergonomics, 13, 265-272.

Further Reading
Savelsburgh, G., & Davids, K. (2002). "Keeping the eye on the ball": The legacy of John Whiting (1929-
2001) in sport science. Journal of Sport Sciences, 20, 79-82.
Snyder, C., & Abernethy, B. (1992). The creative side of experimentation: Personal perspectives from
leading researchers in motor control, motor development, and sport psychology. Champaign, IL:
Human Kinetics. ·

and cats. Records from electrodes implanted behaviors and neurological processes in order to
in the brain, spinal cord, or muscle were taken provide a more complete understanding of how
while the animal was engaged in motor activity. movements are controlled. This emphasis marked
Representing this approach are Grillner and his a refreshing change from the earlier research in
colleagues (1972, 1975) and Smith and her col- which the movements per se were hardly consid-
leagues (1986), who studied locomotion in cats; ered. The association between motor behavior and
Evarts (1972, 1973), who studied a number of motor control resulted in several reviews written
separate brain structures in monkeys; and Houk toward the end of the 1970s, such as those by
(1979) and Granit (1970), who studied the gamma Brooks (1975, 1979, 1986), Grillner (1975), Wetzel
motor system in monkeys and humans. and Stuart (1976), and Gallistel (1980). Behaviorists
The essential feature of all this work is the strong and neurophysiologists participated in a number
attempt to find an association between movement of scientific meetings, and the results appeared in
 Evolution of a Field of Study !I'! 17

edited volumes (e.g., Gandevia, Proske, & Stuart, The late 1970s and early 1980s were also char-
2002; Stelmach & Requin, 1980, 1992; Swinnen, acterized by a general decline in interest in motor
Heuer, Massion, & Casaer, 1994). learning, with a corresponding increase in issues
An additional change occurred toward the of movement control or human performance. This
end of the century-one far more subtle than was unfortunate, because the issues involved in
those just mentioned. Rather than remaining a learning have perhaps the most practical appli-
mere blending of two different fields, the field cation to training, rehabilitation, and teaching
of motor control acquired an independent iden- in general. But there was a renewed interest in
tity. It became a field of study in its own right, learning toward the end of the century, sparked
complete with its own journals and methods for in part by counterintuitive findings regard-
asking research questions and collecting data. ing how practice scheduling (Shea & Morgan,
Such methods involve the use of sophisticated 1979; Lee & Magill, 1983b; Magill & Hall, 1990)
techniques for recording and analyzing move- and augmented feedback (Salmoni, Schmidt, &
ments (such as electrophysiological recordings), Walter, 1984; Schmidt, 1991a) could be organized
cinematographic and three-dimensional analyses, to optimize the learning environment. Much of
measurement of the kinematics of movement, and this work is described in chapters 11 and 12.
advanced methods for examining the involve-
ment of brain structures, integrated with the more Motor Control and Learning
traditional techniques for studying learning (e.g., Research Today
Corcos, Jaric, & Gottlieb, 1996; see also chapter 2).
The influence of Bernstein (and others) resur- The integration of motor control research that
faced in the writings of a number of scientists who developed toward the end of the 20th century
conducted motor control research (e.g., Greene, continues today. For example, behavioral studies
1972; Kelso, 1995; Kugler & Turvey, 1987; Reed, of motor control and learning now appear in jour-
1988; Turvey, 1977). According to Turvey (1990), nals that in the past were strictly oriented toward
Bernstein's legacy resulted in two rounds of theo- neuroscience (e.g., Experimental Brain Research,
rizing and experimentation. The first round dealt Neuroscience Letters, Journal of Neuroscience), and
with the degrees of freedom problem-research neuroscience-oriented studies appear in journals
addressing how a system with many independent that were formerly primarily behavioral (e.g.,
parts could be controlled without the need for an Journal of Motor Behavior, Human Movement Sci-
executive "decision maker." The second round ence). Studies of the specific brain mechanisms
extended Bernstein's thinking on coordination involved in the performance of movement tasks
and the degrees of freedom problem to a search have appeared in specialist journals (e.g., Neuro-
for laws and principles of self-organization. Much Image) and represent a rapidly growing approach
of this work uses physical biology as its basis. The to the study of motor control. Studies of motor
dynamical-systems perspective (e.g., Kelso, 1995) control and learning also continue to be published
suggests that coordinated movement evolves in journals devoted to specific professional topics
over time as a function of the interaction among (e.g., Physical Therapy, Human Factors), and this list
the body parts, and between the body parts and continues to expand to include journals that did
the physical world. Also associated with this not previously publish motor behavior articles
view are the ideas that perception and action are (e.g., American Journal of Surgery). New societies
functionally inseparable-that understanding (e.g., International Society of Motor Control) and
the motor system depends on understanding the journals (e.g., Motor Control, Frontiers in Movement
physical principles of our actions and how they Science and Sport Psychology) have emerged. In
interact with biological functions. Advocates short, more research in the area of motor control
of these traditions showed a reluctance to use and learning is being published now than ever
cognitive-psychological styles of inquiry with before. And that research is being conducted by
hypothetically defined brain mechanisms such as researchers who have a greater diversity and
memory, motor programs, schemas, and the like. breadth of knowledge than ever before.
This approach contributed a different emphasis to The computer, employed in relatively few
the attempt to understand motor behavior (e.g., laboratories for research investigations prior
Anson, Elliott, & Davids, 2005). to the 1980s, is now used almost universally
18 Motor Control and Learning

for conducting research and organizing data. less of the changes that occur, however, the lessons
Analyses that used to take many hours can now of history tell us that the scientific disciplines
be performed in a matter of minutes. In short, the combining to define motor control and learning will
capacity to conduct research of very high quality continue to undergo refinement with further accu-
has grown at a dizzying rate. mulation of knowledge and application.
This is an exciting time for research in motor
control and learning, an area of literature now so
broad in scope that it would seem to be more and Summary
more difficult for the student to remain aware of
the most recent publications. Ironically, however, This text is fundamentally concerned with move-
it has never been easier to retrieve research infor- ments of human beings. Some of these movements
mation in this area. The Internet, which did not are probably genetically defined while others
even exist at the time the second edition of this are skills, requiring practice or experience. Even
book was published, has revolutionized the way though most of the field of human behavior deals
in which motor behavior research is made public. in one way or another with movement, in this text
E-mail alerting services deliver the tables of con- we focus primarily (but not exclusively) on those
tents of new journal issues as soon as they are movements in which cognitive involvement is
published and previews of soon to be published relatively slight, and for which the nature of the
papers. Almost all journals are now available in movement itself-rather than the choice of the
electronic formats, and the articles are available movement from already learned alternatives-is
to consumers at the click of a mouse. Moreover, the primary determinant of success. We focus on
many journals are not only publishing current movements that do not have a heavy concentra-
issues in electronic formats but also archiving tion on cardiovascular endurance or strength, as
all of their pre-electronic issues. General search these activities seem to be more closely aligned
engines (such as Google) and specialized search with other fields of study. Finally, the focus is on
engines (such as Google Scholar, PubMed, Web many different movements that fall within the
of Science) make the literature retrieval for a categories mentioned, such as those in musical
particular topic, author, or paper very fast and performance, work, industry, and sport and other
effective. The Internet has also made databases activities of daily living.
easy to access. Carlton and colleagues (1999) The field of movement control and learning,
described a Web-based laboratory, for example, viewed from a historical perspective, emerged
where a student can retrieve and analyze a set of from the separate but parallel fields of motor
data that addresses a particular research topic. behavior and neurophysiology. Both fields
However, the student of motor behavior must showed steady growth through the beginning
also be aware, now more than ever, that what of World War It then increased in growth and
is available online can be of variable quality. sophistication after the war and through the 1960s
Respected journals maintain the peer review and 1970s. The two.fields, however, were largely
process, whereby a paper submitted for publica- separated until the early 1970s, when they began
tion undergoes careful scrutiny by a small com- to share common problems and methods. Motor
mittee of established peers ("referees") before it learning research, largely forgotten during the
can appear in press under the journal's title. This '70s, reemerged as a strong research area toward
scrutiny is maintained regardless of whether the the end of the century. Today the computer (and
published work appears in paper or electronic the Internet) has revolutionized the way in which
format. However, not all articles that one can research is conducted and disseminated.
retrieve on the Internet have undergone this peer
review process. As in the previous centuries, the
best and most influential research articles are
Student Assignments
almost always those that have undergone careful
and rigorous scientific scrutiny. 1. Answer the following questions and bring
It is impossible to know what changes will the information to class for discussion:
emerge in the next decade. Perhaps these words a. How does a behavioral level of analysis
will no longer be available in paper format. Regard- of motor control and learning differ
 Evolution of a Field of Study m 19

from biomechanical and neurophysi- www.i-s-m-c.org

ological levels of analysis?
These journals specialize in publishing research
b. If you could interview one of the people in motor control and learning:
featured in the highlight boxes (Bern-
stein, Fitts, Henry, Keele, Adams, Whit- www.tandf.co.uk/journals/titles/00222895.asp
ing), what would you ask him about
www.humankinetics.com/MC/journalAbout.
motor control and learning?
cfm
c. What has the proliferation in publica-
www.elsevier.com/wps/find/joumaldescription.
tion outlets contributed to our current
cws_home/505584/description#description
understanding of research in motor
control and learning?
2. Find a research article authored by any of Notes
the six people listed in question lb. Are the
findings of that research article still relevant 1 Learning the rules (grammar) of a language would

today? probably not be of much relevance for the area of motor

control, but learning to make the guttural sounds involved
in German or the nasal sounds inherent in French could
logically be included in the area of motor control (Mac-
Web Resources Neilage, 1970).
2 Now called Perceptual and Motor Skills.
These are some of the professional organizations 3 Now called Research Quarterly for Exercise and Sport.
that promote research in motor control and learn-
4 Now subdivided into five different journals (one of
ing through newsletters and annual conferences:
which is entitled Journal of Experimental Psychology: Human
www.naspspa.org Perception and Performance).
www.scapps.org
 major goal of this book is to present not (i.e., the class of task) under consideration. That
only relevant principles and theories about is, the relation between certain independent
the nature of motor performance, motor control, and dependent variables is often different for
and motor learning but also the research evidence one class of task or behavior when compared to
that supports (or, in some cases, refutes) these another. Without classification, the laws of motor
principles and theories. In evaluating this evi- control would be far more difficult to understand.
dence, it is necessary to understand some of the Movement behaviors have been classified
methods involved in the research and the ways in in various ways. Two important classification
which the motor behaviors are measured, so that schemes are the discrete I continuous I serial
we can more effectively establish the relevance of dimension, which is based on the particular move-
the evidence to the particular principle or theory ments made, and the open/ closed dimension,
in question. Later in the book (chapter 10), we which is determined by the perceptual attributes
focus on the methods and paradigms used spe- of the task.
cifically in the study of motor learning.
Discrete, Continuous,
Classification of Motor Skills and Serial Skills
Discrete movements are those with a recognizable
In any field of study, the objects under investi- beginning and end. Kicking a ball, throwing,
gation are usually classified according to some striking a match, and shifting gears in a car are
scheme or framework in order to simplify discus- examples (figure 2.1). The end of the movement is
sion. The field of motor behavior is no exception. defined by the skill in question, not arbitrarily by
Classification of movements and motor tasks is the time at which an observer ceased examining
important for two fundamental reasons. First, in it, as would be the case for swimming or jogging,
the research literature on motor behavior and con- for example. Discrete skills can be very rapid,
trol, various terms are used to describe the tasks requiring only a fraction of a second to complete
and movements. These terms must be understood (e.g., kicking, blinking an eye); but they can also
if we are to communicate about the field. The require much more time for completion, as in
second reason is that the laws of motor behavior writing your signature. Discrete skills can also be
seem to depend on the kinds of performances quite cognitive. For example, one laboratory task

Ill 21
22 Motor Control and Learning

is to press one of four buttons when one of four or an uncertain way on the screen. The second
lights comes on; the problem for the subject is to cursor is moved by the subject via a hand control,
decide which button to press in response to which and the subject's task is to minimize the distance
light. Thus, the decision about which button to (or error) between the two cursors.
push is paramount, and the "how" of pushing the Two kinds of tracking tasks are used com-
button is clearly secondary in importance. While monly in motor behavior research: pursuit track-
many discrete skills have large verbal-cognitive ing and compensatory tracking. In pursuit tracking,
components, there are certainly examples of dis- experimenter-produced actions of the target and
crete skills that are highly "motor" as well. the subject's own movements are both displayed.
Continuous movements-defined as those that The previously mentioned task of steering a car is
have no recognizable beginning and end, with a good example of pursuit tracking. In compensa-
behavior continuing until the movement is arbi- tory tracking, the experimenter-produced varia-
trarily stopped-are at the opposite end of the tions in the track are combined with the subject's
continuum (in figure 2.1). Swimming, running, movements to produce a single displayed value,
and steering a car are examples of tasks that have and the subject's goal is to maintain this value
arbitrary ends. Continuous tasks tend to have at some constant location. Practical examples of
longer movement times than do discrete tasks compensatory tracking are often seen in aircraft
(they might even continue all day). This, however, instruments, such as the glide slope indicator.
should not be taken as basic to their definition. Here only the difference between the proper
A common class of continuous skills, both in altitude and the actual altitude is displayed; and
everyday experience and in the laboratory, involves when the pointer is in the middle of the screen,
tracking tasks. The tracking task is characterized the pilot's altitude is correct. Compensatory track-
by a pathway (track) that the individual intends ing tasks are almost always more "difficult" than
to follow and a device that the person attempts to pursuit tracking tasks, particularly if the behavior
keep on the track via certain limb movements. In of the track is irregular and unpredictable.
steering a car, for example, the track is the road, Tracking tasks also vary in terms of the aspect
and the device is the car, steering wheel, and so of the display that the subject controls. The most
on. A very common laboratory example involves simple is the zero-order, or positional, display. If
two cursors on a computer monitor. One of the the subject moves the handle from one position
cursors is moved by the experimenter (or by the to another and then stops, the indicator on the
computer), and it can move in either a predictable display moves a proportional amount and also

FIGURE 2.1 The discrete/serial/continuous classification for motor behavior.

