THE DEVELOPMENT

OF ARITHMETIC
CONCEPTS AND SKILLS:
CONSTRUCTING ADAPTIVE EXPERTISE
 STUDIES IN MATHEMATICAL THINKING AND LEARNING

Alan H.Schoenfeld, Series Editor

Artzt/Armour-Thomas Becoming a Reflective Mathematics Teacher: A Guide for

Observation and Self-Assessment

Baroody/Dowker (Eds.) The Development of Arithmetic Concepts and Skills: Constructing

Adaptive Expertise

Boaler Experiencing School Mathematics: Traditional and Reform Approaches to Teaching

and Their Impact on Student Learning

Carpenter/Fennema/Romberg (Eds.) Rational Numbers: An Integration of Research

Clements/Sarama/DiBiase (Eds.) Engaging Young Children in Mathematics: Results of the

Conference on Standards for Pre-School and Kindergarten Mathematics Education

Cobb/Bauersfeld (Eds.) The Emergence of Mathematical Meaning: Interaction in Classroom

Cultures

English (Ed.) Mathematical Reasoning: Analogies, Metaphors, and Images

Fennema/Nelson (Eds.) Mathematics Teachers in Transition

Fennema/Romberg (Eds.) Mathematics Classrooms That Promote Understanding

Lajoie Reflections on Statistics: Learning, Teaching, and Assessment in Grades K–12

Lehrer/Chazan (Eds.) Designing Learning Environments for Developing Understanding of

Geometry and Space

Ma Elementary Teachers’ Mathematical Knowledge and Its Relationship to Teaching

Competence: A United States-China Comparison

Martin Mathematics Success and Failure Among African-American Youth: The Roles of
Sociohistorical Context, Community Forces, School Influence, and Individual Agency

Reed Word Problems: Research and Curriculum Reform

Romberg/Fennema/Carpenter (Eds.) Integrating Research on the Graphical Representation

of Functions

Schoenfeld (Ed.) Mathematical Thinking and Problem Solving

Senk/Thompson (Eds.) Standards-Based School Mathematics Curricula: What Are They?

What Do Students Learn?

Sternberg/Ben-Zeev (Eds.) The Nature of Mathematical Thinking

Wilcox/Lanier (Eds.) Using Assessment to Reshape Mathematics Teaching: A Casebook for

Teachers and Teacher Educators, Curriculum and Staff Development Specialists

Wood/Nelson/Warfield (Eds.) Beyond Classical Pedagogy: Teaching Elementary School

Mathematics
 THE DEVELOPMENT
OF ARITHMETIC
CONCEPTS AND SKILLS:
CONSTRUCTING ADAPTIVE EXPERTISE

Edited by
Arthur J.Baroody Ann Dowker

University of Illinois Oxford University

at Urbana-Champaign

LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS

Mahwah, New Jersey    London
This edition published in the Taylor & Francis e-Library, 2009.

To purchase your own copy of this or any of Taylor & Francis or Routledge’s
collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.

Copyright © 2003 by Lawrence Erlbaum Associates, Inc.

All rights reserved. No part of this book may be reproduced in any form,
by photostat, microfilm, retrieval system, or any other means, without prior
written permission of the publisher.

Lawrence Erlbaum Associates, Inc., Publishers

10 Industrial Avenue
Mahwah, NJ 07430

Cover design by Kathryn Houghtaling Lacey

Library of Congress Cataloging-in-Publication Data

The development of arithmetic concepts and skills: constructing adaptive

expertise/edited by Arthur J.Baroody and Ann Dowker.
p. cm.—(Studies in mathematical thinking and learning)
Includes bibliographical references and index.
ISBN 0-8058-3155-X (cl: alk.paper)
ISBN 0-8058-3156-8 (pbk : alk. paper)
1. Arithmetic-Study and teaching (Elementary)
I. Baroody, Arthur J., 1947– II. Dowker, Ann. III. Series.

QA135.6.D49 2002
372.7’2044—dc21
2002033866
ISBN 1-4106-0721-6 Master e-book ISBN
 Dedication

We dedicate this book to Herbert P.Ginsburg and Neil O’Connor, two exceptional
mentors.
 Contents

Foreword Giyoo Hatano xi

Preface Arthur J.Baroody and Ann Dowker xv

Chapter 1 The Development of Adaptive Expertise

and Flexibility: The Integration of
Conceptual and Procedural Knowledge
Arthur J.Baroody 1

Chapter 2 Does It All Add Up? Changes in Children’s

Knowledge of Addition Combinations,
Strategies, and Principles
Richard Cowan 35

Chapter 3 Two Perspective on Addition Development

Arthur J.Baroody and Sirpa H.Tiilikainen 75

Chapter 4 The Development of Children’s

Understanding of Additive Commutativity:
From Protoquantitative Concept to
General Concept?
Arthur J.Baroody, Jesse L.M.Wilkins, and
Sirpa Tiilikainen 127

Chapter 5 “You’ve Got to Carefully Read the Math

Sentence…”: Classroom Context and
Children’s Interpretations of the Equals
Sign
Kyoung-Hye Seo and Herbert P.Ginsburg 161

Chapter 6 Basic Multiplication Combinations: Passive

Storage or Dynamic Reorganization?
Brian Butterworth, Noemi Marschesini, and
Luisa Girelli 187

Chapter 7 Young Adults’ Strategic Choices in Simple

Arithmetic: Implications for the Develop-
ment of Mathematical Representations
Jo-Anne LeFevre, Brenda L.Smith-Chant,
Karen Hiscock, Karen E.Daley, and Jason
Morris 203

vii
viii Contents

Chapter 8 Language Supports for Mathematics

Understanding and Performance
Irene T.Miura and Yukari Okamoto 229

Chapter 9 Young Children’s Estimates for Addition:

The Zone of Partial Knowledge and
Understanding
Ann Dowker 243

Chapter 10 Multidigit Addition and Subtraction

Methods Invented in Small Groups and
Teacher Support of Problem Solving and
Reflection
Karen C.Fuson and Birch H.Burghardt 267

Chapter 11 Children’s Invention of Multidigit

Multiplication and Division Algorithms
Rebecca Ambrose, Jae-Meen Baek, and
Thomas P.Carpenter 307

Chapter 12 The Early Numeracy of Children With

Specific Language Impairments
Chris Donlan 339

Chapter 13 Mathematical Thinking and Learning

Difficulties
Nancy Jordan, Laurie B.Hanich, and
Heather Z.Uberti 361

Chapter 14 Neuropsychological Findings on

Conceptual Knowledge of Arithmetic
Margarete Delazer 385

Chapter 15 Arithmetical Savants

Lisa Heavey 409

Arithmetical Development: Commentary on Chapters 1

through 8 and Reflections on Directions
Jeffrey Bisanz 435

Arithmetical Development: Commentary on Chapters 9

through 15 and Future Directions
David C.Geary 455
 Contents ix

Glossary of Abbreviations 467

Author Index 469

Subject Index 481

 Foreword

Many educational researchers are interested in how students can be taught

subjects so that they develop adaptive expertise—the ability to apply meaningfully
learned procedures flexibly and creatively. From this perspective, school
instruction is successful when students are able to use what they have learned
to invent effective procedures for solving new problems. Simply being able to
complete school exercises quickly and accurately without understanding—what
we call routine expertise—is not particularly valuable. This is because such
competence is extremely limited in that it can be applied effectively only to
familiar tasks. Unfortunately, little is known about the process of cultivating
adaptive expertise.
Researchers who have studied transfer have almost unanimously concluded
that, when solving real-world problems, students seldom aptly apply schooltaught
procedures that were learned in a short period of time. Similarly, people who
have had years of experience solving problems in a given domain may be unable
to solve problems outside their experience. Such “experts” can solve familiar
types of “problems” quickly and accurately but may not understand why their
procedures work. As a result, when these experts are faced with a changed
condition or a new problem, they are unable to modify known procedures or
invent new ones. This is particularly true for experts in knowledgelean domains,
such as abacus operation, which I studied in detail in the early 1970s. However,
even in knowledge-rich domains, some experts’ knowledge may consist of
narrow, unconnected problem-solving schemas. As a result, these experts may
simply classify problems and apply the single-routine solution associated with a
particular problem type.
The notion of adaptive experts, which I introduced in Hatano (1982), was a
theoretical ideal rather than a model derived from a series of empirical studies.
I began by considering the following two questions: What kind of knowledge
do flexible and inventive experts construct? And how do they construct it? In
other words, I speculated about both the product of adaptive expertise and
its acquisition process. Regarding the former, my attention focused on the
conceptual knowledge underlying procedures. “Flexibility and adaptability seem
to be possible only when there is some corresponding conceptual knowledge to
give meaning to each step of the skill and provide criteria for selection among
alternatives possibilities for each step within the procedures” (p. 15). I assumed
(and still assume) that conceptual knowledge enables experts to construct
mental models of the major entities in a content domain, models that can be
used in mental simulations. I was very pleased to see that, in their formulation
of adaptive expertise in the Preface and chapter 1, the book’s editors emphasize
the importance of connections among pieces of knowledge, especially between
procedural and conceptual knowledge.

xi
xii Foreword

I also want to emphasize differences in the learning process between routine

expertise and adaptive expertise. Those who are becoming routine experts
(i.e., are acquiring knowledge in a conceptual vacuum) often fixate on a single
procedure, whether or not it makes sense, and care little about comprehending it.
In contrast, those who are constructing adaptive expertise (acquiring meaningful
knowledge) often explore a variety of possibilities and try to make sense of their
actions. Only the latter acquire the knowledge that can provide the direction and
constraints needed to solve novel problems.
I hypothesized that if people ask why a skill works or why each step is
needed during its application, such questions can lead them to develop some
conceptual knowledge related to that procedure. This is similar to what Donald
Schoen (1983) called “reflection-in-action,” which characterizes professionals
and distinguishes them from technicians. What is common between Schoen’s
and my formulations is that experts are seldom taught conceptual knowledge
in the verbalized form. Instead, they may construct it in the process of solving
problems or performing tasks in the domain.
In subsequent work, Kayoko Inagaki and I tried to specify motivational
conditions for adaptive expertise—that is, how to nurture this type of expertise.
In one such attempt (Hatano & Inagaki, 1992), we proposed the following four
conditions for promoting adaptive expertise: (a) encountering novel problems
continuously, (b) engaging in interactive dialogue, (c) being freed from urgent
external need, and (d) being surrounded by a group that values understanding.
In retrospect, such attempts were too individualistic and needed to be placed
in sociocultural contexts. We should have paid greater attention to the nature
of the socioculturally organized practice in which participants are learning. For
example, when the educational environment is oriented toward solving a fixed
class of problems skillfully, participants tend to become experts defined in terms
of speed and accuracy (i.e., routine experts). In contrast, when the educational
environment requires meeting varied and changing demands, they acquire
flexible and adaptive skills (adaptive expertise).
The most critical limitation of the formulation above was that the notion of
adaptive expertise was never applied to a variety of subject matter contents.
My coauthor and I focused on a single nonacademic example of cooking. This
book is an important contribution because its contributors extend our analysis
across a range of arithmetic topics, age levels, and contexts. These chapters
illustrate that flexibility and inventiveness in mathematics are closely linked to
a rich understanding of a topic—to a rich web of connections within a topic
(particularly those between procedures and concepts) and to other domains of
knowledge. Many of the chapters have clear implications about how to foster
adaptive expertise, and several include specific instructional guidelines. I am
honored that authors have found the notion of adaptive expertise useful.
—Giyoo Hatano
Keio University
 Foreword xiii

References
Hatano, G. (1982). Cognitive consequences of practice in culture specific procedural skills. The
Quarterly Newsletter of the Laboratory of Comparative Human Cognition, 4, 15–18.
Hatano, G., & Inaeaki, K. (1992). Desituating cognition through the construction of conceptual
knowledge. In P.Light & G.Butterworth (Eds.), Context and cognition (pp. 115–133). Hemel
Hempstead, England: Harvester Wheatsheaf.
Schoen, D. (1983). The reflective practitioner. New York: Basic Books.
 Preface

Educators and other scholars have long debated, often heatedly, how formal
instruction can best promote intellectual expertise (see, e.g., Wilson, 2003).
From ancient times to the present, the conventional wisdom has been that the
best way to promote school-taught arithmetic (and other content) is through
direct instruction and drill. For instance, whether the basic “number facts” were
presented on a clay tablet (as in Babylonian times and illustrated here) or as a
printed table of facts (as is the modern custom), the goal of traditional instruction
has been the same—memorize them by rote. Throughout history, critics of
traditional schooling have proposed reforms only to be greeted by resistance
or even vilification. For instance, in his comedy The Clouds, Aristophanes
described a debate between a proponent for the “Old Education” (traditional
schooling) and a reformer who supports the “New Education” (Socrates’ “Think
Academy”). The less than objective criticism leveled at the latter is that it would
“subvert manly self-control, turn young people into sex-obsessed rebels, and
destroy the city” (Nussbaum, 1997, p. 1).