 Methodology for Studying Motor Performance Ill 23

stops; that is, the handle movements control the starting a car, preparing and lighting a wood
position of the pointer. In first-order, or velocity fireplace, and many tasks involved in production
control, movement of the handle causes changes lines in industry. Serial tasks can be thought of as
in the velocity of the pointer. Moving the handle a number of discrete tasks strung together, and
further in one direction causes the velocity of the order (and sometimes timing) of the actions
the pointer to increase in the same direction, and is important.
stopping the handle movement off center results
in a constant velocity of pointer movement. In a Open Versus Closed Skills
second-order task, the movements of the control Environmental predictability during the per-
produce changes in the pointer's acceleration. formance provides another basis for classifying
Keeping the handle centered produces zero movement skills (Poulton, 1957; Gentile, 2000).
acceleration, but moving the handle to a new Open skills are those for which the environment
position off center accelerates the pointer in the is constantly (perhaps unpredictably) changing,
same direction. Each of these kinds of tracking so that the performer cannot effectively plan
tasks is used in research, and there are real-world the entire movement in advance (figure 2.2). A
examples of each in various control systems (see good example is the penalty shot in ice hockey.
Poulton, 1974, for more details). While skating toward the goalie, the player may
One final type of tracking task is step tracking. make a general decision about whether to go
In this task, the track "jumps" from one fixed left or right, but the final decision may depend
location to another, often unpredictably, and the on what the goalie does. Another example is
subject's task is to move the control as quickly driving on a busy freeway. Although you may
as possible to correct this sudden change in the make a general plan about what you want to do,
track's location. Step tracking tasks can be either such as pass another car, your precise plans must
pursuit or compensatory. be left flexible enough to deal with unexpected
Serial movements are neither discrete nor con- actions of other drivers. Success in open skills is
tinuous, but usually are comprised of a series of largely determined by the extent to which the
individual movements tied together in time to individual is successful in adapting the planned
make some "whole." These types of movements motor behavior to the changing environment.
appear in the center of the continuum in figure Often this adaptation must be extremely rapid,
2.1 because they can be rather long in duration and the effective responder must have many dif-
but are not stopped arbitrarily. Examples are ferent actions ready to implement.

FIGURE 2.2 The open/closed continuum for motor behavior.

24 Motor Control and Learning

Closed skills, for which the environment is pre- Objectivity

dictable, are at the other end of the continuum
shown in figure 2.2. An environment may be pre- The term objectivity is important in measurement
dictable because it is almost perfectly stable-for because of the scientific demand that observations
example, the environment in which one performs be subject to public verification. A measurement
skills like archery, bowling, or signing one's name system is objective to the extent that two observ-
to a check. An unstable yet predictable situation ers evaluating the same performance arrive at
can also arise when the environment is variable the same (or very similar) measurements. For
but the changes are very predictable, or have been example, using a tape measure to determine the
learned as a result of practice, or both; examples distance a javelin was thrown yields very simi-
are juggling and industrial production-line tasks. lar results regardless of who reads the tape. By
Here, the essential feature is that the environment comparison, evaluation of performances such as
for the next few seconds or so is essentially pre- diving, gymnastics, and figure skating is more
dictable, so that the movement can be planned subjective-although elaborate scoring rules,
in advance. Of course, some skills have environ- complete with certification tests for judges, help
ments that are semi-predictable, and these can make it more objective. From the point of view
be classified somewhere between the ends of of research in motor behavior, it is important to
the open/ closed continuum in figure 2.2. Farrell use performances in the laboratory for which the
(1975) has provided additional distinctions that scoring can be as objective as possible; and this
help us to classify movements on this dimension. necessarily limits the usefulness of some tasks for
The observation that open skills seem to require providing an understanding of motor behavior
rapid adaptations to a changing environment, in general.
whereas closed skills require very consistent and A second aspect of objectivity relates to the
stable performances in a predictable environ- sensitivity of the measuring device to changes in
ment, raises interesting questions about how the the skill of the performer. How high did the girl
two classes of skills might best be learned and jump when she set the school record in the high
taught. Should the methods for teaching open and jump? The record books report that she jumped
closed skills be different? Do different individuals 5 ft 6 1/2 in. (168 cm). But it is more accurate
perform better in one of these skill classes or the to say that she jumped at least 5 ft 6 1/2 in. The
other? Are the laws of performance different for measurement system for the high jump (and for
the two kinds of skills? Evidence suggests that the pole vault) indicates only whether or not the
some of the answers to these questions are yes, bar was knocked off during the jump, and it is
and we will discuss these issues in more detail possible that the successful jumper cleared the
in later chapters. bar by 6 in. In such situations, the scoring system
is somewhat insensitive to the variations in the
performer's actual l~vel of skill.
Basic Considerations In this example, the scale of measurement itself
in Measurement was acceptable in terms of sensitivity; in fact, we
often see high jump officials measuring the height
A fundamental issue in any science concerns of the bar very carefully, perhaps to the nearest
how the behaviors of the objects of study are 1/ 4 in. Yet this method is not totally objective, as
measured, and motor behavior is no exception. we have just argued. However, sometimes the
We often need to operationalize skills so that we scale of measurement itself is lacking in preci-
can assign numerical values to certain perfor- sion. Often, continuous measures are artificially
mances based on the quality of the movements. categorized, as in hit/ miss scoring for basketball
Scientists must be able to measure the degree of and golf putting. For these examples the details of
skill exhibited by a performer in scientifically the movement, together with information about
acceptable ways. Some of the criteria for good the performer's skill level, can be lost in the over-
measurement systems are (a) the objectivity of simplified measurement method. Such methods
the system, (b) the reliability (or stability) of the make it more difficult to determine whether an
measuring system, and (c) the validity of the individual has improved on the task with prac-
system. tice, or to identify which of two individuals is
 Methodology for Studying Motor Performance 25

the more skilled performer-both of which may practical situations, but the alternative of study-
be critical questions in the study of movement. ing skills during a game adds sources of variation
that reduce the reliability of the measures.
Reliability The procedures mentioned can reduce variabil-
A second aspect of the measurement system that ity in experimental settings. But even when the
is important to motor behavior is reliability-the task is well learned and simple, when the experi-
extent to which the measurement is repeatable mental situation is well controlled, and when the
under similar conditions. A lack of reliability can subject is trying to do well, there remains a great
result from random technological error, such as deal of variability because biological systems
the stretch in measuring tapes, errors in clocks, are inherently somewhat unstable. Experimen-
and human errors in reading instruments. These tally, the best method for countering this type
errors, while they might seem to be important of variability is to record many observations of
sources of unreliability, probably contribute the "same" behavior on the same subject, taking
very little to unreliability as long as a researcher the average of a large number of measurements
uses quality recording apparatus and careful under essentially identical conditions. With this
procedures. The most important source of unreli- procedure the variations in the subject's perfor-
ability manifests itself when the performer does mance tend to "average out," raising the reliabil-
not perform the same action twice in exactly the ity of the measurement system, so that the mean
same way. Some of these intrasubject variations of a large number of observations more closely
are caused by momentary changes in the internal represents the construct being measured.
state of the subject (degree of attention, fatigue, or
boredom, for example), while others are caused
Validity
by systematic changes, such as alterations in strat- Another aspect of the measurement process is
egy, the amount of practice, and the like. Both of validity, the extent to which the test measures
these factors tend to obscure the constructs that what the researcher intends it to measure. An
scientists are attempting to measure. important aspect of validity (called construct
Experimenters seek to minimize these sources validity) is the extent to which the measures
of variability through experimental control in taken actually reflect the underlying construct
the testing situation. Researchers typically use of interest. We would be reasonably comfortable
written or prerecorded instructions in order to with a 10 min typing test to operationalize typing
eliminate variability in what and how informa- skill, but we would perhaps be less comfortable
tion is presented; they use testing rooms that are with a measure of finger tapping speed to assess
either silenced or sound deadened; subjects are typing skill. There are, on the other hand, situa-
tested one at a time to eliminate variability due tions in which validity does not seem to present
to another person's presence in the room; and the much of a problem. One of these involves what
entire experimental session is often quite formal are often called face valid tests, which are so obvi-
and impersonal. This is the primary reason some ously measures of the concept of interest that
researchers tend not to measure skills in every- they usually are not questioned. For example, if
day settings-at a ball game or on an industrial we wish to determine which member of a group
production line, for example. In these situations, of individuals has most skill in javelin throwing
the environment is not well controlled; there are (a construct), we might have them all throw the
many sources of variation from other players or javelin as a standardized test.
workers, from changes in the score of the game, Another class of measurement situations in
from the level of proficiency of opponents, from which the importance of validity is minimal is in
day-to-day changes in the weather, and so on. experiments on motor learning (Schmidt, 1989a).
Primarily for reasons of experimental control, In these situations, an arbitrary task is created
motor behavior tends to be most profitably stud- that represents a motor performance novel to
ied in the laboratory, away from these sources of the subject, and the experimenter studies how
variability. To be sure, there is a trade-off in this the subject attempts to learn it or what variables
approach: Laboratory studies tend to make the influence that learning. The particular constructs
situation less natural and more artificial, and the being measured (e.g., balance, timing, movement
measures taken are not quite as directly related to speed) frequently are not very important to the
26 Motor Control and Learning

experimenter, because the primary focus is on act at a particular time (e.g., hitting a baseball). A
the variables that affect performance and learn- particular force, distance, speed, or time can be
ing generally. defined as the subject's target; then, deviations
of the subject's performances with respect to this
target are measured. The level of analysis that is
Measuring Motor Behavior least sensitive comprises dichotic outcomes, such
as when the performances are scored as hit/miss
In the field of motor behavior and control, mea- or right/wrong, as in shooting a basketball or
surement can be approached in essentially three judging which of two lifted weights is heavier.
different ways. At the most general level, we can We can refine the accuracy score by dividing the
describe how well a movement achieved some possible outcomes into hit/almost hit/miss or
environmental goal that was inherent in the task by dividing a bull's-eye into 10 or more zones,
(e.g., whether or not a target was struck). Here for example. But motor performance is complex,
the emphasis is on the outcome of movement. and more sophistication in the measurement of
At a more specific level, we may be concerned accuracy is usually required.
with quantifying the actual movements the In the discussion that follows, assume that a
person made. In this case, the focus of analysis is single performer is striving for accuracy in arriv-
describing the movement itself. The third level of ing at some target (e.g., a force, a speed, a location
analysis entails the study of the brain and central in space) and that the movement outcomes can be
nervous system prior to and during the produc- placed along some measurable dimension (e.g.,
tion of movement. At this level, researchers are kilograms, centimeters per second, centimeters)
interested in the neural activities involved in as in figure 2.3. Let the correct value along this
planning and executing movements. dimension-the target-have the value T. The
values that the performer actually achieves are
Describing the Outcome abbreviated by xi' where i is a subscript notating
of Movements a particular trial (i.e., the ith trial). For example,
The first aspect of measurement in motor behav- x23 is the score on the 23rd trial. In the simple for-
ior is quantification of the extent to which a given mulas that describe these fundamental statistical
movement achieved the goal that was intended accuracy scores, the symbol~ means "the sum
or instructed. For example, did the movement of." For example, ~xi means to add up all of the
result in striking the target, or was the movement values xi' where i ranges progressively from 1
made at the right time? Such measures generally through n, with n = 5 in the following example:
concern the movement in relation to some object (2.1)
or to another performer in the environment,
although some movements (e.g., modern dance, In the following explanation, assume that the
diving) may not be so closely associated with target (T) is 100 units ~nd that the individual does
other environmental elements. The achievement not always achieve this target score. In figure 2.3
of such environmental goals can be assessed in there are five scores: 93, 103, 99, 105, and 96 units
essentially four fundamental ways-through for trials 1through5, respectively.
measures of (a) error, (b) time and speed, (c) It is obvious that no single trial will be very
movement magnitude, and (d) performance on effective in describing the subject's behavior, as
secondary tasks. the scores possess a great deal of variability. One
solution is to combine these scores to achieve a
Measures of Error for a Single Subject more representative measure of the subject's capa-
Many performances require the subject to per- bility. In the study of motor behavior, researchers
form some action with maximum accuracy. Thus, have typically focused on five methods for com-
the performance measures represent the degree to bining scores into measures of "error," and each
which the target was not achieved-a measure of one has a slightly different meaning in terms of
error. The accuracy goal can be imposed in many the performer's capability. These methods are
ways; for example, subjects can be asked to move described in the next sections as (a) constant error,
with a certain amount of force, hit a certain spatial (b) variable error, (c) total variability, (d) absolute
target, move at a certain speed, or perform some error, and (e) absolute constant error.
 Methodology for Studying Motor Performance 27

FIGURE 2.3 An arbitrary measurement scale, showing locations of a target (T) and of five hypothetical movement
attempts (x1 , . • . , X5 ).