A nine-times table circa 1,800 B.C. (Jacobs, 1982)

More recently, some proponents of reform (e.g., National Council of

Teachers of Mathematics, 2000) have recommended, for instance, that teachers
encourage children to discover relations and to invent procedures. Critics of this
view question why this is important when teachers can simply have students
memorize standard and efficient procedures by rote. For example, the columnist
Thomas Sowell (a senior fellow at the Hoover Institute, Stanford University)
noted, “Only people ignorant of history and too shallow to understand what
an enormous task it is to create anything could believe that children can reach
any serious understanding of even basic concepts by the faddish methods of
‘discov-ery’ and ‘creativity’ being used in public school classrooms…. Most of
us will never discover or create anything significant that was not understood”
(2000, p. A6).

xv
xvi Preface

What is the validity of each of these views? The focus of this book is two
related questions that lie at the heart of the current debate about reforming
mathematics education: What is the nature of arithmetic expertise? How
can instruction best promote it? We have brought together scholars from a
variety of specialties, including (cognitive, developmental, educational, and
neurological) psychology and (mathematics and special) education to offer
theoretical perspectives and much needed empirical evidence about these issues.
Both indicate the nature of arithmetic expertise and how to best promote it are
far more complex than conventional wisdom and many scholars—both past and
present—have suggested.
In the introductory chapter (chapter 1), Baroody traces the history of the debate
in the United States about whether computational skills should be memorized
by rote or learned in a meaningful manner. He draws three conclusions. The
first, in effect, is that children’s powerful informal knowledge is typified by what
Hatano (1988) called “adaptive expertise” (meaningful procedures that can be
flexibly applied to new, as well as familiar, tasks) rather than routine expertise
(rotely learned knowledge that can be applied to familiar tasks only). (Adaptive
expertise is similar to what Kilpatrick, Swafford, and Findell, 2001, called
“computational fluency,” which is the ability to use procedures appropriately,
efficiently, and flexibly.) The second is that adaptive expertise depends on
conceptual understanding (well-connected knowledge) and its integration with
procedural knowledge. The third entails providing an explicit definition of what
is meant by a conceptual basis for a procedural advance, something not done
in previous discussions of the relations between conceptual and procedural
knowledge (cf. Baroody & Ginsburg, 1986; Carpenter, 1986; Rittle-Johnson &
Siegler, 1998).
In chapter 2, Cowan summarizes an analogous historical debate in the
United Kingdom between educators who advocate a focus on computational
competence and those who advocate a focus on understanding and thinking.
(It is worth noting that there are wide cultural and historical variations in the
relative emphasis of adaptive expertise and routine expertise. For example,
unlike the United States and some other European countries such as France
and Italy, the United Kingdom has emphasized cultivating adaptive expertise
since the 1960s.) Cowan reviews the research literature in three areas critical
to early childhood mathematics instruction: the mastery of single-digit (basic)
addition combinations, the development of mental-calculation strategies, and
the construction of arithmetic generalizations (principles). As he notes, expertise
in all three areas is characterized by flexibility.
In chapter 3, Baroody and Tiilikainen use the conclusions drawn in the
introductory chapter to explain why constructivists are not satisfied with
more deterministic (associative-learning or information-processing) models of
strategy development. The latter do not adequately consider the construction of
conceptual knowledge or its integration with procedural knowledge. Baroody
and Tiilikainen note, for example, that even preschoolers have conceptual
under-standing of arithmetic, understanding that permits them to invent their
 Preface xvii

own procedures and adapt them when necessary—albeit not always successfully.
They further conclude that the stark contrast between preschoolers’ mathematical
competence and schoolchildren’s mathematical incompetence (found in the
United States) is not a paradox when analyzed in terms of Hatano’s (1988)
constructs of adaptive and routine expertise.
In chapter 4, Baroody, Wilkins, and Tiilikainen summarize an influential
model of how arithmetic thinking develops, namely Resnick’s (1992) view
that this development can be characterized by increasingly general or abstract
and, hence, flexible thinking. They then review how the existing research on
the development of a key arithmetic concept (namely, additive commutativity,
which can be symbolically represented as a+b=b+a) is consistent or inconsistent
with this model. Baroody et al. conclude by speculating on how knowledge
of addition and commutativity progresses from weak schemas (relatively
incomplete, context-bound, and unconnected representations) to strong
schemas (relatively complete, abstract, and well-connected generalizations)—a
development that embodies the construction of adaptive expertise and the
acquisition of greater flexibility.
In chapter 5, Seo and Ginsburg describe a long-term case study of a class
that explored a key arithmetic symbol, the equals sign. The children’s initial
interpretation of this symbol was limited and inflexible. Seo and Ginsburg
chronicle how context enables and constrains children’s understanding of the
equal sign. They document the persistence of children’s initial understanding of
this symbol and how textbooks and teaching practices can reinforce a limited
and inflexible view of it. Seo and Ginsburg also describe how teachers can use
meaningful and inquiry-based instruction, including cognitive conflict and
peer-peer dialogue, to prompt the construction of a broader and more flexible
understanding of the equals sign, the adaptive expertise essential for correctly
understanding, for instance, algebraic equations later. A key observation they
make—one echoed in the chapters by Miura and Okamoto (chapter 8) and Fuson
and Burghardt (chapter 10)—is that a crucial role for teachers is helping children
to make connections. Seo and Ginsburg also make the important observation
that typical testing procedures are well suited for measuring routine expertise but
not adaptive expertise. To gauge the latter, more flexible assessment methods,
such as the clinical interview, are needed.
In chapter 6, Butter worth, Marschesini, and Girelli present evidence that
fostering the mastery of basic multiplication combinations, a key goal of early
childhood mathematics instruction, is not simply a matter of providing sufficient
practice. Their data indicate that internalizing relational knowledge, namely
the commutative property of multiplication, can be an important component
in achieving such mastery and may even affect how the basic multiplication
combinations are represented in long-term memory. In other words, Butterworth
et al.’s results are inconsistent with the traditional view that mastering the basic
number combinations is merely a matter of developing rou-tine expertise and are
consistent with the view that learning even this basic arithmetic knowledge is a
matter of fostering adaptive expertise.
xviii Preface

In chapter 7, LeFevre, Smith-Chant, Hiscock, Daley, and Morris summarize

additional evidence contradicting the traditional view that mastery of the basic
multiplication combinations entails a single process, namely fact retrieval. Their
results indicate that even adults flexibly use a variety of strategies to efficiently
generate the 100 basic products. A key implication drawn from their results
is that researchers need to be more flexible and creative in measuring even
knowledge as basic as mastery of the single-digit number combinations.
In chapter 8, Miura and Okamoto discuss how a key sociocultural factor,
namely, language, can influence the construction of mathematical understanding
and its application. More specifically, because Asian counting-word sequences
are highly regular and clearly reflect the grouping-by-ten nature of multidigit
numbers and the English counting-word sequence is less so, Asian children can
rely more heavily on inducing counting patterns and rules than can English-
speaking children. This may give the former an edge in noticing the underlying
grouping-by-ten structure of the counting and written number sequence and
in constructing thinking (reasoning) strategies for basic addition combinations
with sums of more than 10 and their corresponding subtraction combinations.
This latter advantage may, in turn, given Asian children an advantage in mental
multidigit arithmetic. Put differently, learning to count in an Asian language
provides a better basis for acquiring adaptive expertise with numbers that can
then be applied to multidigit-number tasks. Miura and Okamoto also discuss how
fraction terminology and linguistic conventions can help or hinder mathematical
understanding and performance.
In chapter 9, Dowker reviews the evidence on the early development of single-
and multidigit computational estimation. She notes that children can use their
understanding of numbers and operations on numbers to produce reasonable
estimates to arithmetic problems for which they cannot yet calculate an answer.
Dowker emphasizes the following point that has often been overlooked by
scholars of arithmetic cognition: Children’s ability to estimate effectively
increases with computational experience and the resulting growth in general
arithmetic knowledge. She notes further that a child’s estimation competence
drops off gradually as increasingly larger and less familiar numbers are involved.
In other words, when children can draw on their number sense (a rich network
of concepts and facts), they flexibly choose appropriate estimation strategies.
When they cannot, children resort to using inflexible and ineffective strategies.
In chapters 10 and 11, Fuson and Burghardt and Ambrose, Baek, and Carpenter
adduce evidence that suggests critics of mathematics reform who claim most
people cannot discover or invent anything significant (such as the colum-nist
Thomas Sowell quoted earlier) underestimate children’s mathematical ability.
The authors of these chapters detail a variety of children’s invented procedures
for operating on multidigit written numbers, some of which are as, or more,
efficient than the standard algorithms taught in school. They note that discovery
and invention can be powerful tools in helping children to achieve genuine
mathematical proficiency (cf. Kilpatrick et al., 2001). Specifically, these processes
can promote further meaningful learning or conceptual understanding, which
 Preface xix

can provide a basis of additional adaptive expertise and computational fluency.