Constant Error (CE)-Computation The first ference scores are shown in column C, headed
statistic to be considered as a measure of the (xi - T). It is important to notice in this column
subject's accuracy is the constant error (CE), which that for CE, the sign (+ or - ) of the difference
measures the average error in responding. Its is retained. Next, the summation sign calls for
formula is adding the values (using the signed values) for
Constant error= CE= 2. (xi - T) In (2.2) each of the trials (in this case for each of the five
trials), and this sum is presented at the bottom
where xi is the score on trial i, Tis the target, and n of column C (-4.0). Then the formula calls for
is the number of trials the subject performed. It is dividing by n, the number of trials, in order to
very easy to compute this measure from table 2.1, get the average CE over trials. The final CE score
which can serve as a work table for computing all is -4.0/5, or -0.80.
of the statistics presented in this section on error
measures. The trial numbers are listed in column Interpretation of Constant Error The CE score
A; the scores obtained (x) are given in column B. of -0.80 indicates that, on average, the subject fell
All other values in the table are computed from slightly short of the target (by 0.80 units). Notice
these initial values (remember that T = 100 in that the CE is given in units that represent the
this example). amount and direction of deviation relative to the
To compute the CE, the numerator calls for target, sometimes called bias. One could also ask
finding the difference between each of the scores for the subject's scores on the average by consult-
on the test (x) and the target (T = 100); these dif- ing the mean for column B. Thus, the average
28 Motor Control and Learning

score was 496 I 5 = 99.2 units, meaning that the are given in column E. Next, obey the summa-
subject fell short of the target by 99.2 -100 units, tion sign and add the squared values, the sum of
which is also -0.80 units. The CE represents the which (96.80) is shown at the bottom of column
average magnitude of the movement and mea- E. Then divide this sum by the number of cases
sures the direction of the errors on the average. (n = 5) to get 19.36 and take the square root to
While a measure of average error bias might, arrive at the final answer of 4.40 units.
at first, seem satisfying to students as a measure
Interpretation of Variable Error The VE reflects
of accuracy, notice that the value computed for
the variability, or inconsistency, in movements,
the subject (-0.80) was far smaller than the error
as can be seen from the "ingredients" in the
for any of the single movements that contributed
formula. The important feature is the difference
to the average. The movements were scattered a
between the subject's score on each trial and his
great deal, with the center of movements being
own average score. Thus, if one subject always
roughly the target that was the goal. What the
moves very consistently, the VE will tend to be
CE does not consider is this amount of scatter,
small. If the subject always receives the same
variability, or inconsistency in performance of
score, even though it is not the correct one (such
the movements. Consider a second hypothetical
as a score of 99 on all five trials), then the VE will
subject with scores of 99, 99, 99, 99, and 100. These
be zero. This is so because the subject's average
scores represent a very small scatter but would
result in precisely the same CE score as for the score will be 99, and the difference between each
subject we have just been considering (-0.80). of the scores and the average will always be zero
For this reason, another measure of error, the as well.
variable error, is used to describe the subject's Thus, VE does not depend on whether or not
inconsistency. the subject was close to the target, since it is the
measure of spread about the subject's own aver-
Variable Error (VE)-Computation The vari- age. To illustrate, the VE for the set of scores 43,
able error (VE) measures the inconsistency in 53, 49, 55, and 46 achieved during aiming at a
movement outcome. It is the variability of the target of 100 units will be precisely the same (4.40)
subject's performances about the mean value and as that calculated in the previous example. (We
is calculated by the formula obtained these five new values by subtracting 50
Variable error= VE=~~ (x; - M)2 I n (2.3) from each of the raw scores in table 2.1.)
where x; and n are defined as in the previous Using Constant Error and Variable Error An
example. The Mis the subject's average move- additional aspect of error scores is important
ment, measured in the same units as the scores from the point of view not only of research but
for the task, so that for this example the M has also of practical application. Compare two rifle
the value of 99.2 units. To compute the VE for this marksmen: Marksman Ahas a large VE and small
subject, use table 2.1 once again. Notice that the CE, whereas marksman B has a small VE and
formula indicates first to compute the difference large CE. This situation was described years ago
between the performance score and the subject's by Chapanis (1951) and is illustrated in figure
own mean (M), so the first step is to compute the 2.4 (see "The Relative Importance of Constant
subject's M. As noted in the previous section, the and Variable Errors"). Which marksman, A or B,
computed M for these trials was 99.2 units. Now, appears to be the more skilled? 1
the values in column D of table 2.1 represent the The study of motor learning will show that
differences between each of the scores on the the measure of error that is most sensitive to the
trials and 99.2 (that is, the difference between each effects of practice is consistency (VE); bias (CE)
individual score and the mean of all that person's often changes quickly in the first several trials and
scores). For example, 93.0 - 99.2 equals -6.2, the remains near zero thereafter, even after years of
first entry in column D. Since the deviations from practice. There are some situations, however, in
the mean for each of the individual trials must which CE is preferred to VE; but these are special-
cancel each other out, by definition the sum of the ized applications. Thus, these two measures of
values in column D must equal zero. So, the next error, CE and VE, seem to represent two distinct
instruction from the formula is to square each of aspects of performance-bias and variability,
the values in column D, and these squared values respectively. But sometimes it is more desirable
 Methodology for Studying Motor Performance 29

THE RELATIVE IMPORTANCE

OF CONSTANT AND VARIABLE ERRORS

Alphonse Chapanis was one of the pioneers in the emergence of human factors research. His
analysis of errors in movement is as important today as it was over a half century ago.
"Having defined constant and variable errors, we might ask: Which is the more important? Let
us return for a moment to the target patterns shot by the two riflemen [figure 2.4]. At first glance,
you might say that B is a very inaccurate shooter. And yet any rifleman will tell you that this is
not the case at all. B is a much better shooter than A The reason is this: The large constant
error in the trial shots fired by B can be compensated for very easily by simple adjustments in
his sights. With suitable corrections in elevation and windage, rifleman B will turn in a perfect
score. In rifle shooting, then, constant errors are not the important ones, because they can be
very easily adjusted for by changing the position of the sights on the gun. The really important
errors are the variable errors. No correction of the sights on P.s gun will make all of his shots
fall in the center. He is inherently much too variable" (Chapanis, 1951, p. 1187).

Marksman A Marksman B

FIGURE 2.4 Distribution of rifle shots. Marksman A has a small constant error (CE) and large vari-
able error (VE). Marksman B has a large CE bias, but a small VE.
Reprinted, by permission, from A. Chapanis, 1951, "Theory and methods for analyzing errors in man-machine systems," Annals of the New York
Academy of Sciences 51: 1181.

to have a single measure of "overall error" that E2 = VE2 + CE 2 (2.4)

combines both of these performance indicators E can also be computed directly from the
rather than using separate measures of accuracy formula:
and variability.
Total variability= E = ~ ~ (x; - T)2 I n (2.5)
Total Variability (E)-Computation The total
variability around a target (or error) for a set where X;, T, and n are defined as before. To apply
of movements (labeled E by Henry [1975] and the formula, we can use table 2.1 again. Notice
sometimes called root mean square error) can be that the major "ingredient" is the difference
thought of as the measure of "overall error" (see between the score and the target (x; - T), and
also Chapanis, 1951). E can be defined as the this difference (with the sign included) is given
square root of the sum of VE 2 and CE2, or in the in table 2.1, column C-the same values used to
following formula when expressed as E2: compute CE. Next, square each of these values;
30 Motor Control and Learning

these squared values are given in column F. The Interpretation of Absolute Error In providing
summation sign then says to add the squared an interpretation of the AE, it will be helpful to
values, and the sum (equal to 100) is given at the consider the ways in which it is similar to E. First,
bottom of column F. Next, divide by n (5), which notice that the numerator is essentially the same
results in a value of 20; then find the square root. for the two statistics, each numerator having a
The final value for E is fiQ, or 4.47. difference between the obtained score (x) and the
Interpretation of Total Variability The total target (T) as the major "ingredient." Second, the
variability, E, is the total amount of "spread" of xi values for E and AE (4.47 and 4.0, respectively)
the movements about the target, so it represents are very similar; the two values will be equal only
an overall measure of how successful the subject in special circumstances but will be very close
was in achieving the target. The key to under- in most situations. Third, both of the formulas
standing this formula is the expression in the involve methods for eliminating the sign of the
numerator (xi - T)2. Eis based on the sum of a difference between the score and the target; for
group of squared differences, where each differ- the AE, the method is to take the absolute value,
ence is the amount by which the subject missed while for E, the method is to square the values
the target. This contrasts with VE, where the in the numerator.
numerator (xi - M)2 represented the deviations Absolute Error Versus Total Variability The
from the subject's own average, which is not neces- AE is a very "logical" measure to use to describe
sarily equal to the target. In cases when CE is close the subject's overall accuracy in a task because it
to zero (i.e., the mean of the trials approximates is sensitive to the extent to which the subject was
the aimed-for target [Mand Tare nearly equal]), "off target." It was used far more commonly than
then E and VE come to represent very similar E in the early research and for many different
aspects of the subject's performance. But, in cases applications. A controversy, however, has arisen
when the CE is very different than the target, then about the use of AE (Schutz & Roy, 1973). The
E will represent the combination of this error in mathematical properties of AE have been shown
bias plus the variability about the CE (i.e., E is a to be a complex combination of CE (accuracy or
combination of CE and VE). bias) and VE (variability), and it is difficult to
Absolute Error (AE)-Computation A statistic be certain of the relative contribution of each.
closely related to the total variability (E) is abso- Because of the precise relation among E, CE, and
lute error (AE), which can also be thought of as VE (namely, E2 = VE 2 + CE2), Eis always an exact
a measure of overall accuracy in performance. It combination of the variability and bias, and thus
is the average absolute deviation (without regard is preferred to AE (Henry, 1975). The tendency
to direction, or sign) between the subject's move- today, when a researcher wishes to present a
ments and the target, and its formula is as follows: combined measure of accuracy and variability,
is to prefer E, for two reasons: First, E measures
Absolute error= AE = 2, IX; - TI In (2.6) essentially the same -component of movement
where X;, T, and n are defined as before. The as AE, and second, E is more easily interpreted
important difference here is the presence of the since it represents a simple combination of CE
vertical bars (I I), which are the symbol for and VE. However, we will use AE a great deal
"absolute value of" and mean that we should take in this text, because much of the earlier research
away the sign of the difference before summing. reported only this measure.
To compute AE, refer again to table 2.1. The Absolute Constant Error (I CE I) One final
first step is to compute the values for the numera- measure of accuracy is merely a transformation
tor terms, and this is done in column G, headed of constant error, CE:
IX; - TI · Notice that the value in this column is
Absolute constant error = ICE I (2.7)
the same as the corresponding value in column
C, except for the sign. The summation sign ~ is Thus, for a single subject, the absolute constant
an instruction to add up the values from each of error ( ICE I) is just the absolute value of the CE,
then trials (recall that n = 5 in this example), and which is simply 0.80 in the present example. Be
the sum is given at the bottom of column Gas 20. careful to note, however, that ICE I is not cal-
The next step is to divide by the number of trials culated in the same way as AE. For AE we take
included (n = 5), and so the final answer is 4.0. the sign away immediately after calculating the
 Methodology for Studying Motor Performance 31