Discovery and invention involves children in mathematical inquiry (e.g., provides
practice in conjecturing, logical reasoning, and problem solving) and, thus, can
foster strategic mathematical thinking. Discovering or inventing knowledge for
one’s self is more likely to promote confidence and a positive disposition toward
learning and using mathematics than does being spoon-fed mathematical facts
and prescriptions. These benefits are reaped even if students “merely” rediscover
concepts or reinvent a standard procedure.
Furthermore, Fuson and Burghardt note that conceptual understanding is
essential for inventing effective strategies and rejecting ineffective ones. In a
similar vein, Ambrose et al. describe how not applying conceptual knowledge
can limit children’s invention of arithmetic strategies and how applying such
knowledge can lead them to devise flexible and creative procedures.
The last four chapters touch on the arithmetic learning and knowledge of
special populations and what research in this area tells us about the topic in
general. In chapter 12, Donlan reviews the evidence regarding the arithmetic
development of children with specific language impairments. Consistent with
Miura and Okamoto’s chapter, he concludes that such impairments can lead to
an impoverished mental representation of number and limit the construction of
informal arithmetic procedures. However, these deficiencies are neither inevitable
nor total. Indeed, certain aspects of mathematics are surprisingly unimpaired in
this population. In brief, the construction of arithmetical adaptive expertise can
be facilitated by language in some critical ways.
In chapter 13, Jordan, Hanich, and Uberti review research on children with
mathematics disabilities in three areas: mastery of basic number combinations,
ability to solve word problems, and skill at multidigit calculation. Learning
difficulties with these three areas can often be traced back to the failure of schools
to promote adaptive expertise. A particularly interesting aspect of this chapter
is their review of Huttenlocher and colleagues’ research on nonverbal addition
and subtraction among preschoolers, competencies that may underlie children’s
adaptive expertise with informal counting-basic arithmetic solution methods.
In chapter 14, Delazer makes the important point that neuropsychological
research has not paid sufficient attention to the role of conceptual knowledge.
Her summary and analysis of two influential models in the field and her review
and analysis of the interesting evidence on the mental-arithmetic performance
of brain-injured adults reinforce this point. Delazer’s review indicates that
there is a neuropsychological basis for distinguishing between routine and
adaptive expertise.
In chapter 15, Heavey reviews the fascinating evidence regarding arithmetical
savants and arrives at the surprising conclusion that their arithmetic proficiencies
are not the product of skills learned by rote but the result of a rich (extensive
and well-connected) number sense. She also makes the interesting observation
that the calculational prowess of arithmetical savants does not exceed those of
skilled nonsavant mental calculators and that the skill of both groups is shaped
by an early interest in numbers, counting, and mathematical analysis. Like
xx Preface

people in general, arithmetic savants use a variety of strategies, many of which

are self-invented.
The results of psychological, educational, and clinical studies using a wide
variety of arithmetic tasks and populations (including “normally” and atypically
developing children, noninjured and brain-injured adults, and savants), then,
all point to the conclusion that a key basis for the flexible and creative use of
arithmetic procedures (adaptive expertise or computational fluency) is conceptual
understanding, knowledge that connects a procedure to a rationale, everyday
experiences, analogies, and other skills and concepts. Bisanz and Geary, cognitive
psychologists whose work relies more heavily on informationprocessing theory
than constructivism, provide further comment and perspective on chapters 1 to
8 and 9 to 15, respectively. In doing so, they raise additional issues and questions
for future research.
This book should be of interest to those who are curious about the
nature of arithmetic expertise and those who are interested in promoting it
effectively, including:

• researchers whose focus is mathematical cognition or development such as

developmental or educational psychologists and neuropsychologists.
• researchers whose focus is mathematics instruction such as mathematics or
special educators.
• practitioners, administrators or others interested in improving arithmetic
instruction such as teachers, curriculum supervisors, principals, other
school administrators, and officials in national and local government
education departments.
• graduate students who are preparing for a career in research (in any of the
fields noted earlier), teaching, or administration.

—Arthur J.Baroody
University of Illinois at Urb ana-Champaign

Ann Dowker
Oxford University

References
Baroody, A.J., & Ginsburg, H.P. (1986). The relationship between initial meaningful and
mechanical knowledge of arithmetic. In J.Hiebert (Ed.), Conceptual and procedural
knowledge: The case of mathematics (pp. 75–112). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Carpenter, T.P. (1986). Conceptual knowledge as a foundation for procedural knowledge:
Implications from research on the initial learning of arithmetic. In J.Hiebert (Ed.), Conceptual
procedural knowledge: The case of mathematics (pp. 113–132). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Hatano, G. (1988). Social and motivational bases for mathematical understanding. In G.B. Saxe
& M.Gearhart (Eds.), Children’s mathematics (pp. 55–70). San Francisco: JosseyBass.
Jacobs, H.R. (1982). Mathematics: A human endeavor (2nd ed.). San Francisco: Freeman.
 Preface xxi

Kilpatrick, J., Swafford, J., & Findell, B. (Eds.). (2001). Adding it up: Helping children learn
mathematics. Washington, DC: National Academy Press.
National Council of Teachers of Mathematics. (2000). Principles and standards for school
mathematics. Reston, VA: Author.
Nussbaum, M.C. (1997). Cultivating humanity: A classical defense of reform in liberal education.
Cambridge, MA: Harvard University Press.
Resnick, L.B. (1992). From protoquantities to operators: Building mathematical competence
on a foundation on everyday knowledge. In G.Leinhardt, R.Putnam, & R.A.Hattrup (Eds.),
Analysis of arithmetic for mathematics teaching (pp. 373–425). Hillsdale. NJ: Lawrence
Erlbaum Associates.
Rittle-Johnson, B., & Siegler, R.S. (1998). The relation between conceptual and procedural
knowledge in learning mathematics: A review of the literature. In C.Donlan (Ed.),
The development of mathematical skills (pp. 75–110). Hove, East Sussex, England:
Psychology Press.
Sowell, T. (January 11, 2000). “Why insist on reinventing wheel?” The News-Gazette, p. A6.
Champaign-Urbana.
Wilson, S.M. (2003). California dreaming: Reforming mathematics education. New Haven:
Yale University Press.
 CHAPTER 1
THE DEVELOPMENT OF ADAPTIVE
EXPERTISE AND FLEXIBILITY: THE
INTEGRATION OF CONCEPTUAL AND
PROCEDURAL KNOWLEDGE
Arthur J.Baroody
University of Illinois at Urbana-Champaign

In Hard Times (Ford & Monod, 1966), Charles Dickens described a teacher by
the name of Mr. M’Choakumchild in order to satirize teachers trained to focus
on memorizing facts by rote at the expense of all else:

He…had taken the bloom off…of mathematics…. He went to work… looking at all
the vessels ranged before him…. When from thy boiling store [of facts] thou shalt
fill each [vessel] brimful by-and-by,…thou… kill outright… Fancy [imagination,
curiosity, creativity] lurking within—or sometimes only maim [it]. (p. 6)

Those interested in improving mathematics instruction have long been concerned

about the ineffective, if not harmful, way traditional instruction promotes
arithmetic expertise (e.g., Brownell, 1935; Davis, 1984; Erlwanger, 1975; Holt,
1964; Kilpatrick, 1985; Wertheimer, 1945/1959). Consider, for example, two
vignettes with dramatically different outcomes.

Vignette 1: A Case of Inflexible Mechanical Learning. Wertheimer (1945/1959)

told of a visit he took to a classroom that had just learned how to determine
the area of a parallelogram. The students had been taught to measure the length
of a parallelogram’s base and height and multiply these two values (see Fig. A).
After watching them successfully complete several “problems,” Wertheimer sought
permission from the teacher to ask the class a question. Proud of his students, the
teacher readily agreed. Wertheimer stepped to the board, drew a picture similar
to the one in Fig. B and asked what its area was. Some students were “obviously
taken aback. One pupil [noted]: ‘Teacher, we haven’t had that yet’” (p. 15). Other
students tried to apply the procedure they had been taught but quickly became
bewildered. In brief, confronted with a somewhat novel task, the students did not
see how their school-taught procedure applied, leaving them utterly helpless.

1
2 The Development of Arithmetic Concepts and Skills

Vignette 2: A Case of Flexibly Applied Learning. Arianne, a fourth grader, was

attempting to divide 901 by 2 in order to solve a problem. Her initial effort (shown
in Fig. C) resulted in an answer of 45 with a remainder of 1. Her teacher noticed that
she spontaneously revised her answer (shown in Fig. D). Asked why she had changed
her answer, Arianne explained that she had checked it. This process was described
cryptically in a journal entry the girl recorded later: I had ansewer of 45R1. I added it
together and didn’t get 901. So I added a zero and got 900 and added the 1.

Several months earlier Arianne’s class had learned that subtraction could be checked
by adding the answer (difference) to the subtracted amount (the subtrahend). If the
resulting sum was the same number as the starting amount (minuend), then the
answer was presumably correct. The teacher had used small numbers to illustrate
the rationale for the procedure (e.g., subtracting 2 from 5 can be undone by adding
2) before demonstrating this checking procedure with multidigit examples like that
shown as follows:
 The Development of Adaptive Expertise and Flexibility 3

Without further instruction, Arianne applied the principle for checking subtraction to
the task of checking her division. Instead of using the inverse operation of multiplication
(2×45=90), she checked her answer informally by using its mathematical equivalent,
namely repeated addition of a like term. (Adding 45+45 is what she meant by “I
added it together and didn’t get 901.”) After recognizing that 45 could not be the
correct solution and adding a 0 to make her answer 450, the girl proceeded to check
her new answer in the following manner:

To check whether Arianne really understood the checking procedure for division
she had invented, her teacher inquired, “What if we had divided by three?”
Arianne promptly indicated that she would have to add her answer three times.
“Or multiply by three,” interjected Alison who had been listening intently to
the dialogue.

Note that because these students understood the rationale for checking
subtraction, they recognized that, in principle, it could be applied to the related,
but different, task of checking a division outcome. Moreover, their understanding
allowed them the flexibility to adapt their learned procedure so that i t could be
used in this new context.
Vignette 1 illustrates what Giyoo Hatano (1988) called “routine expertise,”
knowledge memorized by rote, knowledge that can be used effectively with
familiar tasks but not with novel ones, even those that differ only slightly
from familiar ones. Vignette 2 illustrates what he called “adaptive expertise,”
meaningful knowledge that can be applied to unfamiliar tasks as well as familiar
ones. In contrast to rote knowledge, which is learned and stored in isolation,
meaningful knowledge is learned and stored in relation to other knowledge.
That is, it is grasped and represented in connection with other knowledge
(Ginsburg, 1977; Hiebert & Carpenter, 1992). If meaningful knowledge is
the basis of adaptive expertise (the flexible application of knowledge) and can
be characterized as well-connected knowledge (e.g., Ginsburg, 1977; Hiebert
& Carpenter, 1992; Hiebert & Lefevre, 1986), then the construction of
4 The Development of Arithmetic Concepts and Skills

well-connected knowledge should be the basis for fostering adaptive expertise

or flexibility, as measured by transfer (e.g., understanding new subject matter or
inventing a strategy or procedure to solve a novel problem). Analogously, the
acquisition of unconnected knowledge should be the basis for promoting routine
expertise or inflexibility.
This chapter begins with a historical sketch of the debate about the
developmental relations between conceptual and procedural knowledge.
Tentative theoretical conclusions about these relations are then outlined. Finally,
key implications of these conclusions for current reform efforts are discussed.
The aim of this chapter is to provide a general framework for understanding
the significance of the chapters that follow and for further explorations of the
development of arithmetic concepts, procedures, and adaptive expertise.

Historical Perspective
Researchers interested in the teaching and learning of arithmetic have
long distinguished between knowledge memorized by rote and knowledge
acquired meaningfully. The former has been equated with computational
skill or procedural knowledge; the latter, with understanding or conceptual
knowledge. In this section, the early and then the more recent view of this
dichotomy are examined.

EARLY VIEWS: SKILL VERSUS

UNDERSTANDING
Resnick and Ford (1981) pointed out that “the relationship between
computational skill and conceptual understanding is one of the oldest concerns
in the” field of mathematical psychology (p. 246) and that, in the past, the focus
was on which was better. In 1935, William A.Brownell outlined three views
of arithmetic learning, each of which reflected different assumptions about the
importance of computational skill and understanding. After briefly describing
these views, we summarize the early skills versus concepts debate.

Three Theories of Arithmetic Learning

In the early 20th century, the following three theories of arithmetic teaching and
learning were prominent in the United States: drill, meaning, and incidental-
learning theory (Brownell, 1935; cf. Cowan, chap. 2, this volume).
Drill Theory. According to drill theory (Model A in Fig. 1.1), a product of
associative theories of learning, instruction should focus on ensuring the (rote)
memorization of computational skills (see Cowan, chap. 2, this volume, for
a detailed discussion). The basic assumptions of this theory were (a) children
must learn to imitate the skills and knowledge of adults, (b) what is learned are
associations or bonds between otherwise unrelated stimuli, (c) understanding is
 Figure 1.1: The Psychological Roots of Various Theories of Arithmetic Learning

Empiricist
Empiricist . . . . Na tivist
Nativist
Tradition
Tradition Tradition

Theories of Learning
that
that focus on
focus On
observable behavior
observable behavior
factors)
(external factors)
(e><tl.'maJ
~ con$itWr
are called
menta! processes
are called (inter",,1 factors)
arecaUed

Behavioral Cognitive
Theories Theories

that 31'f! based

that are bued onon the
the
assumption
assumption that lnat the
the unit
unit of
of
learning
learning is Is an
an association
association
between
between stimuli Of a
stImuli or a stimulus
stimulus
and a response are call~
~ponse
___.,..____ are~
Associative- lnforma!ion- "Radical
Learning Processing COll5tructivisf'
Theories Theories Je.g., J. Piaget)

an examp!.t of
wh.lCh. uSed 10
expWn lhe
leamingof school
mathematICS was
A /Driii , c~
Meaning D /Schema-Based
Theory View (e.g., A. J.
Theory
(e.g., E. L. Baroody
(e.g., W .A.
Thorndike) &H.P.
Brownell)
~

5
6 The Development of Arithmetic Concepts and Skills

not necessary for the formation of such bonds, and (d) the most efficient way
to accomplish bond formation is through direct instruction and drill. In this
view, the learning of single-digit (basic) number combinations such as 5+3 =8
and multidigit renaming procedures such as “carrying” or “borrowing,” for
example, could be achieved quickly by a well-organized regimen of instruction
and practice (Thorndike, 1922). Self-invented procedures by children, such as
counting- or reasoning-based computational strategies (e.g., Arianne’s procedure
for checking division described earlier) were viewed as impediments (see, e.g.,
Smith, 1921; Wheeler, 1939).
Meaning Theory. Dissatisfaction with traditional mathematics instruction
and its theoretical basis (drill theory) led Brownell (1935) to propose meaning
theory (Model C in Fig. 1.1). This theory is akin to the social-learning theory
of Lev S.Vygotsky (1962) and foreshadowed Hatano’s (1988) ideal of adaptive
expertise. In Brownell’s view, instruction should focus on promoting the
meaningful memorization of skills:

The “meaning” theory conceives arithmetic as a closely knit system of

understandable ideas, principles, and processes. According to this theory, the
test of learning is not mere mechanical facility in “figuring.” The true test is an
intelligent grasp upon number relations and the ability to deal with arithmetical
situations with proper comprehension of their mathematical as well as their
practical significance, (p. 19)

According to meaning theory, teachers must take into three interrelated factors
in order to promote the learning of arithmetic with understanding:
• The complexity of arithmetic learning. In the traditional drill approach, for
instance, instruction on counting and numbers is followed by drill of “number
facts.” This approach “almost totally neglects the element of meaning and the
complexity of the first stage in arithmetic learning” (Brownell, 1935, p. 21).
According to meaning theory, children are, at first, expected to engage in
“immature” strategies, such as self-invented counting and reasoning strategies
(see, e.g., Baroody, 1987b, 1999). These informal methods provide the basis
for mature knowledge, including mastery of basic number combinations,
understanding of arithmetic principles, and meaningful memorization of
multidigit procedures (see also, e.g., Ambrose, Baek, & Carpenter, chap. 11,
this volume; Baroody, 1987a; Cowan, chap. 2, this volume; Fuson & Burghardt,
chap. 10, this volume).
Brownell (1935) noted that children may initially need to rely on countingbased
arithmetic strategies because such strategies may be their only meaningful means
of operating on numbers. As soon as they are ready, however, students should be
encouraged to use more advanced strategies, such as reasoning (e.g., strategies
transforming an unfamiliar problem into a familiar one: 7+5=[7+3]+[5–
3]=10+2=12). Eventually, children come “to a confident knowledge of [a number
combination], a knowledge full of meaning because of its frequent verification.
 The Development of Adaptive Expertise and Flexibility 7

By this time, the difficult stages of learning will long since have been passed,
and habituation occurs rapidly and easily” (p. 24). Drill may serve to increase
the facility and permanence of recall. In sum, “children attain [skill] ‘mastery’
only after a period during which they deal with [arithmetic] by procedures less
advanced (but to them more meaningful) than automatic responses” (Brownell,
1941, p. 96).
• Pace of instruction. In the drill approach, children are told or shown,
for instance, a “number fact” once or twice and expected to memorize it
quickly (within a week), if not almost immediately (a day or so). According
to meaning theory, children are not expected to imitate immediately the skill
or knowledge of adults. Learning arithmetic—including mastering the basic
number combinations and multidigit procedures—is viewed as a slow, protracted
process. In the meaning approach, then, time is allowed for children to construct
an understanding of arithmetic ideas and to discover and rediscover arithmetic
regularities before practice makes the basic number combinations automatic.
The result is a more secure knowledge of these combinations, knowledge that is
more easily transferred.
• Emphasis on relations. According to drill theory, for example, 6+5=11 and
7+4=11 should not be taught simultaneously because of “associative confusion” or
“associative interference.” Brownell (1935) argued that addition and subtraction
combinations such as 3+2=5 or 5–3=2 are not facts but generalizations and,
for teaching purposes, should be treated as such. To help children make such
generalizations, teachers need to help them discover a regularity such as 3+2=5
many times in many different situations. Furthermore, they should help students
see the relations among combinations so that they came to know the basic
combinations as a system of knowledge.
Incidental-Learning Theory. Incidental-learning theory (Model E in Fig. 1.1),
an analog of Jean Piaget’s (radical) constructivist theory, was embodied in John
Dewey’s early progressive-education movement. This theory was also a reaction
to drill theory. According to this view, children should be free to explore the world
around them, notice regularities (patterns and relations), and actively construct
their own understanding and procedures. In other words, mathematical learning
should be the incidental result of satisfying their natural curiosity.
Brownell (1935) concluded that the incidental-learning approach was
impractical for three reasons. The first is that “incidental learning, whether
through ‘units’ or through unrestricted experiences, is slow and time-consuming.
[A second is that the] arithmetic ability as may be developed in these [unfocused]
circumstances is apt to be fragmentary, superficial, and mechanical. [A third
is] that teachers generally lacked the expertise to implement such an approach
effectively” (pp. 17–18; cf. the more recent analyses by Fuson & Burghardt,
chap. 10, this volume; Howe, 1999; Ma, 1999; Seo & Ginsburg, chap. 5, this
volume). For these reasons, Brownell advocated direct instruction. Moreover, he
felt that interests were socially determined.
8 The Development of Arithmetic Concepts and Skills

Skills Versus Concepts Debate

Proponents of drill theory believed that mathematics instruction should focus
on promoting the mastery of basic skills, not on cultivating understanding of
mathematical concepts (e.g., Smith, 1921; Thorndike, 1922; Wheeler, 1939).
In Hatano’s (1988) terms, the aim of instruction was routine, not adaptive,
expertise. In contrast, proponents of incidental-learning theory believed the focus
of instruction should be fostering conceptual understanding, not on memorizing
basic skills.
In advocating meaning theory, Brownell (1935) took a middle course. His
meaning approach was a combination of a drill theory and incidental-learning
theory. On the one hand, he recommended the judicious use of textbook-based
instruction and drill to promote skill mastery. On the other hand, Brownell
recognized the value of building on children’s experience and discoveries in
order to foster conceptual understanding. Put differently, unlike many of
his contemporaries, he did not value computational skill over conceptual
understanding or vice versa. Brownell appears to have presaged the modern view
that both are important—that mathematical competence requires “knowing both
what to do and why” (Skemp, 1978, p. 9) and that instruction should focus on
promoting adaptive expertise (Hatano, 1988). Although Brownell seems to have
favored fostering conceptual knowledge first, he left unclear how it developed
and how this development was related to skill development. These issues were
taken up in the last two decades of the 20th century.

RECENT VIEWS: SKILL AND UNDERSTANDING

In the early 1980s, Resnick and Ford (1981) called for a cognitive psychology
of how computational skills and conceptual understandings of arithmetic are
acquired, including the transitions in competence. They further concluded that,
whereas past research efforts focused on which was better to promote, future
research efforts should focus on how understanding influences learning of
computational routine and how skilled computation, in turn, affects conceptual
learning. Next, I first outline two modern but different approaches to a cognitive
psychology of arithmetic development. I then summarize four perspectives
that have emerged from the recent debates about how the developments of
mathematical concepts and skills may be related.

Two Modern Views of Arithmetic Development

Geary (1994) noted that currently there are two general theoretical models of
arithmetic development: a strategy-choice model (e.g., Siegler, 1986; Model
B in Fig. 1.1) and schema-based models (e.g., Baroody & Ginsburg, 1986;
Steffe, 1992). Although “both approaches seek to specify the cognitive changes
associated with children’s arithmetical development” (Geary, 1994, p. 81), each
reflects its psychological roots.
 The Development of Adaptive Expertise and Flexibility 9

Strategy-Choice Model. The strategy-choice model developed by Siegler

and his colleagues reflects its roots in associative-learning theory by focusing
on children’s overt strategies and the practice and reinforcement of arithmetic
responses. This model also reflects its origins in information-processing theory
with its effort to delineate the internal cognitive mechanisms that govern a child’s
choice among alternative strategies. See chapters 2 (Cowan) and 3 (Baroody &
Tiilikainen) for a detailed discussion and analysis of this model.
A Schema-Based View. One schema-based view (e.g., Baroody & Ginsburg,
1986; Model D in Fig. 1.1) reflects the influence of (a) Piaget’s (e.g., 1964)
constructivist theory with its focus on the construction of abstract structures
(relational knowledge that summarizes information about many particular cases),
(b) Vygotsky’s (e.g., 1962) social-learning theory with its concern for learning
potential or readiness (how much children can benefit from intervention), and
(c) information-processing theory with its accounts of the limits of working
memory and the drive for cognitive economy.
In chapter 3, Baroody and Tiilikainen compare this schema-based view
with the strategy-choice model on three key issues: (a) How do counting-based
addition strategies and strategy choice develop? (b) How does mentalarithmetic
competence with basic and multidigit combinations evolve? and (c) How are
the developments of addition concepts and procedures related? In chapter 4,
Baroody, Wilkins, and Tiilikainen extend the discussion of the schema-based
view to the development of early arithmetic understanding in general and the
concept of additive commutativity in particular.
In chapter 5, Seo and Ginsburg discuss how school instruction can expand
a child’s conceptual schema for a key formal symbol, namely the equals sign.
In chapter 6, Butterworth, Marschesini, and Girelli discuss how conceptual
understanding of multplicative commutativity may impact the learning and
representation (schemas) of basic number combinations. In chapter 7, LeFevre,
Smith-Chant, Hiscock, Daley, and Morris extend the discussion of strategy
choice and mental arithmetic to provide a dramatically different view of adult’s
basic number-combination facility. In chapter 8, Miura and Okamoto explain
how language can help shape conceptual schemas of multidigit numbers and
fractions (cf. Donlan, chap. 12, this volume). In chapter 9, Dowker considers
how conceptual understanding can influence estimation efforts on familiar,
somewhat novel, and novel estimation tasks and describes key components of
this conceptual knowledge. In chapters 10 and 11, Fuson and Burghardt and
Ambrose et al. discuss critical elements of the conceptual schemas underlying
nonstandard or standard multidigit procedures and provide insight into the
development of these schemas.
In chapters 12 and 13, Donlan and Jordan, Hanich, and Uberti discuss the
nonverbal or precounting bases for constructing number and arithmetic schemas
and come to different conclusions about the roles of language and instruction
(social) factors that may underlie mathematical learning difficulties. In chapter
14, Delazer reflects on the inadequacies of information-processing models
that have dominated the field of neuropsychology and takes a step toward a
10 The Development of Arithmetic Concepts and Skills

schema-based view by explicitly considering the role of conceptual knowledge

in neuropsychological disorders. In chapter 15, Heavey makes the startling
conclusion that what underlies the calculational competencies of idiot savants is
an intricate web of interconnected number knowledge, not the recall of isolated
facts memorized by rote.