difference between a single score and the target, direction, the measures of overall error (E and
prior to summation. For ICE I the sign is not taken AE) tend to represent the magnitude of the bias,
away until after the average over a series of trials and VE alone represents the variability.
has been calculated. Now consider subject 2 in table 2.2. This subject
Interpretation of Absolute Constant Error The has the same spread of outcomes as subject 1, but
situation sometimes arises (e.g., Newell, 1976a) in with much less bias. We obtained these scores
which approximately half the subjects in a group by adding 15 to each of subject l's scores to get
have positive CE scores while the other half have subject 2's scores: 95, 105, 100, 97, 102. Table 2.2
negative CE scores. If one characterizes the aver- gives the error measures of this set of scores. Now
age bias for this group of subjects by calculating notice that the measures of overall error tend to
a group mean score (the average score of all the be very close to the VE, all around 3. The CE (and
subjects), the positive and negative signs will ICE I), however, is now nearly zero. Here the
tend to cancel" each other out. This could give
/1 measures of overall error (E and AE) represent
rise to the misleading conclusion that the average the variability of the movements (VE), exactly the
bias for the group was nearly zero when in fact opposite of the situation with subject 1.
every subject in the group might have shown a Therefore, when CE is large in either direction,
bias of considerable size. In this case, it is useful the measures of overall error (E and AE) tend
to also compute ICE I, which tells the researcher to represent the amount of bias in the scores.
the mean amount of bias for the group of sub- When CE is small, E and AE tend to represent the
jects without regard to its direction, and which amount of variability (VE) in the scores. When
will not fall prey to this canceling" effect of the
/1 CE is intermediate in value (with some bias, but
positive and negative CEs. In cases in which with scores falling on both sides of the target), the
group data have been the focus of a researcher's measures of overall error represent an unknown
experiment, there has been a trend to report ICE I combination of bias and variability. This should
(as the measure of accuracy) along with VE (as make clear why simply examining overall error
the variability score) (see Schutz, 1977, for more statistics does not provide a very complete picture
statistical details). of performance.
Relationships Among the Error Scores One Other Measures of Accuracy There are many
way to evaluate the relative contributions of the tasks in the motor behavior literature that could
various measures of error is to consider the fol- not be scored so simply. A task for which accu-
lowing cases. At one extreme, when CE is very racy is important is the tracking task; in this case,
large (an extreme is the situation in which all the performance is ongoing, thus preventing the
person's movements lie on one side of the target), computation of a discrete performance error. A
then the absolute error (AE), the total variability commonly-used tracking task is the pursuit rotor,
(E), and the constant error (CE) all tend to mea- shown in figure 2.5a. There are many varieties of
sure the same component of performance-the pursuit tracking tasks, and figure 2.5a illustrates
bias or directional deviations of the errors. In the just one type. Here, a target (e.g., a small circle)
following case, the target is again 100, but subject is embedded in the surface of a turntable-like
1 produces five movements, 80, 90, 85, 82, and structure that rotates at various speeds. The
87-each smaller than the target. Table 2.2 gives subject holds a stylus in the preferred hand and
the measures of error for this subject. Notice that attempts to keep its tip in contact with the target
the statistics E, CE, AE, and ICE I are all around as the turntable rotates. A trial might last from
15, but that the VE is very much lower at 3.54. 10 s to 1 min, and performance is scored in terms
This suggests that when the CE is large in either of the amount of time in the trial that the subject

Subject 1
Subject 2 3.55 -0.2 3.54 0.2
32 Motor Control and Learning

maintained contact with the target. The perfor- scored in terms of the number of correct matches
mance measure is usually called time on target that can be achieved in a trial of fixed duration.
(TOT) and can range from zero (if the subject The stabilometer is shown in figure 2.Sc. The
never touched the target) up to a value equal to standing subject attempts to keep an unstable
the duration of a trial (if the subject was always platform level; the scores denote either time in
in contact with the target). But notice that time balance or the number of times the platform edge
on target can represent a complex combination touches the floor (indicating extreme loss of bal-
of bias (if the subject is consistently behind the ance) during a trial of perhaps 30 s. The number
target, for example) and variability (if the subject of times a stylus touches the sides of a maze is
is alternately ahead of and behind the target). another example of this kind of measure. Figure
Other common variations of tracking tasks are 2.Sd shows the two-hand coordination task, in which
shown in figure 2.5. The Mashburn task is shown the subject attempts to follow a target by moving
in figure 2.Sb. It was designed to simulate certain a pointer with two crank handles. One handle
features of airplane controls. The control panel controls the right-left movement, and the other
contains three double rows of lights. One row of controls the forward-backward movement (as in
each pair is controlled by the movements of the the "Etch-a-Sketch" toy). The score is again TOT,
subject (left-right and forward-backward move- or the amount of time in a trial that the subject was
ments of the stick, and right-left movements of the over the target. All these measures, including TOT,
pedals), while the other row of each pair is con- are measures of overall error, and they tend to
trolled by the experimenter. The subject attempts confound the bias with variability in performance.
to match the experimenter-determined lights with In each of the foregoing examples, the experi-
appropriate movements of the controls. The task is menter does not need to keep a record of the sub-

Mashburn task

a Pursuit rotor b

coordination task
c Stabilometer d

FIGURE 2.5 Four movement tasks frequently used in motor behavior research: (a) pursuit rotor, (b) Mashburn task, (c)
stabilometer, and (d) two-hand coordination task.
 Methodology for Studying Motor Performance 33

ject's actual behavior. However, there are other can accomplish more in a given amount of time,
tracking tasks for which a continuous record of or who can produce a given amount of behavior
the movements of the subject and the target is in less time, is the more skillful. These two kinds
kept. From these data a measure of overall accu- of measures are essentially the same, since a time
racy, the root mean square (RMS) error, can be measure (time/unit) can easily be converted to a
computed. (Notice that RMS error is analogous speed measure by taking the reciprocal; that is, 1
to E, described earlier. Both are root mean square I (time I unit)= units I time, which is a measure
deviations of the behavior from some target, com- of speed. Both speed and time measures have
puted on successive "trials.") Essentially, the RMS been used a great deal in motor behavior research.
error is based on taking small "slices" of time Reaction time and movement time are common
and measuring the deviation of the subject's line examples, described next.
from the target at each of these times, as shown
Reaction Time Reaction time (commonly abbre-
in figure 2.6. Depending on the capacity of the
viated as RT) is a measure of the time from the
recording system, these slices can be taken every
arrival of a suddenly presented and unanticipated
few milliseconds or so over the entire course of
signal to the beginning of the response to it. In
a 20 s trial, providing many measures of error.
the RT paradigm shown in figure 2.7, the subject
To compute the RMS error, square each of these
is given a warning signal, and after a randomly
deviations from the track, add up the squared
deviations, divide by the number of measures, determined foreperiod (perhaps ranging from 1 to
and then take the square root, giving a measure 5 s), the stimulus is presented. Using the variable
of the amount of deviation over the course of foreperiod represents an attempt to prevent the
the trial. Root mean square error in each of these subject from anticipating when the stimulus will
cases represents essentially (but not exactly) the arrive (temporal anticipation). Sometimes" catch
area between the subject's movements and the trials" are used, in which the stimulus is not
target, as shown by the shaded portions of figure presented at all; this allows the experimenter to
2.6. As with TOT, the RMS error is a measure of "catch" a subject who is anticipating, and thus its
overall error and is sensitive to both the bias and use tends to prevent anticipation. The introduc-
the variability in performing. tion of catch trials, given randomly perhaps on
15% of the total number of trials in an experiment,
Measures of Time and Speed improves the experimental control. Also, subjects
The second fundamental way of assessing skills can be prevented from anticipating which move-
is by measures of time and speed. Basic to this ment to make (i.e., spatial or event anticipation)
idea is the assumption that the performer who through the use of two or more choices, so that the

c:
0
:;::;
'iii
0
c..

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Time (s)

FIGURE 2.6 Hypothetical record from a tracking task, showing the basis for computation of root mean square (RMS)
error.
34 Motor Control and Learning

Stimulus Response Response

Warning presented begins ends

Response time

1 - - - - - - RT _ _ _ ____,,_,_..,,___ _ __ MT --------c..-

_.___ Premotor ---)!- -<Oii(- Motor ___..._

RT RT

EMG

FIGURE 2.7 Critical events involved in the reaction-time (RT) paradigm.

proper response is signaled by the stimulus itself the signal to the first change in EMG is termed
(e.g., red light means move left, blue light means "premotor RT" and is thought to represent central
move right); this is termed the choice-RT method processes involved in making the response (e.g.,
(when only one option or "choice" is available, perception, decisions). The interval from the first
this is termed a simple-RT method). change in EMG to finger movement is termed
Reaction-time measures are common in many "motor RT" and represents processes associated
sport settings; an example is the interval between with the musculature itself. Such methods are
the starter's gun and the first movement in a useful in gaining further information about the
swimming race. This is an example of a simple-RT effect of an independent variable on RT (e.g.,
task in which there is only one response to make. Fischman, 1984).
The starter varies the time between the "ready" Reaction-time measures are very common in
command and the auditory signal to start in order research on skills, for two basic reasons. First,
to reduce temporal anticipation. The decision of RT measures are components of real-life tasks
a soccer goalie to dive to the left or to the right (e.g., sprint starts), so they often have high face
in a penalty kick is an example of a choice-RT validity. Amore imp0rtant reason (which we will
task in sport. Reaction-time measures are also amplify in chapters 3 and 4) is that RT presumably
used extensively in the laboratory as measures measures the time taken for mental events, such
of information-processing speed (see chapter 3). as stimulus processing, decision making, and
One variation of the RT method is to partition movement programming. These two motivations
the latency of the response initiation into "cen- for using RT measures differ considerably. In the
tral" and "peripheral" components (Weiss, 1965). first case, RT is a measure studied for its own
The bottom of figure 2.7 shows a hypothetical sake; in the second case, RT allows the researcher
electromyographic (EMG) trace taken from a to understand the kinds of mental processes that
muscle involved in the movement to be made lead to movement (e.g., Posner, 1978). Regardless
(EMG indicates the electrical activity in a muscle). of the motivation, the measurement of RT is the
The EMG is mostly silent during a substantial same.
part of the RT, indicating that the command to
move the finger, which is initiated in the brain, Movement Time Movement time (commonly
has not yet reached the finger musculature. The abbreviated as MT) is usually defined as the
muscle is activated late in the RT, but no move- interval from the initiation of the response (which
ment occurs for 40 to 80 ms. The interval from defines the end of the RT) to the completion of
 Methodology for Studying Motor Performance 35

the movement (figure 2.7). Clearly, MT can be just via instructions, so that accuracy can be assessed
about any value, ranging from a few milliseconds (e.g., Quinn, Schmidt, Zelaznik, Hawkins, &
for a very quick movement to several weeks if McFarquhar, 1980; Schmidt, Zelaznik, Hawkins,
the movement being studied is jogging from Frank, & Quinn, 1979).
Los Angeles to Chicago. Some sport skills have
minimal MT as a goal (e.g., time to run 100 m, Measures of Movement Magnitude
or time for a quarterback to "set up" for a pass), A third way of measuring skills is by the mag-
and MT is used a great deal in skills research as nitude of behavior that the performer produces,
a result of its overall external validity in these such as the distance that a discus was thrown
practical settings. Sometimes researchers use RT or the amount of weight that was lifted. These
and MT tasks together in the same performance, measures have particularly important applica-
as in requiring the subject to lift a finger from a tions to sport settings, as many sports use such
key and move to a button as quickly as possible measures as the primary determinants of success
after a stimulus. in the activity. One example of these measures in
The sum of RT and MT is termed response time research is the Bachman (1961) ladder climb task
(figure 2.7). Research has consistently shown that (figure 2.8), which uses a specially constructed
very different processes or abilities are required ladder. At the beginning of, say, a 30 s trial,
in reacting quickly as opposed to moving quickly the subject begins to climb the (unsupported)
once the reaction is over, and this has justified ladder without skipping rungs until balance is
separating response time into RT and MT. What lost and the subject topples over. The subject
is frequently called "brake reaction time" in the quickly returns to the starting position and begins
field of automobile accident analysis is really climbing again, and so on until the trial has been
response time, because it consists of the time used completed. The score is the number of rungs
to initiate the foot movement from the accelerator accumulated in a given trial. A variant of the task
pedal plus the time required to move the foot to is to climb as high as possible in a single attempt.
the brake pedal and press it. The ski simulator (figure 2.9) is a task in which
Often the degree of accuracy in the task must large rubber bands keep a platform centered. The
be taken into account when measures of speed subject's task is to displace the platform as far
are used. A well-known phenomenon in motor as possible from side to side using whole-body
behavior is the speed-accuracy trade-off, mean- movements.
ing simply that when performers attempt to do
something more quickly, they typically do it less
accurately. In most measures of speed, therefore,
accuracy requirements are kept to a minimum
so that speeding up the movement (which is the
major goal for the subject) does not seriously
affect accuracy. In some situations, though, mea-
sures of speed are confounded with measures of
accuracy, and the speed with which the subject
performs is dependent on the amount of error
she is willing to make or the amount of error
the experimenter will tolerate. Such trade-offs
are particularly troublesome for experimenters,
because it is not always clear to subjects how
much error will be tolerated, and experimenters
are unsure about how to interpret an indepen-
dent variable that produces increases in speed
but decreases in accuracy. One solution to this
problem is to hold accuracy constant by various
experimental techniques so that a single depen-
dent variable of speed can be assessed (e.g., Fitts,
1954). Another solution is to hold speed constant, FIGURE 2.8 The Bachman ladder climb task.
36 Motor Control and Learning

FIGURE 2.9 Ski simulator being used by Gaby Wulf.

Reprinted, by permission, from G. Wulf, 200~ Attention and motor ski/I learning (Champaign, IL: Human Kinetics), 9.

While it might seem that tasks with goals of measurement will be sensitive to differences in
maximum movement magnitude are consider- skill among individuals or to differences in skill
ably different from those requiring speed or caused by some independent variable. Generally
accuracy, their fundamental determinants may these situations involve tasks for which differ-
not be all that different. At first glance, producing ences in performance are not evident because
maximum movement magnitude would seem the tasks are well learned (driving a car down
simply to be a matter of generating more force. an open road), or involve tasks that do not "tax"
But these skills certainly require precise timing the motor system very much because they are so
of the forceful contractions and accurate coordi- simple (drinking a glass of water without spill-
nation among the various participating limbs. ing). How are skills assessed in such situations?
Therefore these precise muscular activities might One method is to use some measure of critical
be essentially the same as those required in tasks incidents. In the driving example, accident rates
that seem to necessitate only accuracy. Of course, from statistical databases for a particular kind
inconsistency (in terms of VE) in these processes of car, or for certain types of people (e.g., drunk
will degrade performance, and such inconsis- drivers), might be used as measures of skill; with
tency is probably related to the VEs that are seen pilots, "near misses" (midair near-collisions)
in the outcomes of simpler tasks. might be used. But these techniques are difficult
to utilize in the laboratory because (fortunately)
Measures of Secondary Tasks such critical events occur so infrequently. Thus,
There are instances in both practical and research they are far more useful for groups of people
settings in which none of these basic methods of taken over relatively long stretches of time.
 Methodology for Studying Motor Performance 37

Another useful technique is to employ perfor- (as a measure of overall effort), or even EMC
mance on some sort of secondary task, performed from the muscles of the forehead (to indicate the
simultaneously with the primary task, as a mea- level of concentration) can be used, depending
sure of the skill in the primary task. For example, on the particular situation. Neurophysiological
Brown (1962) used a verbal task in which the indi- techniques, such as the recording of event-related
vidual was presented with eight-digit numbers at potentials from the scalp, are also used. One such
4 s intervals. Each number contained seven of the potential has been called P300, as it occurs as a
same digits as the previous one, and the subject's positive voltage about 300 ms after the presenta-
task was to detect the digit that was different and tion of a stimulus (Duncan-Johnson & Donchin,
to provide a response. Errors were counted as 1982). The latency of the P300 and the magnitude
omitted responses, incorrect responses, and late of its amplitude appear to be highly correlated
responses. In this experiment, Brown obtained with RT, thus providing an unobtrusive way to
measures of performance when subjects per- measure processing activities (e.g., Radlo, Janelle,
formed only this verbal-numerical task, and also Barba, & Frehlich, 2001). In all these cases, the
when subjects performed this task while driving secondary measures become the focus of the
under various conditions. The mean percentage investigator, especially when the main task does
of correct responses when the verbal task was not provide sensitive measures of the subject's
performed alone was 90.6%. When the task was performance.
performed during driving in quiet residential A variation of this technique is to use a sec-
areas, the mean percentage dropped to 83.8%. ondary task as a distracter in order to increase
And when the task was performed in heavy traf- the overall "load" on the performer. Normally,
fic conditions, the percentage again dropped, to fatigue may not have any obvious effect on the
79.5%. Yet it was very difficult to see any differ- well-learned task of driving. However, if the
ences in vehicle control in light and heavy traffic, driver is required to perform a simultaneous
largely because driving is so well learned by most mental arithmetic task at a predetermined level,
people. This secondary task provided evidence then large differences between fatigued and
about the difficulty of the driving conditions rested driving may be seen. In this situation,
when the driving task itself would not have pro- unlike the others, the major interest is in the per-
vided such a measure. formance of the main task, and the secondary task
Some experimenters have used RT tasks has increased the sensitivity of the measurement
(so-called probe tasks-e.g., Kerr, 1975) inserted system for the main task. However, care must be
during a performance of the primary task. Others taken with these techniques, as Brown (1962) has
have used measures of finger tapping regularity shown. When truck drivers were fatigued, their
(e.g., Michon, 1966) as the secondary task. In performance on the secondary digit detection
these latter cases, the implication is that these task actually improved, suggesting that they were
tasks require some of the subject's limited capacity devoting less capacity to the driving task and
to process information; presenting the second-
overcompensating by devoting more capacity to
ary task simultaneously with the primary task
the secondary task. While these techniques can
necessitates the use of some of this capacity, and
be somewhat tricky to use, they have served well
lowers the performance on the secondary task in
in a number of situations.
relation to the amount of capacity demanded by
the primary task. (See chapter 4 for more on this Describing Characteristics
assumption and general method.)
Rather than a task, secondary physiological mea- of Movements
sures of effort can be used during the performance Countless methods could be employed to
of a main task. One technique is to measure pupil describe movements, depending on the charac-
diameter by one of various recording methods. teristics of the movement that are of interest to the
Pupil dilation is associated with circumstances observer. At the most fundamental level, one can
in which effort, arousal, or information process- use verbal descriptors to characterize movement.
ing is demanded (e.g., Beatty & Wagoner, 1978; For example, movements have been described in
Kahneman, 1973). Similarly, measures of heart dance notation created by Laban (1956) and in
rate, heart rate variability, oxygen consumption terms of units of work behavior called "Therbligs"
38 Motor Control and Learning

in early industrial time-and-motion studies (Gil- 1979) using series of still photos (see "The Move-
breth, 1909; note that "Therblig" is the author's ment Photographs of Eadweard Muybridge,"
name spelled backward, almost). Another way is figure 2.10), or with videotape or digital images.
to illustrate movement with photographs, as was Such methods are of some use in describing or
done over a century ago by Muybridge (1887, illustrating the basic forms of movement, but

The Movement Photographs

of Eadweard Muybridge
In 1887 Eadweard Muybridge published what probably was the first photographic analysis of human
and animal locomotion. In some series, Muybridge used 36 separate shutters, arranged so that
three different angles could be photographed simultaneously, positioned at the side, front, and rear
of the subject. At each angle a series of 12 shutters was electrically arranged to open in a timed
sequence after a signal had been sent to the first shutter. Muybridge determined the length of the
time intervals on the basis of the speed at which the subject performed the activity (e.g., picking up
an object vs. running at full speed). These methods resulted in a sequence of still photographs that
produced simulated motion, which one could study further by examining the changes in the action
as it evolved over time. In all, Muybridge published 781 plates, each plate a series of photographs of
the subject. The subjects were both. humans and animals. Humans of all ages, males and females,
and of many different body types, performed actions such as walking, running, jumping, skipping,
lifting objects, and hammering. Many of the humans were photographed in the nude, which provided
rather explicit details of body actions. Animals from a wide range of species were photographed in
various natural and contrived settings. It is rumored that Muybridge settled a bet concerning whether
or not all of a galloping horse's feet are ever off the ground at the same time. According to his photos,
they are. Muybridge's series of still-sequence photographs remains as a magnificent legacy in the
history of human and animal motion analysis.

FIGURE 2.10 A famous series of Muybridge photos, clearly revealing a period of time when the horse has no
contact with the ground.
 Methodology for Studying Motor Performance 39

have limited value in detailed assessments of or the entire body. The locations of various parts
performance. Improvements in photographic of the body during the movement, the angles of
techniques occurred early in the 20th century; the various joints, and the time relations between
some were pioneered by Bernstein (Bernstein & the movement in one joint and the movement in
Popova, 1930/2003). In this section we focus on another are examples of the many ways move-
kinematics, a branch of mechanics in physics that ment kinematics can be recorded.
involves the description of "pure" motion without
regard for the forces and masses that produced the Location Perhaps the most common of the kine-
motion. The devices that can be used to collect this matic methods entails recording the locations of
information also vary widely, and we will describe the limbs during a movement. An early example
some of the more common ones. of research using movement kinematics is that
of Lindahl (1945; see "Lindahl's Study Using
Movement Kinematics Movement Kinematics"). Early in the history
As applied to movement behavior, kinematic of motor behavior and biomechanics, research-
measures describe the movement of the limbs ers used high-speed cinematography to record

Lindahl's Study Using Movement Kinematics

One of the early investigations using movement kinematics as a research tool was published by
Lawrence Lindahl in 1945. His study concerned a very practical problem: an industrial task in which
a factory worker operated a machine that cut very thin slices off tungsten rods. The key aspect of
the job involved using a foot pedal action that moved an unseen cutting wheel through the tungsten
rod. Success at the task required carefully coordinated displacement-time foot actions. Failure
to coordinate these actions resulted in wasted material (improperly cut discs) and damage to the
equipment (broken wheels). Skilled machine operators made very stereotypical actions of the
foot, which Lindahl captured nicely using a writing apparatus that recorded the foot displacements
over time (see figure 2.11 ). These recordings were then used for comparison with the recordings
that beginning and less skilled workers produced in their disc-cutting performances. This type of
recording, using kinematics as a means to provide objective feedback and goal-related movement
information (so-called kinematic feedback), predated modern investigations by about 40 years (see
discussion in chapter 12).

Machine table

Paper tape

FIGURE 2.11 Experimental setup used by Lindahl (1945) to record kinematics of the disc-cutting task.
40 Motor Control and Learning

movements. Often the subject being filmed wore from the output of a device called a potentio-
tape markers over certain landmarks (e.g., the meter that signals angular position. This trace,
wrist or ankle) so that the locations of these body read from left to right, represents an arm move-
parts could be studied frame by frame. These ment of about 17 cm. The movement began at the
positions on successive frames were separated time when the trace left the horizontal axis. The
by nearly fixed periods of time, so a graph of the largest amplitude (about 20 cm) was achieved
position of the landmark against time could be about 125 ms after the movement started, and then
generated from the data. the limb stabilized its position at the final location.
Figure 2.12 shows an example of this type of But examining the location of a limb in space
graph, taken from Wadman, Denier van der Gon, may mask some of the more subtle factors that
Geuze, and Mol (1979). For now, consider only determine its control. For this reason motor
trace a, which represents position; it is derived behavior researchers often examine variables that

0.3

:[ 0.2
c:
~
"(ij
0 0.1
D..

a O
3.0

-!!! 2.0
.§_
~
"ij
0 1.0
Qi
>
0

+50.0

N'
i c:
0
~.
Q)
Qi
0

0
0
<(

-50.0

0.0 0.1 0.2 0.3 0.4

c Time (s)

FIGURE 2.12 Position, velocity, and acceleration traces representing a rapid 17 cm elbow extension movement.
Reprinted from W.J. Wadman et al.. 1979, "Control of fast goal-directed arm movements;· Journal of Human Movement Studies 5: 5. By permission of W.J. Wadman.
 Methodology for Studying Motor Performance Ill! 41