The Skills-First Versus Concepts-First Debate

and Its Aftermath
In the last quarter of the 20th century, the debate about whether skills or
concepts is more important was replaced by a debate about their developmental
order. Initially, two polar camps, namely the proponents of the skills-first view
and those of the concepts-first view, dominated this debate. In time, two more
moderate views were advanced. Summarized next—and discussed further in
chapter 3 (Baroody & Tiilikainen) of this volume—are four possible ways skills
and concepts can be related.
Possibility 1: Skills First. More behaviorally oriented researchers (e.g., Briars
& Siegler, 1984) have proposed that the development of mathematical skills
predates and underlies the development of concepts. In this view, arithmetic
skills are learned by rote memorization through imitation, practice, and
reinforcement. The result is that skill learning is piecemeal. Through applying
their skills, children discover arithmetic regularities or concepts. As conceptual
knowledge slowly accumulates, it increasingly links together procedural
knowledge. A parent or teacher might teach a child, for instance, to count-on
from the larger addend (COL; e.g., for 3+5: “5, 6, 7, 8”). Then, by using this
procedure to compute sums, the child might discover that both 3+5 and 5+3
have the same sum and that addend order does not affect the sum (the principle
of additive commutativity).
Possibility 2: Concepts First. Some more cognitively oriented researchers (e.g.,
Briars & Larkin, 1984) have proposed that conceptual knowledge precedes and
guides the construction of procedures. For example, Riley, Greeno, and Heller
(1983) hypothesized that an understanding of additive commutativity enables
children to invent COL. Indeed, more nativist-oriented cognitive psychologists
have argued that recent evidence suggests a version of the concepts-first view
called the “privileged-domain hypothesis” (e.g., Model F in Fig. 1.1). According
to this hypothesis (Geary, 1995; Gelman, 1993; Gelman & Meek, 1992; all
cited by Rittle-Johnson & Siegler, 1998; Wynn, 1998), some (mathematical)
competencies have been favored by our evolutionary history and, thus, are
innate or have an innate basis.
Possibility 3: Iterative Development. One result of a colloquium series at the
University of Delaware in the mid 1980s (titled Conceptual and Procedural
Knowledge: The Case of Mathematics) and its by-product (a book by the same
name, edited by James Hiebert, 1986) was an understanding that the developmental
relations between concepts and skills are not as straightforward or easily
characterized as skills-first or concepts-first proponents suggested. Baroody and
 The Development of Adaptive Expertise and Flexibility 11

Ginsburg (1986), for instance, proposed an iterative relation between concepts

and skills. In this moderate or intermediate perspective, conceptual knowledge
can lead to advances in procedural knowledge, the application of which can lead
to a conceptual advance, and so forth (see also Baroody, 1992; Fuson, 1988).
In this iterative view, children use their conceptual knowledge, for example, to
construct a concrete counting-all strategy (CCA; representing each addend with
objects and then counting all these objects). Computational experience with this
basic strategy can lead to insights that prompt the development of somewhat
more advanced strategies. Their use, in turn, can lead to the discovery of additive
commutativity, which can prompt the adoption of the advanced strategy COL
(e.g., Neches, 1987).
Possibility 4: Simultaneous Development. Another intermediate position
proposed by Rittle-Johnson and Siegler (1998) is that procedural and conceptual
knowledge develop concurrently. According to this simultaneous-development
model, the development of the COL strategy, for instance, is intertwined and
inseparable from the development of the additive-commutativity concept.

Tentative Theoretical Conclusions About the Relations Between

Conceptual and Procedural Knowledge
Rittle-Johnson and Siegler (1998) noted that existing evidence bears largely
on the skills-first and concepts-first views (Possibilities 1 and 2, respectively),
because of the difficulties of adequately testing the iterative and simultaneous
views (Possibilities 3 and 4, respectively). However, they further concluded, “An
iterative relation between procedural and conceptual mastery, seems an especially
plausible path of development” (p. 78). Next, I discuss three conclusions drawn by
Hiebert and Lefevre (1986) regarding the relations between skills and concepts—
conclusions that serve as premises for the iterative (or the simultaneous) view. I
then briefly summarize the case for Possibility 3 (and that against Possibilities 1
and 2) in the form of three key guidelines for constructing or evaluating models
of arithmetic development.

KEY CONCLUSIONS BY HIEBERT AND LEFEVRE (1986)

Hiebert and Lefevre (1986) defined procedural knowledge (skills) as knowing
how-to and conceptual knowledge (understanding or concepts) as knowing why.
Fig. 1.2 summarizes their detailed description of these aspects of mathematical
knowledge. On the surface, conceptual and procedural knowledge appear to be
clearly distinct. However, Hiebert and Lefevre made the following important
points about these two types of knowledge:

1. Whereas conceptual knowledge, by definition, involves interconnected and,

hence, meaningful knowledge, procedural knowledge may or may not be
connected to other knowledge and, thus, may be meaningful or not.
12
Figure 1.2: Summary of the Hiebert and Lefevre (1986) Characterizations of Procedural and Conceptual Knowledge

Knowledge

that involve> that ilwo!ves

know,",,""'" undom.tanding
to.i:il ca l!oo' iscalled

fl'Ol..'edural Conceptual
knowledge knowlf'dge
~Dfknowin3tL
consists of knowing the -:1 1;;-
oo~ of mowing S!'ql-by-s~ tNt is oor.strude<i by . Chat ~ OONotruct.ed by IlnkifIg
forms. 0(
forms m .. t.hantb<::s.
of mathematics, litJe.u (sequentiillJ o-r rueraich.ial e:Wting tut tsolared ~ new Infom\i!l~Ofl to existing
wnK ....
which f1:"<:'i'Itl..J
are called the th~ pn!Sl"riptioM ..-alIN 0' m.u..-loo~ is called I 'ledge is CilHed
I

Symbol PloductiOI'l )~_ _---, Tn:s ight, di.scaYery Meaningful Jearning

represtntatlon sp:l't:Tn:i leaming.. ur .integration or assimilation
'Y'''''" ---,..- ~
th3. is
that is that is
that i~ not
not
tied to a 1 - - - - 1 - - - - - - . . . . l
O""" j ) tied
tiftJlo
to aa
Prob l em~tvins
specific
sp«iilc
Rules AJsorithms mnl~t
context specific
specific
str"'egies. '------------1 context
CMtext
---- -- Qn
can Primary-le\'eI Reflec:tive-ll!vl!l
o~l1Ite
operate concepts concl'plo;;
••on {less abstr~(t) (more abstract)

or
0'
 The Development of Adaptive Expertise and Flexibility 13

2. The distinction between these two types of knowledge can, in some cases, be
fuzzy (see, e.g., Baroody & Ginsburg, 1986).
3. Linking conceptual and procedural knowledge can greatly benefit the
acquisition and application of the former, as well as the latter.

THREE ADDITIONAL CONCLUSIONS

In an effort to make sense of what research data are available on the topic, I
offer three additional conclusions about the relations between conceptual and
procedural knowledge.

1. Conceptual Knowledge Usually Underlies Procedural Innovations

Although children seem to learn many mathematical skills without fully
understanding their conceptual basis (e.g., Briars & Siegler, 1984; Fuson, 1988),
the skills-first view does not provide an adequate explanation for how they can
readily devise or adapt appropriate procedures for various mathematical tasks.
(For a detailed critique of this view see, e.g., Gelman and Meek, 1983, 1986,
1992.) Carpenter (1986) rejected the skills-first position:

It is inappropriate to conclude that conceptual knowledge of some kind is

unnecessary for procedural advances…. When invention occurs at a procedural level,
it often results in errors…. If children were simply attempting to increase cognitive
economy, they presumably would construct routines that were inappropriate more
frequently than they do. (p. 120)

In brief, although Carpenter allowed that rote memorization of computational

procedures can occur, he concluded that the learning of procedures, particularly
informal ones, has at least some conceptual basis or underpinning (in this volume,
see, e.g., Ambrose et al., chap. 11; Fuson & Burghardt, chap. 10).

2. Adaptive Expertise Involves the Integration of Conceptual

and Procedural Knowledge
Carpenter (1986) concluded, “There is ample evidence that the development
of conceptual knowledge alone does not drive the acquisition of procedures
for solving [arithmetic] problems (Baroody & Ginsburg, [1986]; Carpenter &
Moser, 1984)” (p. 119). He further noted that the concepts-first “views proposed
by Briars and Larkin (1984) and Riley et al. (1983) are overly parsimonious”
and that “children’s behavior is not as orderly as” they prescribe (p. 119). In
particular, such they do not account for systematic differences. For example,
although Riley et al. (1983) posited that the same conceptual knowledge
underlies children’s ability to solve missing-addend problems and counting-on,
evidence indicates that they can do the former before the latter (Carpenter &
Moser, 1984).
14 The Development of Arithmetic Concepts and Skills

Carpenter (1986) went on to offer a modified concepts-first view that

helps account for the variability in children’s performance not predicted by
conceptsfirst accounts:

Advances in children’s ability to solve addition and subtraction problems are based
on conceptual knowledge, but connections between conceptual and procedural
knowledge are established gradually and on a piecemeal basis. Conceptual
knowledge is not immediately integrated with all related procedures; rather initially,
it is applied locally to individual problems and procedures, (p. 120)