can be derived from location information-veloc- actions of one body part are controlled together
ity and acceleration. with the movements of another body part. Many
Velocity Trace b in figure 2.12 is a record of types of coordination exist, such as the coordina-
velocity of the movement at each moment in time, tion of two or more joints in one limb (as when
placed on the same time scale as the positional one is reaching for a cup) or of different limbs
trace for easy comparison. The velocity trace was simultaneously (as of the arms and legs during
determined by computer, which read in the posi- walking), or even more subtle coordinations (such
tion information from the potentiometer and then as the movements of the eye and head while
calculated the slope or inclination of the line at one is reaching for a glass in the cupboard). For
each moment. The slopes (called derivatives) of movements that are oscillatory, one measure of
the positions at each moment in time represent the coordination is to describe the temporal phasing
velocities at corresponding moments, indicating between the two body parts.
the rate of change in position. Then this information Consider the simple action of tapping two
is output onto the same record as the position fingers on a table. Suppose we plotted the dis-
information. Such a trace is useful here in showing placement records of the up-and-down tapping
that the maximum velocity (Vm) was about 2.7 ml s cycles of the right finger along the abscissa and
and that the achievement of maximum velocity the cycles of the left finger along the ordinate.
occurred at about 75 ms through the movement. Plotted separately, each would be represented as
Also shown are a gradual increase in velocity until a back-and-forth, overlapping straight line along
the peak velocity (the midpoint of the movement) its respective axis. However, to assess how these
is reached and then a decline toward the end. Such two fingers are coordinated, the time records
a trace gives a more complete description of the of one finger can be plotted relative to the time
movement than does positional information alone. records of the other finger. At any point in time,
the position of the left hand and the position of
Acceleration Trace c in figure 2.12 is a record the right hand are represented as a single data
of the acceleration at each moment of time. This point on the graph. When the points are combined
record was also obtained by the computer, which over time they produce one continuous trace, as
calculated the slope or inclination of the velocity illustrated in the two examples in figure 2.13 (see
curve at each moment. The slopes of the velocity also Winstein & Garfinkel, 1989). In figure 2.13a,
curve at each moment yield the accelerations and the two fingers tap the table at approximately
represent the rate of change in velocity. This output the same time, are at maximum height above the
is plotted along with the other two traces on the table (in the "up" position on the graph) at
same time scale. Initial acceleration lasts about about the same time, and seem to be moving
100 ms until the acceleration trace returns to zero. within their respective cycles at about the same
Then there is a deceleration (a negative accelera- time-the coordinated motions are simultaneous.
tion trace) that lasts for about the same length of The example in figure ~.13b illustrates alternate
time. Also, the peak velocity of the movement is tapping-one finger taps the table at about the
achieved at the point at which the acceleration same time that the other finger reaches maximum
changes to deceleration (where the acceleration height, then vice versa. This figure illustrates
curve crosses the zero baseline). just two of the types of temporal coordination
These kinematic variables, with simultaneous patterns that exist between oscillating effectors
recording of position, velocity, and accelera- (see chapter 8).
tion as a function of time, provide a reasonably One can also obtain a quantitative measure
complete picture of these movements. Scientists of temporal coordination by considering the
often search for changes in these kinematic displacements of each cycle over time. This is
variables when certain independent variables represented in figure 2.14 in two ways. In figure
are changed-for instance, instructions to the 2.l4a, displacement of one finger is plotted over
subject or the size of a target to which the person time-position A represents the finger at the time
is moving. Examples of this kind of research are of a tap, position Bis about halfway up, position
provided later in the text. C represents the "up" point, and position D is
Coordination Kinematics In chapter 8 we halfway back down again. Figure 2.14b repre-
will focus on movement coordination-how the sents the same data by plotting these one-finger
42 Motor Control and Learning

Up

Tap

Tap Up Tap Up
a Right finger b Right finger

FIGURE 2.13 Sample displacement plots of two fingers moving simultaneously. (a) In-phase coordination; (b) anti-phase
coordination.

displacements against their own velocities (called phase angle (<I>) indicating the progress through
a phase-plane representation). Now positions A a cycle, or a circle, containing 360°.
and C represent zero velocity, and positions B Since the phase planes of each finger can be
and Dare the maximum upward and downward determined independently, the measure of coor-
velocities, respectively. The value of phase-plane dination is simply the difference between the
representations is that the position of each finger phase angle for the left finger (<PL) and the phase
at any point within its cycle can be described as a angle for right finger (<PR). This measure is called

c:
0
E
Ill
~

Time

Position

a b

FIGURE 2.14 Repetitive movement of a finger: displacement (a) relative to time and (b) relative to its velocity.
Figure 2.14a Adapted from Kelso et al. 1985.
Figure 2.14b Adapted from Burgess-Limerick, Abernethy, and Neal 1991.
 Methodology for Studying Motor Performance 43

relative phase, because the measure represents the Goniometers can be wired with potentiometers
position of one finger within its cycle relative to that send voltage information (proportional to
the position of the other finger within its cycle. the joint angle); this can be accumulated and
From figures 2.13a and 2.14, it is clear that at analyzed by a computer. Potentiometers are also
any one time, the two phase angles are about the used in many other direct measurement devices,
same. That is, when both fingers are down," each
/1
such as a computer mouse, that signal changes in
phase angle is 0°; when both fingers are "up," position when physically moved by the subject.
each phase angle is 180°, and so on. Thus, the A graphics tablet is another device that records
relative phase (<I\ - <l>R) will always be close to the changes in contact positions with the surface.
0° whenever a sample is measured. This simul- Muybridge (see "The Movement Photographs
taneous pattern of coordination is often referred of Eadweard Muybridge" on p. 38) is often cred-
to as moving in-phase. However, for figure 2.13b, ited with initiating the analysis of human move-
notice that the temporal phasing of one finger is ment through imaging techniques. High-speed
always nearly exactly opposite that of the other cinematography introduced a way to capture
finger. That is, when one phase angle is at 0°, the images of moving limbs many times each second.
other is at 180°; then when the first is at 180°, the However, frame-by-frame analysis methods were
other has reached 360° (or 0°). The relative phase tedious ways of examining changes in locations
for this series of estimates produces an average over time (Bernstein & Popova, 1930/2003).
relative phase of about 180°, which is sometimes Fortunately, technology has introduced more
referred to as moving in an anti-phase pattern of automated ways of performing these analyses.
coordination. Such instruments have relieved the scientist of the
But notice also that there is some variability in time-consuming job of reading locations from each
the plots presented in figure 2.13. Thus, research- frame of film, a method that also made the cost of
ers often calculate the standard deviation of these such analysis systems prohibitive for many years.
relative-phase samples in order to supplement the Computer systems that analyze such data are now
description of a pattern's average relative phase more powerful and much cheaper, and the use of
with an estimate of its stability. The measure of these measurement techniques is now the norm.
relative-phase variability is determined in a way The most common imaging devices use video
similar to that for the measures of VE, discussed and optoelectric methods. Video methods were
earlier in the chapter. Both the mean and standard straightforward, as movements could be cap-
deviation measures of relative phase will be used tured on relatively inexpensive videotape using
to describe important features of coordination in VHS and 8 mm formats. Typically, the subject
chapter 8. was recorded on videotape wearing pieces of
tape or fluorescent "markers." The locations of
Movement Measurement Devices these markers were later digitized for analysis.
Measurement recording systems, used to collect For optoelectric method_s, tiny light bulbs (called
data about human movement, have undergone light-emitting diodes or LEDs) are attached to
many changes over the years. Just as "moving the subject on various body parts. Light-sensing
picture" technology evolved from the efforts devices then detect and record automatically the
of Muybridge and others to capture on film the locations of the LEDs during the movement. As
moment-to-moment changes in body positions, illustrated in figure 2.15, an advantage of these
new developments in measurement technology newer systems is the capability to record infor-
have often been created to satisfy the needs of mation in three dimensions, providing greater
researchers to make recordings with increased information about movements and greater flex-
precision and economy (in terms of time and ibility in the types of actions that can be studied
effort). in experiments.
One of the simplest methods to collect body The most recent advances in movement record-
position information is to acquire the information ing technology have combined the use of robotics
directly through various means. For example, a and virtual reality environments (see figure 2.16).
goniometer is a hinged device that, when strapped These systems not only record movement with
to the side of a body joint, physically changes fine precision, but also have great flexibility to
in angle along with changes in the joint angle. provide diverse perceptual inputs. For example,
 FIGURE 2.15 Three-dimensional optoelectric system.
Photo courtesy of Northern Digital Inc.

FIGURE 2.16 Robotic system in a virtual environment.

© SensAble Technologies, Inc. www.sensable.com/documents/images/premium_3.0_6doUarge.jpg

44
 Methodology for Studying Motor Performance 45

the robotic system can induce different types perform a four-phase movement involving
and magnitudes of haptic feedback (information rotation of the right wrist. The record at the top
about touch) during movement. When combined of the figure shows the clockwise (supination)
with virtual reality feedback information, these and counterclockwise (pronation) movements.
robotic systems can simulate the look and feel of The EMGs are from the pronator teres muscle,
complex skills (e.g., surgical techniques) without which acts as the pronator (counterclockwise),
the dangers that might otherwise be present if and the biceps muscle, which acts as the supina-
practice were undertaken in the "real world" tor (clockwise). Once the movement begins, one
(e.g., on a patient). can see marked activity in the various muscles;
the activity is dependent on the particular action
Electromyography
being performed. The pronator teres is the first
Another method for describing movement char- muscle to act, throwing the wrist into pronation;
acteristics is to measure the involvement of a then the pronator is turned off and the biceps
muscle in a movement by recording the electrical acts to brake the action and reverse it; then the
activity associated with its contraction. The sim- pronator brakes and reverses that action, and so
plest method is to attach (with adhesive collars) on. These records describe the temporal patterning
recording electrodes to the skin surface over the of the movement segments. Information about
involved muscle; then, this weak signal from the the intensity of contraction is also provided by
muscle is amplified and recorded on a polygraph the amplitudes in these records, with larger EMG
recorder or computer for later analysis. Occa- amplitudes being generally indicative of larger
sionally subcutaneous electrodes are used; the forces. However, while the relation between
electrode is placed just under the skin but above EMG amplitude and force under static, controlled
the muscle belly. Or a small wire electrode can conditions within a given muscle is good, many
be embedded within the muscle so that electrical situations arise that can degrade this relation,
activity in small portions of the muscle can be so that the amount of force produced is usually
recorded. not accurately reflected by the amount of EMG
A recording using surface electrodes, taken being produced.
from a study by Carter and Shapiro (1984), is A record of transformed EMG activity, taken
shown in figure 2.17. Subjects were asked to during a rapid elbow extension (from Wadman

FIGURE 2.17 A typical electromyogram tracing taken during a movement.

Journal of Neurophysiology by M.C. Carter and D.C. Shapiro. Copyright 1984 by AMERICAN PHYSIOLOGICAL SOCIETY Reproduced with permission of AMERICAN
PHYSIOLOGICAL SOCIETY in the format Textbook via Copyright Clearance Center.
46 mi Motor Control and Learning

et al., 1979), is depicted in figure 2.18. A number other records of kinematic information, so that the
of changes were made in the raw EMG signals changes in the muscle actions can be associated
before they were plotted. First, the EMGs were with the resulting actions of the limbs.
rectified; that is, the negative voltage values were
Measures of Eye Movements
given positive signs so that the resulting record
would be completely positive. (Notice that figure Motor skills involving the use of the upper
2.18 has two such records, with the biceps record limbs constitute a large proportion of our daily
inverted so that the two patterns can be com- activities. How we use our hands and fingers to
pared more easily.) When the EMG is rectified, press the buttons to make a phone call, pick up a
the pattern of electrical activity can be seen more cup from the table, or move a mouse-controlled
readily than with the raw signals shown in figure pointer across a computer monitor depends con-
2.17. Second, these records were averaged for a siderably on our visual system. We use vision to
number of similar movements, mainly so that determine where we want the hand to go and
the important patterns of contractions could be then to update the relative success of that move-
seen over and above the trial-to-trial variations. ment as it approaches the target. Researchers
These patterns are more reliable than are those have used eye movement recording systems in
for a single trial. their investigations to determine where someone
is looking (that is, where the eyes are directed),
for example during an aiming movement of the
hand or in making a left turn in an automobile.
These recording systems can provide accurate
measures of what the subject sees, through two
video imaging techniques. One camera mounted
on the head or helmet of the subject provides an
image of the subject's line of vision. This visual
gaze is then coordinated with an eye-tracking
device that measures the movements of the eyes
by means of corneal reflection. The calibration of
the two recording devices provides an accurate
measure of where the eye is directed during the
performance of a motor task, sometimes referred
to as point-of-gaze information. One caveat here
is that what the person "sees" (i.e., perceives visu-
ally) is not perfectly related to where the eye is
pointed. In some situations, even though the eye
is directed at an object, the object is for various
reasons not actually "seen." The mere fact that a
FIGURE 2.18 Rectified and averaged electromyogram person is looking at an object does not guarantee
signals from the triceps and biceps muscles during a rapid
that he actually perceives it.
elbow extension.
Reprinted from W.J. Wadman et al., 1979, "Control of fast goal-directed arm
Figure 2.19 provides a good example of how
movements," Journal of Human Movement Studies 5: 10. By permission of W.J. the information from eye-tracking devices can be
Wadman.
coordinated with other information to provide
a more informative description of the motor
Such records are useful in that they provide behavior of a subject in a particular task. In the
one kind of description of what the central ner- example in figure 2.19, the subject is moving a
vous system "tells" the muscles to do. In the hand forward to point at a target that is 40 cm
example shown, it appears that the triceps muscle away from the initial "home" location. The top
contracted for about 100 ms; then it turned off half of the figure provides data from the eye-
and the biceps muscle contracted for about 50 tracking measurement system, showing that the
ms; and then the triceps muscle came on again for eye performs a rapid relocation from the home
another burst of about 100 ms. These records are position to a point about 4 cm short of the 40 cm
even more helpful if they are superimposed on target (indicated by the arrow denoting point of
 Methodology for Studying Motor Performance II! 47

associate the movements of body parts with the

'E perceptual information that may be received
~ 50 during action. Such techniques may be informa-
-
CJ
c.. tive when one is determining, for example, how
c 40
CD visual information is used during action (chapter
E
CD
u 5), or how athletes of varying levels of skill use
ltl 30 visual information in performance (chapter 9).
c.
Ill
:a Figure 2.20 illustrates the use of eye tracking to
-
iii
N
c
0
·;::
20

10
measure point-of-gaze information during the
performance of a golf putt.
0
::c
Measures of Brain Activity
15,000 The most dramatic advances in measurement in
recent years have been the various methods of
10,000 measuring brain activity. Their use in describing
N'
~
the brain activity that underlies motor perfor-
E 5000 mance is still in development; and the prom-
.§.
c ise of faster, more flexible, and more accurate
0
;;
0 advances in technique remains a technological
ltl
.... challenge. Each of these techniques has advan-
CD
Cii -5000
u
u tages and disadvantages, although changes in
ct methodology are occurring at a rapid pace and
-10,000
many of the current disadvantages may yet be
-15,000 overcome.
200 400 600 800 Electroencephalography (EEG), which has
Time (ms)
been around the longest, involves the recording
of electrical changes that occur in the brain as
FIGURE 2.19 Acceleration traces of the finger, elbow,
recorded from the scalp, more or less as EMG
and shoulder, relative to the changes in horizontal dis-
placement in point of gaze, during a rapidly aimed move- measures electrical activity of a muscle via the
ment. skin adjacent to the muscle. Another type of
Adapted from Helsen et al. 2000 and 1998. encephalography is known as magnetoencepha-
lography (MEG). As the name suggests, this tech-
nique measures changes in the magnetic fields
gaze). The bottom half of the figure illustrates involving brain activities. Both techniques have
kinematic information regarding the accelera- the advantage of working very quickly, providing
tion profiles of the finger, elbow, and shoulder precise measures of the timing of mental events.
as the upper arm moves toward the target. These However, their capacity to allow inferences about
kinematics were provided by an optoelectric localized anatomical structures is significantly
measurement system that, while separate from less than that of other, albeit slower, methods
the eye-tracking system, was calibrated in real (Haynes & Rees, 2006).
time in order that these measurements could In contrast, methods such as (a) positron
be coordinated later to provide the kinds of emission tomography (PET), (b) single-photon
data illustrated in figure 2.19. The dotted line emission computed tomography (SPECT), and
in the figure illustrates that the end of the initial (c) functional magnetic resonance imaging (fMRI)
rapid-eye movement toward the target (called are all neural imaging techniques that provide
a "saccade") corresponded very well with the much more detailed information about the local-
point of peak acceleration of the finger. A second ization of brain structure and activity. The latter
saccade of the eye, initiated about 100 ms later, technique (fMRI) is based on the fact that when a
again corresponded with an increase in the specific part of the brain is active in the process-
acceleration profile of the finger. Methods such ing of information, neural activity increases and
as these have provided a rich source of informa- consequently oxygenated blood to that region of
tion about motor behavior, as researchers can the brain increases. The fMRI records the "BOLD"
48 Motor Control and Learning

FIGURE 2.20 Eye-tracking setup.

Reprinted, by permission, from J. Vickers, 2007, Perception, cognition, and decision training: The quiet eye in action (Champaign, IL: Human Kinetics), 59.

(blood oxygen level dependent) signal as vectors TMS). Recent investigations have combined
of three-dimensional pixels (called "voxels"), TMS with £MRI so that the effects of stimulation
allowing researchers to map the location and on both behavioral and brain responses can be
volume of brain regions that are actively process- observed (Sack & Linden, 2003).
ing information during task performance. The These neural imaging and simulation tech-
use of new brain recording techniques in motor niques are useful in localizing the parts of the
control and learning experiments has increased brain that are active during the performance of
dramatically in the past decade, although some many activities. Indeed, many motor behavior
challenges remain, such as temporal resolution researchers are including these techniques as
and the nature of tasks that can be used with routine components of their investigations to
these devices. provide neural-anatomical evidence that aug-
Although not a brain activity "measurement" ments the motor behavior data obtained in their
device, an approach to studying brain function experiments.
that is frequently used today is called transcranial
magnetic stimulation (TMS). The TMS method
sends a brief pulse through a magnetic coil Measuring and Evaluating
that has a temporary effect on the region of the
brain above which the coil has been positioned. Relationships
Depending on where the magnetic coil is posi-
tioned, the brief pulse can either excite or inhibit An important process for evaluating the out-
the activity of that brain area. In contrast to what comes of various experimental procedures begins
happens in nonstimulated trials, the TMS-stim- after the main performance measures are gener-
ulated locations of the brain are assumed to be ated from the experiment. This process involves
active if a change in task performance is observed determining the relationship between some inde-
(see Hallett, 2007, for an excellent primer on pendent variable and a dependent variable on the
 Methodology for Studying Motor Performance 49

basis of the empirical data. One important kind of In this equation, Y represents the values on
relationship is linear. In a graph, the dependent the y-axis (error), X represents the values on the
variable plots essentially as a straight line with x-axis (average velocity), and a and b are con-
the independent variable. stants (figure 2.21). The constant a is termed the
y-intercept, and it refers to the value of Y when
linear Relationships the line crosses the y-axis; here the value is about
In figure 2.21 we have plotted data from an 2 mm. The constant bis called the slope and refers
experiment by Schmidt and colleagues (1979), to the amount of inclination of the line. The slope
for which the error (measured as VE and called can be positive (upward and to the right, as in this
effective target width, We) in hitting a target with example) or negative (downward to the right),
a handheld stylus is plotted as a function of the associated with either positive or negative values
average velocity of the movement. A quick exami- of b, respectively. Once these values are specified
nation of the plot in figure 2.21 indicates that from a given set of data, the empirical equation
the relationship between these two variables is that describes the linear relation between values
essentially linear. A line has been placed through of X and Y can be written.
the points that seems to represent their general
direction, and this line is called the line of best fit. Computation of Constants
The actual placement can be done accurately by Computation of the constants needed for the
various statistical techniques (e.g., regression) or empirical equation is simple. After the line of best
can be done "by eye" for a rough approximation. fit has been applied to the data points, extend it
The goal in this section is to express this line of leftward until it crosses the y-axis and read off
best fit in terms of what is known as an empirical they-intercept, or a. In the data shown in figure
equation, a kind of shorthand that enables us, with 2.21, a equals 2.08 mm.
but two numbers, to convey information about a Next, draw two lines, one perpendicular to the
linear relationship for an empirically determined y-axis and one perpendicular to the x-axis, form-
set of data points. These two numbers also will ing the shaded triangle as shown. The length of
have special meaning in terms of various theo- the line forming the base of the triangle will be
ries; that is, these numbers will be measures of called AX, and the length of the line forming the
certain hypothetical constructs. We begin with side of the triangle will be called AY. The symbol
the general equation for a line: A means "change in"; measures of the changes
Y = a + bX (2.8) in Y (6.7 mm) and the corresponding changes in

8
~7
E
E 6
c:

is
0.... 4
w3
2

50 100 150 200 250

Average velocity (cm/s)

FIGURE 2.21 Graphical method for determining constants for linear empirical equations.
Reprinted, by permission, from A.A. Schmidt et al., 1979, "Motor-output variability: A theory for the accuracy of rapid motor acts,'' Psychological Review 86: 427.
50 Iii Motor Control and Learning

X (201 cm/ s) can be seen. Then, the slope of the Interpreting Empirical Equations
line is defined as In addition to the benefits provided by empirical
b=~Y I ~X (2.9) equations in terms of description of experimental
results and prediction of new findings, the values
That is, the slope is defined as the change in Y of the constants a and b often have special theoreti-
divided by the corresponding change in X. Here, cal meaning, depending on the nature of the data
the slope (b) is computed as 6.7 I 201 = +0.033. collected and the kind of independent variable
The interpretation of this slope is that each time studied. In the present example, the meaning of the
the value of X increases by 1 cm/ s, there is a 0.033 constant a (the intercept) is related to the amount
mm increase in the Y value (error). of error for the slowest movement possible, and
Uses of Empirical Equations thus the intercept seems to represent a kind of
The slope and intercept are the only two values "background" or "baseline" error. On the other
needed to determine the linear relationship. Put- hand, the value of the slope (b) refers to the amount
ting the slope and the intercept together into the of increase in error as the velocity increases, and it
general equation for a straight line results in the represents a measure of the "difficulty" of the task.
empirical equation for these data: In this and other similar situations to be discussed
later, the slope and intercept describe two distinct
Y = 2.08 + 0.033X features of the task or the subject's behavior.
Having been provided the calculated values of
a and b found by a person in California, someone Correlation and Regression
in Munich can reconstruct the line of best fit by A statistical tool that is used often in motor
using the linear equation. This is done by picking behavior research, and that has mathematical
any two arbitrary values of X (say, 50 and 250 properties similar to those of the simple linear
cm/s) and calculating the values of Y for these regression method, is the correlation. The correla-
values of X: tion and regression methods are used to establish
Y = 2.08 + 0.033 (50) = 3.73 mm the degree of association between two measures,
as seen in the study of individual differences,
Y = 2.08 + 0.033 (250) = 10.33 mm for example (in chapter 9). For a relatively large
Then, on a new graph, these data points (X = group of subjects (e.g., 50), we begin with two
50, Y = 3.73; and X = 250, Y = 10.33) can be plot- different tests administered to each person. The
ted, and the line in figure 2.21 drawn between degree to which the performances by individuals
them. Thus, saying that the intercept was 2.08 on one test are related to the performances of the
and the slope was 0.033 can convey a great deal same individuals on the other test is reflected in
of information about the experimental results to the size of the correlation coefficient. The correla-
someone who does not have access to the nine tion coefficient expresses the amount of shared
actual data points. association that exists between the two data sets,
In addition, this relation can be used to predict with no implications about whether or not one
new values of error before they are found. If we variable caused the other to change.
wanted to choose a velocity value so that the error Scaffergrams
was only 5.00 mm, we could take the empirical
equation and substitute the value of the error as One of the ways in which data from two tests can
follows, then solve for the value of the velocity: be described is by a special kind of graph called
a scattergram. Consider the data shown in table
5.00 = 2.08 + 0.033X 2.3, which have been plotted on the scattergram
x = (2.08 - 5.00) I (-0.033) in figure 2.22. The two axes in figure 2.22 are the
scales of the two tests, respectively; and each
X = 88.48 cm/ s
of the subjects is represented as a dot, located
Thus, if we wanted the error to be about 5 according to his scores on the two tests. The data
mm, we would use a velocity of about 88 cm/ s. are hypothetical scores that might be obtained
Having an empirical equation makes it possible on a common playground and consist of age (in
to predict this result without actually going into years) and the time for a 100 m dash (in seconds).
the laboratory to measure it directly. In figure 2.22, the scores for these 10 people are
 Methodology for Studying Motor Performance 51

12.8
12.0
10.5
16.1 22.4
15.2 23.7

7 16.0 22.5

8 14.1 25.5

9 13.0 27.7

10 18.0 20.0
Note: Speed data were computed from the time data.

plotted, so that each of the 10 dots on the graph between these two variables, and a line could be
represents each person's joint scores on the two drawn through the points to represent the "direc-
variables (age and running time). tion" in which the "cloud" of points is oriented.
A relationship apparently exists between the This "line of best fit" can be determined in exactly
score on the age variable and the score on the run- the same way as discussed earlier for empirical
ning test, indicating that as the age score becomes equations, and often involves computing regres-
larger, the number of seconds on the running test sion equations.
tends to become smaller, with some exceptions. Direction of the Relationship In this example, as
In general, the 10 subjects showed a relationship the value of the age variable increases, the value of
the running time variable tends to decrease. This
kind of relationship is called an inverse, or negative,
relationship, and the equation representing this
18
relationship has a negative slope constant (b). In
other situations, we might find that as one of the
variables increases, the value of the other variable
16
tends to increase as well; this is a direct, or positive,
~
Cl> relationship. In such cases, the slope constant of
:S 14 the regression equation has a positive value, with
Cl
c: the line of best fit sloping upward to the right.
·~ 12
::i
The direction of the relationship shown in such
a: data is often dependent on the scoring system
10 used. Consider the data in table 2.3. In the fourth
column we have expressed each subject's 100
8 m dash scores as average running speed (km/h)
rather than as the time required to travel 100 m.
9 12 15 18
This change in scoring system "inverts" the group
6
Age (years) of scores, so that the person who had the largest
time score has the smallest speed score, and so
FIGURE 2.22 A scattergram showing the relationship
on. When the age data are plotted against the
between age and running time. (Data are from table 2.3, average running speed scores in figure 2.23, the
and each dot represents one of the 10 subjects.) relationship becomes positive, and the empirical
52 Motor Control and Learning