The development of arithmetic adaptive expertise, then, cannot merely be equated

with the construction of understanding. It also involves building links between
conceptual knowledge and procedural knowledge. Two key implications follow
from the hypothesis that the development of arithmetic adaptive expertise can be
characterized by the increasing integration of these two types of knowledge:
• Whereas the distinction between conceptual and procedural knowledge will
initially be fuzzy regardless of the nature of instruction, whether the distinction
becomes clear later depends heavily on the nature of instruction. Even with
rich instructional experiences, early conceptual development (assimilation and
integration) may proceed relatively slowly. Put differently, despite the good
intentions of parents, preschool teachers, and others, a child’s initial concepts
will have relatively few links. Thus, the understanding represented by such
“weak schemas” will be relatively isolated, superficial, and inflexible (Baroody
& Ginsburg, 1986; Baroody et al., chap. 4, this volume; Mix, Huttenlocher, &
Levine, 2002)—not altogether different from procedural knowledge that has few
connections to conceptual knowledge.
As development proceeds, the distinction between conceptual and procedural
knowledge should remain fuzzy if children learn procedures in a meaningful
manner. Otherwise, conceptual knowledge (what there is of it) and procedural
knowledge may well become distinct. From this perspective, the evidence that
indicates arithmetic knowledge has distinct components (in this volume, see,
e.g., Delazer, chap. 14; Donlan, chap. 12; Dowker, chap. 9; Heavey, chap. 15;
LeFevre et al., chap. 7) is—to some extent—a by-product of pervasive educational
practices that promote routine expertise. This conjecture is consistent with
Delazer’s (chap. 14, this volume) observation that braininjury reports include
cases of intact procedural knowledge and impaired conceptual knowledge but
not the reverse. The commonly found pattern of dissociation is what would
be expected of patients whose procedural knowledge was not integrated with
their conceptual knowledge (who have routine expertise of arithmetic skills).
Patients with adaptive expertise might be less prone to exhibit any type of
dissociation. Furthermore, key to gauging dissociation is assessing procedures
learned meaningful and the conceptual knowledge that directly underlies them,
not any procedural and conceptual knowledge.
 The Development of Adaptive Expertise and Flexibility 15

• The integration of conceptual and procedural knowledge should permit

greater flexibility in the invention and use of strategies or procedures. (See, in
particular, in this volume, Cowan, chap. 2; Baroody & Tiilikainen, chap. 3; Seo
& Ginsburg, chap. 5; Butterworth et al., chap. 6; Dowker, chap. 9; Ambrose et
al., chap. 11; Heavey, chap. 15.) Why might more connections yield not only a
more complete and deeper understanding but greater flexibility?
Consider the analogy of moving to a new town (Pólya & Szego, 1972).
Initially, a newcomer’s knowledge of her new hometown is rather incomplete and
unconnected. She may know how to get from her house to her workplace and
from her house to the grocery store. Unfortunately, if she is at work and needs to
go to the grocery store, her only option is to return home, a route that as Frame
A of Fig. 1.3 illustrates, can be relatively inefficient. As the person explores her
new hometown, she discovers landmarks and streets and can better see how
the all fit together. As her knowledge of the town becomes more complete and
interconnected, she can find her way around the town more easily. It allows
her, for example, to determine the most efficient route from her workplace to
the grocery store (see Frame B of Fig. 1.3). Moreover, if this customary path
is blocked, the resident’s well-connected knowledge gives her the flexibility to

Figure 1.3: The Relation Between the Amount of Connected Knowledge and the Degree of
Travel Flexibility

Frame A: Sketch of Unconnected Knowledge of New Hometown

Frame B: Sketch of More Connected Knowledge of a Familiar Hometown

16 The Development of Arithmetic Concepts and Skills

determine the next-best route. For instance, if the portion of West Street south
of Greenview Park is closed for repairs, the well-versed traveler knows that Alt
Avenue provides a good alternative route.

3. Conceptual Knowledge Can Play Either a Direct or an Indirect Role in the

Invention of Procedures
Like Carpenter’s (1986) modified concepts-first view, a schema-based iterative
account is grounded on the assumption that development evolves from localized
understandings or weak schemas into generalized concepts or strong schemas
(Baroody & Ginsburg, 1986; Baroody et al., chap. 4, this volume). Nevertheless,
because this account included nonconceptual factors as a mechanism for
prompting strategy development or choice, Carpenter rejected it for same reason
he dismissed the skills-first view.
However, if a distinction between an indirect basis (conceptual support) for a
procedural advance and a direct basis for (conceptual instigation of) a procedural
advance is made, then the modified concepts-first and the schemabased iterative
views can be reconciled. (For example, recognizing that each addend should
be represented only once may directly guide a child’s effort to invent CCA,
but it may only serve as a constraint in the invention of subsequent strategies
such as COL.) Unlike the skills-first view, the schema-based iterative account
is not based on the assumption that procedural innovations motivated by
nonconceptual factors occur in a conceptual vacuum. Consider, for example,
the argument that the invention of a strategy in which addend order is ignored
(e.g., COL) can be motivated by an effort to achieve cognitive economy and,
thus, does not require knowledge of commutativity (Baroody, 1987b; Baroody
& Gannon, 1984; Baroody & Ginsburg, 1986; Baroody et al., chap. 4, this
volume). Such an argument does not exclude an indirect conceptual basis
for this innovation. Clearly, children must have at least some conceptual
understanding of addition and that disregarding addend order is acceptable,
even if they do not necessarily recognize that commuted expressions such as
5+3 and 3+5 are equivalent.
Two conjectures about the aforementioned distinction seem plausible: (a) A
direct role may be the more important of the two. Put differently, conceptual
knowledge, rather than nonconceptual factors, may be the primary driving force
of self-invented procedures (see, e.g., Fuson & Burghardt, chap. 10, this volume;
Heavey, chap. 15, this volume), (b) Nonconceptually motivated (but conceptually
supported) procedural advances are probably more prone to be inconsistently
applied (generalized), inappropriate (illegal), and less stable (forgotten) than
are conceptually motivated ones (again see, e.g., Fuson & Burghardt, chap. 10,
this volume). Parenthetically, the former may be misinterpreted as evidence for
a conceptual advance (Baroody, 1985) or contribute to the appearance that
children use diverse, conceptually motivated strategies.
 The Development of Adaptive Expertise and Flexibility 17

Educational Implications for

Current Reform Efforts
The psychological debate about the relations between conceptual and procedural
has had an impact on the ongoing argument about how to improve mathematics
instruction (Ginsburg, Klein, & Starkey, 1998)—sometimes characterized as
the “math wars.” In order to better understand the math wars and to put it
in perspective, two frameworks may be helpful. One is a description of four
different instructional approaches. This framework builds on and updates
Brownell’s (1935) three models of teaching mathematics (discussed in the first
section of this chapter). A second is a set of new instructional goals that has
emerged from the math wars. In this section, I outline these two frameworks and
then use them and the earlier conclusions about the relations between conceptual
and procedural knowledge to analyze current efforts to reform mathematics
education in the United States.

A SOURCE OF DISAGREEMENT: DIFFERENT

VIEWS OF INSTRUCTION
In this subsection, I first outline different perspectives on mathematics instruction
and then discuss how these differences of opinion set off the most recent version
of the math wars.

Four Approaches to Mathematics Instruction

Baroody (with Coslick, 1998) identified the following four qualitatively different
approaches to teaching mathematics: the skills, conceptual, problemsolving,
and investigative approaches. The characteristics of these approaches and how
they contrast with each other are summarized in Table 1.1. In this subsection,
I describe the parallels between these approaches and Brownell’s (1935) three
approaches and outline their philosophical and psychological assumptions about
teaching and learning mathematics.
The Skills Approach. The skills approach, with its focus on the memorization
of basic skills by rote, is analogous to what Brownell (1935) called drill theory.
This approach is based on the assumption that mathematical knowledge is simply
a collection of socially useful information (facts, rules, formulas, and procedures).
Except for the mathematically gifted, children are viewed as uninformed (“empty
vessels”), largely helpless, and perhaps incapable of understanding many aspects
of school mathematics. The aim of mathematics instruction is to inform them
about how to do mathematics. Proponents of the skills approach (e.g., the
“back-to-basics” movement) believe that the most efficient way to transmit
information is by means of direct instruction and practice—without taking the
time to promote understanding about the whys of procedures. A teacher directs
all aspects of the learning process to ensure efficient inculcation (transmittal and
mastery) of basic skills.
18 The Development of Arithmetic Concepts and Skills

Table 1.1: Four Approaches to Mathematics Instruction (Baroody, with Coslick, 1998)

Instructional
Ins truction al Philosophical View
Philosophical View Teaching
Teadl.ing
Approach
Approach Style
S<y~
Name
Name Nature of
Nature of Knowledge
Knowledge View of
View of Aut
Authority
hority

Skills
Skins Dualism
Dualism Rightt or
Righ or wrong
wrong with
with Absolute external
Absolute external Completely
Approach
Approach no shades
no shades of of gray:
gray: authority: As
authority: As the
the authoritarian
There is
There is olle correct
one corr.....:t expert, the teacher is and
,.d
procedure or
procedure or answer.
answer. judge of
the judge of co
correct-
rrect· extremely
ness. Procedu
!leS$. Procedures or
res or teacher
teacher
answers that
answers that di
differ
ffer centered::
centered
from those advocated Direct
Direct
by the teacher are instruction
wrong and not (teaching by
tolerated. Teacher
tolerated. Teacher imposition).
imposition).
provides ddefinitive
efinitive
feedback
feedba ck (e.g., praise
(e.g., praise
for the
for the co
correct
rrect
answer).
Conceptual
Conceptual Pluralism
Pluralism Continuum from
Continuum Tolerant external
Tolerant external Semi-
Semi·
Approach
Approach right
ri to wrong:
gh t to There
wrong: lhere authority:
au Teacher
thority: Teacher authoritarian
is aa choice
is of possible
choiCE' of possible accepts diverse
accepts diverse pro-
pro- and teacher
but not
b\lt not equally
eq ually valid
va lid cedures and
cedures answers
and .l.nSWers centered:
procedures
p r()(edu~ o orr but strives for
but strives for Direct
Diled and
answers. Objectively,
answers. Objectively, perfection,
perfection, namely,
name ly, semi-direct
there
there isis one
one best learning
learning of of the
!he best
best instruction
possibility.
possibility. procedure
p~ure or answer. answer. (teaching by
Teacher provides "careful"
"ca reful"
feedback (e.g., praises
feedbact. (e.g., praises imposition).
imposition).
all
all ideas,
ideas, particularly
particularly
the
the conventional
conventional
one).
one}.
Investigative
Investigative Instrumen-
Instrumen- Many
Many right
right choices:
choices: Open
Open internal
internal author-
author- Semi-
Approach
Approach talism
talism There
There is
is aa choice
choice of
of ity:
ity: Teacher
Teacher or or democratic
possible
possible procedures
procedures or or student
student remains
remains com-
OOffi- and student
answers
answers and
and often
often mitted
mitled toto aa method
method oror centered:
centered:
many
many are
are good.
good. viewpoint as long as itII Semi-indirect
is
is effective. Teacher
Teache r instruction
instruction
responds
responds to incorrect
incorrect (guided
procedures
proredures or or participatory
partiCI patory
answers
answe~ by by posing
posing a democracy).
democracy).
question, problems,
problems, oror
task
task that
that prompts
student
student reflection.
refl oction.
Problem-
PlobJem- Extreme
Extreme No
No right
right oror wrong:
wrong: No
No external
external authority: Completely
Completely
Solving
Solvil"lg Relativism
Relativism There are
are many
many Teacher and
and each democratic
democratic
Approach
Approach possible,
puSbib1e, equally
equally valid
valid student define his
his or
or and
,nd
possibilities.
poso;ibilities. her
her own truth.
truth. extremely
Children evaluate student
student
their
their own centered:
centered:
conclusions. Indirect
Indirect
instruction
instruction
(teaching
(teaching by
by
negotiation).
negotiation ).
 The Development of Adaptive Expertise and Flexibility 19

Table 1.1 continued

Aim of
Aim of Instruction
Instruction Focus of
Focus of Instruction
Instruction Teacher's/Students' Roles
Teacher'sIStudents' Roles

Skills
Skill; Foster routine
Foster routine expertise:
expertise: Procedural content
Procedural content Teacher serves
Teacher serves as
as aa
Approach
Approach the memorization
the memorization of of (e.g., how
(e.g., how to add
to add director: an
director: an information
information
basic skills
basic skills (arithmetic
(arithmetic multi-digit
multi-digit dispenser (informer)
dispenser (informer) and
and
and geometric
and geometric facts,
facts, numbers).
numbers). taskmaster (manager).
taskmaster (manager).
definitions, rules,
definitions, rules, Because children
Because children are
are
formulas, and
formulas, and viewed as
viewed as uniformed
uniformed andand
procedures) by
procedures) by rote.
rote. helpless, students
helpless, students must
must be
be
spoon fed
spoon fed knowledge.
knowledge.