is its strength. By strength, we mean the extent

35
to which the relationship is perfectly linear, or
the extent to which all of the subjects' points fall
exactly on the line of best fit. Figure 2.24 shows
two scattergrams that represent relationships of
different strengths. In figure 2.24a, the relation-
ship can be considered quite strong because
nearly all the points fall close to the line of best
fit. In figure 2.24b, however, the relationship
is not very strong, because the points tend to
fall away from the line of best fit. These two
aspects, direction and strength, are the primary
descriptors used to characterize relationships.
20 They are independent, in that a relationship can
be either positive or negative and either strong or
weak.
6 9
Age (years) Predicting From a Relationship One of the
most important reasons scientists want to know
FIGURE 2.23 A scattergram showing the relationship the nature of the relationship between two vari-
between age and running speed. (Data are from table 2.3, ables is for the purpose of prediction. For example,
where speed is computed from running time; each dot rep-
resents one of the 10 subjects.) if we know that the relationship between age and
running time has been found to be as described
in figure 2.22, then on a new group of children
equation would now have a positive slope we can estimate (or predict) the 100 m time given
constant (b) rather than a negative one. Thus, a the age of a person not in the original data set
positive relationship is not any "better" than a without actually measuring running speed, just
negative one; the sign simply indicates the direc- as with the linear equations discussed earlier.
tion in which the line of best fit is sloped. Such procedures are used extensively in every-
Strength of the Relationship A second charac- day situations, as in predicting the probability of
teristic of the relationship between two variables having an automobile accident from one's age,

a
x b
x

FIGURE 2.24 Hypothetical scattergrams for (a) a strong and (b) a weak relationship.
 Methodology for Studying Motor Performance im 53

or predicting success in graduate school from ship, exactly as described in the previous sections.
achievement test scores. The absolute size of the correlation indicates the
As the strength of the relationship increases, strength of the relationship and hence is critical
the predictability of one variable from the other for evaluating the extent to which one can use
increases as well. When the relationship is perfect the relationship to predict. Figure 2.25 shows five
and all the individual data points fall exactly on hypothetical examples of correlations and the
the line of best fit, perfect predictions with no associated scatterplots. Both a +0.90 and a -0.90
error can be made. When the data are related less correlation are strong relationships, as all the data
perfectly, as in the example shown in figure 2.24b, points fall almost exactly on the lines, although
then more error is introduced into the predictions. the lines are sloped in opposite directions. Cor-
Thus, the strength of the relationship-but not the relations of +0.50 and -0.50 are moderate in
direction-is the primary determinant of the extent strength, and the points fall considerably away
to which a relationship can be used to predict. from the lines of best fit. A correlation of zero is
weakest, indicating that no predictive capability
Correlation Coefficient is possible between these two variables. Formulas
These concepts of strength and direction of for the calculation of correlations can be found
relationships can be quantified using a statistic in most statistics textbooks (e.g., chapter 8 in
called the correlation coefficient. The correlation, Thomas, Nelson, & Silverman, 2005; chapter 7
abbreviated r, ranges from +1.0 through zero to in Vincent, 2005).
-1.0. The two important aspects of the correlation A convenient method for comparing the
are the sign and its absolute size. The sign of the strength of relationships between tests is to square
correlation indicates the direction of the relation- the correlation coefficient and multiply by 100 to

y y

x x

y y

x x

FIGURE 2.25 Scattergrams for hypothetical data showing high, moderate, and low relationships.
54 Motor Control and Learning

convert it into a percentage score. Generally, the error (AE, also a measure of overall error). Each
square of the correlation coefficient indicates the error measure has a different meaning and is used
extent to which two tests can be considered to in different aspects of the measurement process.
measure the same thing, and represents the per- Measures of speed are used when accuracy is
centage of shared variance (or, the percentage "in less important (or is controlled) and when rapid
common") between the two variables. This fits actions are critical. Measures of magnitude are
nicely with the earlier discussion of the strength used when the amount of behavior is critical. A
and direction of relationships. The shared vari- fourth but related measure is based on the analy-
ance for two tests that correlate positively (+0.90) sis of simultaneous secondary tasks, providing a
or two others that correlate negatively (-0.90) is measure of the spare capacity of the performer
the same (e.g., +0.902 x 100 = -0.902 x 100 = 81 %), after she has devoted attention to a primary task.
indicating that the direction of the relationship Movements can be measured in many ways,
(positive or negative) has no effect on the amount but common methods involve the calculation of
of shared variance. kinematic variables (position, velocity, accelera-
The r2 value is useful for interpreting the tion) and the recording of the electrical activity
strength of two relationships. For example, sup- from muscles (EMG). Methods used to assess
pose two variables correlate with a value of r kinematics have changed dramatically with digi-
= 0.50 and two other variables correlate with a tal technology, and precise measures of sensory-
value of r = 0.25. In this case, the r2 value of the motor control are now the norm. Studies of brain
first relationship (0.50 2 = 0.25) indicates approxi- activity are also becoming increasingly popular,
mately four times more shared variance than for and their use in motor behavior research contin-
the second relationship (0.25 2 = 0.063), or that the ues to grow. Linear empirical equations provide a
first two tests had approximately four times more description of a linear relationship between two
in common than did the second pair. variables. The parameters of the equation can be
easily estimated, and they provide a means by
which the relationship can be used to predict facts
Summary that are yet unknown. The parameters can also
be useful in describing the direction and strength
Motor behavior can be classified according to of the relationship between two variables as indi-
several dimensions, such as (a) continuous/serial/ cated by the correlation coefficient.
discrete, referring to the extent to which the move-
ment has a definite beginning and end, and (b)
open/closed, referring to the extent to which the Student Assignments
environment is predictable. Most of the tasks used
in motor control work fall into one or the other of 1. Answer the following questions and bring
these basic categories. In measuring movement, the information to class for discussion:
experimenters devote attention to a measure's a. Compile a new list of examples from
objectivity (the extent to which two independent activities of everyday life that fit each
observers achieve the same score), its sensitivity to of the following task categories: dis-
changes in skill, its reliability (the extent to which crete/open, discrete/closed, serial/
the score is repeatable), and its validity (the extent open, serial/closed, continuous/open,
to which the test measures what the experimenter continuous/closed.
intends it to measure).
The outcome of movements in terms of the b. Close your eyes and draw 10 lines as
environmental goal can be measured in essen- close as possible to 10 cm in length.
tially four ways: in terms of errors, speed or time Do not open your eyes until you are
(or both), or magnitude, or with various second- finished. Calculate the following error
ary tasks. There are many ways to measure errors measures: CE, ICEI, VE, E, and AE.
in movement; chief among these are constant error c. Discuss the relative importance of
(CE, a measure of average error or bias), variable reaction time in the following Olympic
error (VE, a measure of inconsistency), total vari- events: 100 m run, 10,000 m run, 100
ability (E, a measure of overall error), and absolute m butterfly swim, 400 m relay.
 Methodology for Studying Motor Performance 55

2. Find a research article published in the past This site provides some examples of Muybridge' s
five years that uses functional magnetic photographs:
resonance imaging (fMRI) or transcranial
magnetic stimulation (TMS) to better www.masters-of-photography.com/M/muybridge/
explain basic motor control processes. muybridge.html

Web Resources Notes

1 A quantitative analysis of error scores in two dimen-
This resource is related to reaction-time method- sions is provided by Hancock, Butler, and Fischman
ology and the history of RT research: (1995), but is beyond the scope of the present discussion.
www.chss.montclair.edu/psychology/museum/
mrt.html
 CHAPTER 3
·. · •·. · . · ·•. ·.·• .· •·. •·. N·.·•.• ··.o
HUMAN. .•··. ·•. ·.• · . · ·.••·R
·.·.···.·.·.1.···
... F .·• .··.••··. •:·. ·.~l ·!·;f~,
. · . · •. •.· .·• .·•·••.-.•
.· .··. ·.· ·. . •.·•. •. . •.
1

P.ROCESSml~~Ji,.o~:
.•···.•.·····..•.•·.·.•
. ..··•·...•.•.• ·...· .•••·. . .•··•..····:·.··i·····················l····.. .·.:
. .•..•
•·.•.•. ·...·•.•.·.· •· •· · ...•.·· •.•.•.•. .:·.· ·.....·.· •...·•.•.•.·. ·...·.•..•·.·.·.m
•. ....••.·..•.•.•·.·..•.•.·....••.·.•.·
..·•·.·..··.··•···...
·....•·..··.·..··.··...·......
• ·... . •

uman functioning in the environment can cessed information can result in various kinds
be conceptualized and studied in many of movement, just as a computer displays the
ways; one of the most popular is based on the results of its processing on a monitor or sends it
fundamental notion that humans are processors elsewhere as an output. The goal of this chapter
of information. It is assumed that information is is to discuss information processing as it relates
available in the environment; that the individual to human motor behavior (see also Marteniuk,
accepts this information into various "storage 1976)-specifically how information is processed
systems" called memory; and that the informa- for the specific purpose of producing skilled
tion is then "processed" for the purposes of movement.
perception, decision making, and action. Human
information processing is based on a computer
metaphor. According to this metaphor, we "take
The Information-Processing
in" information from outside sources, just as a Model
computer takes in information via input devices.
That information undergoes transformations, The model begins with the input of informa-
uses other information stored in memory, and tion from the environment through one or more
is subject to certain limitations based on the of the sense organs, and then considers what
amount processed and the speed of processing happens to this information once it is inside
that is available; again, this is similar to how the system. A typical "black box" model of the
a computer processes information. And lastly, process is shown in figure 3.1. The individual is
the process by which information is "output" considered to be the box, and information enters
has many analogies to the computer-the pro- the box from the environment. This information

Output
Input
(Signals) --+ --+ (Motor
response)

FIGURE 3.1 The simplified information-processing model.

57
58 Motor Control and Learning

is then processed in various ways, eventually viewpoint, the general concept of stages of pro-
resulting in output as observable motor activity. cessing is quite old, dating back to the research of
This black box model prevailed in the stimulus- Danders (1868/1969; see "Danders' Subtractive
response (S-R) tradition (see chapter 1), in which Method"). This thinking, coupled with efforts
researchers were primarily concerned with the from cognitive psychology (e.g., Kellogg, 2003;
relationship between what went into the box Sanders, 1980; Schweickert, 1993; Sternberg,
(the information, or stimuli) and the output 1969), has led to the view that various processing
from the box (the response). With the emergence stages can be defined and that these stages can
of cognitive psychology, however, interest began be either serial (sequential) or parallel (simultane-
to focus on the processes that occur within the ous) in nature.
box. Obviously, this is an abstract way to study Danders argued that with use of these
human behavior because it concerns processes methods, the durations of the stages of process-
and events that are not directly observable. ing involved in stimulus discrimination and
Knowledge about these processes is inferred response selection, which are not directly observ-
from the overt behavior of the human under able, could be estimated in the laboratory. It is
various experimental conditions. Events occur- remarkable that these insights were developed in
ring inside the box can be studied in a variety of 1868, long predating the cognitive revolution in
ways through use of the cognitive-psychological experimental psychology that made use of these
perspective. and similar concepts.
The most common approach to the study of Danders' general methods and assumptions
information processing is to consider the dura- would later be revived in a classic paper by Saul
tions of these various processes. This chronometric Sternberg (1969) that generated considerable
approach (see Posner, 1978) makes considerable interest and further research, some of which
use of the reaction-time (RT) method, whereby was involved in understanding motor behavior
the chief measure of the subject's behavior is the (e.g., Sanders, 1980). Later thinking and research
interval between the presentation of a stimulus identified a number of flaws in Danders' subtrac-
and the beginning of the response. Many dif- tive method (see Massaro, 1989, for an analysis
ferent information-processing activities occur of Danders' and Sternberg's logic and methods).
during RT; but if the experiment is designed But even so, Danders' basic idea that we could
properly, so that the durations of most other examine the duration of stages by subtracting the
processes are held constant, one can usually RTs in various conditions was remarkable, given
infer that an increase in RT resulting from some the time of his work, and served as the foundation
experimental variable was caused by the length- for more modern analyses of human information
ening of the duration of a particular process. This processing. 1
chapter presents many examples of this type of
research and thinking, and the chronometric The Nature of Serial
approach to studying information processing and Parallel Processing
constitutes a large portion of it. Later in the
Imagine an automotive plant as a model of infor-
chapter we consider other aspects of informa-
mation processing, as in figure 3.2. Some stages
tion processing that are not necessarily studied
occur at the same time in different places, such as
using the chronometric approach but are par-
the fabrication of the electronics, the assembly of
ticularly important in the performance of motor
the engine, and the assembly of the chassis. But
skills.
at various times during the overall process, these
components are combined to complete the final
Three Stages assembly stage. After final assembly, imagine a
test-drive stage, in which the drivers search for
of Information Processing problems before the car is sent to the dealers, as
yet another stage. This simple analogy contains
Although the notion of separate stages or pro- an example of both parallel processing (simulta-
cesses between a stimulus and a response has neous assembly of the electronics, engine, and
been popularized by the cognitive-psychological body stages), and serial processing (successive
 Donders' Subtractive Method
Over a century ago, the Dutch physician F.C. Donders (1868/1969) made the first attempts