Conceptual
Conceptual Foster adaptive
Foster adaptive exper-
exper- Procedural and
Procedural and con-
con- Teacher serves
Teacher as aa shep-
serves as shep-
Approach
Approach tise: the
tise, the meaningful
meaningful ceptual content
ceptual content herd: information
herd: information dis-
dis-
memorization
memorization of of facts,
facts, (e.g.,.. wily
(e.g why you
you penser
penser (informer) and
(informer) and
rules, formulas, and
rules, formulas, and "carry" when
"carry" when add-
add- up-front guide (conduc-
up-front guide (conduc-
procedures.
procedures. ing multi-digit
ing multi-digit num-
num- tor). Because
tor). Because children
children are
are
hers).
bers). seen as
seen as capable
capable ofof under-
under-
standing mathematics
standing mathematics if if
helped, they
helped, are engaged
they are engaged
in quaSi-independent
in quasi-independent
activities and
activities and discussions.
discussions.

Investigative
Investigative Foster all
Foster all aspects
aspects of
of Procedural content,
Procedural content, Teacher serves
Teacher serves asas aa
Approach
Approach mathematical power:
mathematical power: conceptual content,
conceptual content, mentor: activity
mentor: activity organizer
organizer
positive disposition
positive disposition (e.g"
(e.g., and the
and the processes
processes of
of (instigator) and
(instigator) and guide-on-
guide-on-
interest, confidence
interest, confidence andand mathematical
mathematical the-side (moderator).
the-side (moderator).
constructive beliefs
constructive beliefs inquiry (problem
inquiry (problem Because children have
Because children have
about learning
about learning and
and using
using solving, reasoning,
solving, reasoning, informal knowledge
informal knowledge and and
mathematics), adaptive
mathematics). adaptive conjecturing,
conjecturing, an inherent
an inherent need
need to to
expertise
expertise representing, and
representing, and understand, they
understand, are
they are
(understanding of
(understanding of communicating).
communicating). capable of
capable of inventing
inventing their
their
concepts and
concepts and skills),
skills), and
and own solutions
own solutions and
and making
making
mathematical thinking
mathematical thinking (at least
(at some) sense
least some) sense ofof
(the capacity
(the capacity to conduct
to conduct mathematical situations
mathematical situations
mathematical inquiry
mathematical inquiry themselves (i.e.,
themselves (i.e., students
students
including problem
induding problem are engaged
are engaged in in semi-
semi-
solving and
solving and reasoning).
reasoning). independent activities
independent activities
and discussions).
and discussions).

Problem-
Problem- Foster mathematical
Foster mathematical Processes of
Processes of Teacher serves
Teacher serves as
as aa
Solving
Solving thinking-the ability to
thinking-the ability to mathematical
mathematical partner: participant,
partner: participant,
Approach
Approach conduct mathematical
conduct mathematical inquiry: problem
inquiry: problem monitor, and
monitor, and devil's
devil's
inquiry.
inquiry. solving, reasoning,
oolving, reasoning, advocate. Students
advocate. Students
conjecturing
conjecturing engage in
engage in relatively
relatively
representing, and
representing. and independent activities
independent activities
communicating.
communicating. and discussions,
and discussions.
20 The Development of Arithmetic Concepts and Skills

Table 1.1 continued

-
Organizing Principle Methods
Methods

Skills (logically):
Bottom-up (logically): •• Teacher lectures and demonstrates
Approach Sequential instruction •• and largely symbolic
Textbook based and
most basic skills to
from Il106t •• Children work in isolation
most complex skills such •• Practice with an emphasis on written, sterile
on written, sterile
solving.
as problem solving. worksheets
• or technology
Little or no use of manipulatives or technology

Conceptual Bottom-up
Bottom-up •• Didactic instruction supplemented by guided
by guided
Approach (psychologically):
(psychologically): discovery learning
Sequential instruction
Sequential instruction •• Textbook based,
Textbook based, but uses, e.g.,
teacher uses,
but teacher e.g"
on the
based on
based readiness of
the readiness of meaningful analogies and concrete models to
concrete models to
students to
students construct
to construct explain procedures
understanding. •• Whole-class, small-group, and individual
instruction
instruction
•• Children imitate manipulative procedure

Investigative (guided):
Top-down (guided):
Top-down •• Various methods with an emphasis on on
Approach Teacher usually poses a indirect techniques that involve students students inin
"worthwhile task" {one (one exploring, conjecturing about, and
exploring.. debating
and debating
that is challenging
that is and
challenging and ideas (e.g., semi-guided discovery learning)learning)
complex) as
complex) of
way of
as way •• Projects, situations,
everyday situations,
problems, everyday
Projects, problems,
exploring, learning and
exploring..leaming and science
science experiments, children's literature,
experiments, children's literature,
practicing basic concepts
practicing basic concepts math games,
math games, and forth create
so forth
and so for
need for
create aa need
and skills;
and teacher may
skills; teacher may learning and
learning and practicing textbooks serve
math; textbooks
practicing math; serve
take advantage of
take advantage of (e.g., aa source
role (e.g.,
supporting role
aa supporting of worthwhile
source of worthwhile
teachable moments (e.g., tasks and resolving
tasks and disagreements over
resolving disagreements over
question or
question posed
problem posed
or problem definitions)
student).
by student).
by •• Children often work together in groups
in groups
•• encouraged to
Students encouraged
Students to invent, and
share, and
invent, share,
streamline their
streamline own concrete
their own and,
models and,
concrete models
(including the
later, written procedures (including the
conventional one
conventional or equally
one or equally or or more efficient
more efficient
nonconventional ones)
nOn<:onventionalones)
•• Practice purposefully
done purposefully
Practice done
•• Use of technology is a key aim and and central to
central to
many learning tasks
many learning tasks

Problem-
Problem- (unguided):
Top-down (unguided): •• Open-ended or
Open-ended discovery
unstructured discovery
or unstructured
Solving Class tackles
Class of
problems of
tackles problems learning
learning
Approach
Approach their choosing,
own choosing..
their own •• Content done incidentally
instruction done
Content instruction as
incidentally as
whether or
whether students
not students
or not needed; or no
little or
needed; little no use textbooks
of textbooks
use of
have received formal •• Students encouraged to invent, share and
share and
instruction the
on the
instruction on streamline their own concrete models and,
models and,
involved.
content involved.
content later, written procedures
later, written procedures
 The Development of Adaptive Expertise and Flexibility 21

In the skills approach, a teacher simply tells children, for instance, that to
multiply fractions, the numerator and denominator of the answer is determined
by multiplying the two factors’ numerators and denominators, respectively.
Students then complete numerous computations with the procedure until it is
memorized by rote. In brief, because instruction and practice is done without
context (a reason) at a largely symbolic (abstract) level, the skills approach is
not purposeful (in the sense that instruction builds on students’ interests and
creates a genuine need to learn and practice mathematics), nor is it typically
meaningful. As children seldom are engaged in any real mathematical thinking,
the skills approach is almost never inquiry based. (See Jordan et al., chap.
13, this volume, for a discussion of some of the negative consequences of
such an approach.) A modern prototype of the skills approach is the Saxon
curriculum, which involves well-organized and highly repetitive practice (cf.
Thorndike, 1922).
The Conceptual Approach. The conceptual approach, with its focus on the
meaningful memorization of skills, is analogous to what Brownell (1935) called
meaning theory. This approach is based on the assumption that mathematics is a
network of skills and concepts. Children are viewed as capable of understanding
mathematics if told or showed why procedures work. The aim of mathematics
instruction is to help them learn needed facts, rules, formulas, and procedures
in a meaningful way (i.e., with comprehension). A teacher shepherds children
toward understanding and mastery of skills.
In the conceptual approach (as typically implemented), symbolic procedures,
such as multiplying fractions, are illustrated by a concrete model in the form
of a textbook picture or an actual teacher demonstration. Children may even
be encouraged to imitate an illustrated model themselves with manipulatives.
Thus, although instruction and practice is often without context or purpose,
an effort is made to promote meaningful learning. Sometimes students may
also be involved in highly guided discovery-learning activities, which provides
some experience with the processes of mathematical inquiry. Examples of
such an approach are the following textbook series: Glencoe/McGraw Hill’s
Mathematics: Applications and Connections (1999) and Scott Foresman’s
California Mathematics (2001).
The Problem-Solving Approach. The problem-solving approach, with its focus
on the development of mathematical thinking (reasoning and problem solving),
is analogous to what Brownell (1935) called the incidental-learning approach.
Philosophically, this radical approach is at the opposite end of the spectrum
from the skills approach. (This is why it is placed at the bottom of Table 1.1.)
This approach is based on the assumption that mathematics is, at heart, a way of
thinking, a process of inquiry, or a search for patterns in order to solve problems.
Children are viewed, on the one hand, as using immature thinking and possessing
incomplete knowledge and, on the other hand, as naturally curious creatures
who can and must actively construct their own understanding of mathematics.
The aim of mathematics instruction is to immerse mathematical novices in
mathematical inquiry (solving what are to them real and challenging problems),
22 The Development of Arithmetic Concepts and Skills

so that they can develop more mature ways of thinking and incidentally discover
and construct more complete mathematical knowledge. A teacher participates
in the inquiry as a relatively wise partner who pushes the process along but does
not entirely, or even largely, set the agenda or control the inquiry.
In a problem-solving approach, learning content, such as the formal procedure
for multiplying fractions, is secondary to developing children’s thinking processes.
The open-school movement (e.g., Silberman, 1973) or Piagetian curricula (e.g.,
Furth & Wachs, 1974; Neill, 1960) in the recent past embodied the ideals of
such an approach.
The Investigative Approach. The investigative approach, with its focus
on meaningful memorization of skills and the development of mathematical
thinking, can be thought of as a composite of Brownell’s (1935) meaning
and incidental-learning theories (i.e., as the union of the conceptual and
problemsolving approaches). Like the conceptual approach, mathematics
is viewed as a network of skills and concepts. It is also, like the problem-
solving approach, viewed as a process of inquiry. Unlike the conceptual
approach and similar to the problem-solving approach, children are viewed
as capable of actively constructing this understanding. Unlike the problem-
solving approach, children’s active construction of understanding is mediated,
guided, and prompted by the teacher—most often through planned activities,
but occasionally by taking advantage of unplanned “learning moments.”
Like the conceptual approach, an aim of mathematics instruction is to help
students learn needed facts, rules, formulas, and procedures in a meaningful
fashion. Like the problem-solving approach, students are regularly engaged in
mathematical inquiry (to develop problem-solving, reasoning, representing, and
communicating competencies). In the investigative approach, a teacher mentors
children, guiding their meaningful construction of procedures and concepts
and the development of mathematical thinking. Thus, unlike either the skills
or the conceptual approach and like the problem-solving approach, a teacher
indirectly incites doubt, curiosity, or cognitive conflict by posing worthwhile
tasks and by creating a social environment that encourages questioning, inquiry,
and reflection (see e.g., Ambrose et al., chap. 11, this volume; Baroody, with
Coslick, 1998; Hatano, 1988; Miura & Okamoto, chap. 8, this volume; Seo &
Ginsburg, chap. 5, this volume).
In the investigative approach, an instructor uses indirect means to help students
construct knowledge. For example, a teacher might guide children to reinvent a
procedure, such as the algorithm for multiplying fractions. The first step might
be finding or inventing a worthwhile task that creates a genuine need for using
the procedure (e.g., adjusting a recipe with fractional measures by fractional
amounts, determining the area of a wall space or plot that involves a fractional
length and width, or determining the theoretical probability of two independent
events in an effort to solve a problem). The teacher might then encourage children
to invent their own procedures for solving the problem, methods that may well
involve using manipulatives or drawings. Students can then be encouraged to
represent the problem and their informally determined solution symbolically and
 The Development of Adaptive Expertise and Flexibility 23

to look for shortcuts to their concrete procedures. Typically, students should

quickly reinvent the formal procedure for multiplying fractions (cf. Fuson &
Burghardt, chap. 10, this volume; see, e.g., Baroody, with Coslick, 1998, for
similar examples with a variety of content). In brief, the investigative approach
involves purposeful, meaningful, and inquiry-based instruction. A reasonable
approximation of this approach is the Everyday Mathematics program
(Everyday Learning Corporation, 2001). In this elementary-level curriculum, a
teacher encourages self-invention of problem solutions by building on children’s
informal knowledge.

A Reform Effort and a Firestorm of Criticism

In the United States, a number of efforts to reform mathematics instruction

were undertaken during the 20th century (see, e.g., Kilpatrick, 1992, 2001;
Kline, 1974; Wilson, 2003). Nearly all of them had little lasting impact on the
textbooks and the curricula used in the vast majority of classrooms. (See Cowan,
chap. 2, this volume, for a review of reform movements in the United Kingdom.)
Ginsburg et al. (1998), for example, concluded that, although Brownell’s (1935)
meaning theory was embraced by mathematics educators at the university level,
Thorndike’s (1922) associationist view, by and large, won the hearts and minds
of practitioners. Outlined next is a brief review of a recent and ongoing effort to
reform mathematics instruction and the criticisms it engendered.
The NCTM Standards. In the last two decades of the 20th century,
dissatisfaction with the traditional skills approach led the National Council of
Teachers of Mathematics (NCTM) to publish a series of standards documents
(1989, 1991, 1995, 2000) laying out a new vision of mathematics instruction.
Proposed were changes in both the content of school mathematics and, perhaps
more important, the way mathematics is taught. Regarding the former, these
standards outlined ambitious goals for mathematics instruction, including
conceptual understanding and problem-solving and reasoning competence.
Regarding methods of instruction, the proposed reforms include making
mathematics instruction relevant to children (purposeful), conceptually based
(meaningful), and process based (inquiry based). Thus, although the NCTM
recommended using a variety of teaching technique, the general approach
recommended is most similar to what was described earlier as the “investigative
approach.” Like previous reform movements, this proposed approach, which
is based on (moderate) constructivist and social-learning principles, has been
criticized as impractical.
Criticisms of the NCTM Standards. Critics of the NCTM standards often
misinterpreted its 1989 document as advocating the problem-solving approach,
which is based on a philosophy of radical constructivism. Some common
criticisms stemming from this misconception and counterarguments based on
the three conclusions regarding the relations between conceptual and procedural
knowledge noted earlier are discussed next.
24 The Development of Arithmetic Concepts and Skills

1. Conceptual knowledge usually underlies procedural innovations. Geary

(1994), for example, noted that a strict constructivist approach to mathematics
education focuses almost exclusively on the acquisition of conceptual knowledge
and problem solving—at the expense of skill mastery. Such an extreme
position is “based, in part, on a failure to realize that mathematics requires
the acquisition of both procedural skills and conceptual knowledge” (p. 273).
Because constructivist-based instruction shortchanges skill mastery, it “is not
likely to lead to substantial long-term improvements in the mathematical skills
of… children” (p. 266).
Although Geary’s (1994) criticism may be true of the problem-solving
approach advocated by some strict constructivists, it is not true of the
investigative approach advocated by moderate constructivists. As outlined in
Hiebert (1986) and in this chapter, moderate constructivism is clearly based
on recognition that instruction should develop both conceptual and procedural
knowledge. In addition to the development of conceptual understanding and
the ability to engage in mathematical inquiry (e.g., problem solving), a key aim
of the investigative approach is skill mastery. Indeed, by cultivating adaptive
expertise, this purposeful and meaningful approach fosters skill mastery far
more effectively than the skills approach, which focuses on promoting routine
expertise (see, e.g., the evidence adduced by Carpenter, Fennema, Peterson,
Chiang, & Loef, 1989; Hiebert & Carpenter, 1992; Rittle-Johnson & Alibali,
1999). In brief, making concept development a central focus of instruction is not
incompatible with skill development; indeed, it can be the most effective way to
promote skill mastery.
2. Adaptive expertise involves the integration of conceptual and procedural
knowledge. Geary (1994) suggested that strict constructivists overlook the clear
distinction between procedural knowledge and conceptual knowledge and
concluded, “that different teaching techniques are needed for children to acquire
procedural and conceptual competencies” (p. 273). “Except for basic numerical
and arithmetical skills, most children are not likely to be able to develop
mathematical procedures solely on the basis of their conceptual knowledge.
Instead, some procedures and equations simply need to taught and practiced”
(pp. 265–266).
As recent analyses suggest, there is not a serious conflict between developing
procedural knowledge and fostering conceptual knowledge (Howe, 1999; Ma,
1999). As noted earlier in this chapter, flexible and adaptive use of procedures
requires the integration of procedural knowledge with conceptual knowledge.
Put differently, if the aim of instruction is the meaningful learning of skills, the
distinction between procedural and conceptual knowledge becomes blurry and
less useful. With rare exceptions, there is no real need to teach elementarylevel
procedures and concepts in a different manner, nor is it desirable. Using the
investigative approach, even more advanced arithmetic procedures taught in
grades 1–8 can be invented by children, including those for operating on whole
numbers (Ambrose et al., chap. 11, this volume; Fuson & Burghardt, chap. 10,
this volume), rational numbers (fractions and decimals), or variables (algebraic
 The Development of Adaptive Expertise and Flexibility 25

symbols). See Baroody with Coslick (1998) for examples of, and instructional
guidelines for, each.1 In short, when instruction focuses on adaptive expertise
(the integration of procedural and conceptual knowledge), it does not make
sense to use different instructional techniques for fostering skill mastery and
concept learning.
3. Conceptual knowledge can play either a direct or an indirect role in
the invention of procedures. Geary (1994) argued that strict constructivists
overlook the importance of drill and practice. He pointed out that conceptual
knowledge is not sufficient for the development of computational and problem-
solving skills and that drill and practice are necessary for the efficient use of
such skills. Citing the research by Briars and Siegler (1984), Geary further
concluded that mechanistic learning of skills can enable children to induce
basic regularities or concepts.
Moderate constructivists do not disagree that practice has an important role
in mastering computational and problem-solving skills. They would, however,
argue that purposeful practice after meaningful learning is far more likely
to be effective than uninteresting drill of meaningless facts and procedures.
Furthermore, as noted earlier, children probably do not learn even basic
counting, number, and arithmetic competencies, as Briars and Siegler’s (1984)
skills-first model suggests, but do so in an iterative fashion in which procedural
and conceptual knowledge become increasingly intertwined (Baroody, 1992;
Baroody & Ginsburg, 1986; Rittle-Johnson & Siegler, 1998). Thus, a focus of
practice should be on discovering relations and integrating different aspects of
knowledge (fostering adaptive expertise), not merely internalizing isolated facts
and procedures (promoting routine expertise). For example, the practice of single-
and multidigit number combinations, should focus on discovering relations and
automatizing schemas such as the number-after rule for n+1/1+n combinations.
In brief, practicing procedures that are not conceptually instigated or at least
conceptually supported (promoting routine expertise) makes little sense.

1
Geary (1994) further argued that constructivist “fail to distinguish between biologically primary and
biologically secondary cognitive skills” and, thus, naively propose the same instructional techniques
for learning both (p. 264). That is, whereas simple skills (counting, number, and some features of
arithmetic) are innately supported and can be easily discovered by children or learned in appropriate
everyday social contexts, more complex skills (e.g., solving algebraic word problems) require very
different learning conditions. In fact, although the latter require more “groundwork” to ensure
readiness, children can and should be taught more complex mathematics in a manner similar to how
basic mathematics can and should be taught. For example, students should first be encouraged to use
what they know to devise their own informal methods for solving more complex problems involving,
for example, rational numbers (e.g., Mack, 1990, 1993; Post, Cramer, Behr, Lesh, & Harel, 1993;
Streefland, 1993), ratios and proportions (e.g., Lamon, 1993; van de Brink & Streefland, 1979), or
algebra (NCTM, 1992). They can then be encouraged to find shortcuts for concrete models or other
informal methods, a process that can lead them to reinvent formal procedures and formulas (see,
e.g, Baroody with Coslick, 1998, for details). Indeed, the quadratic equation is the only formula
Commonly taught in grades K–8 children are unlikely to re-invent and need to memorize by rote.
26 The Development of Arithmetic Concepts and Skills

A BASIS FOR AGREEMENT: NEW INSTRUCTIONAL GOALS

Although some critics of the NCTM standards promoted a “back-to-basics”
movement—a return to the skills approach—others searched for a middle ground.
The National Research Council (NRC) commissioned a committee of 16, which
included proponents of the NCTM (1989, 1991, 1992, 1995, 2000) standards
and its critics, to search for this middle ground. The resulting NRC report, Adding
It Up: Helping Children Learn Mathematics (Kilpatrick, Swafford, & Findell,
2001), echoed Brownell’s (1935) goal of meaningful mastery of arithmetic
knowledge and Hatano’s (1988) ideal of adaptive expertise by specifying that
the goal of mathematics instruction should be mathematical proficiency. Next, I
first describe mathematical proficiency and then discuss the implications of this
goal for current reform efforts.

Mathematical Proficiency
Four Aspects of Mathematical Proficiency. As conceptualized earlier by
Schoenfeld (1985, 1992), mathematical proficiency involves conceptual
understanding, computational fluency, strategic mathematical thinking, and a
productive disposition. The characterizations of the four strands of mathematical
proficiency in the following are quoted or based on the definitions appearing in
Kilpatrick et al. (2001, p. 5):

• “Conceptual understanding [entails] comprehension of mathematical

concepts, operations, and relations.”
• Computational fluency involves skill in computing efficiently (quickly and
accurately), appropriately, and flexibly.
• Strategic mathematical thinking requires the “ability to formulate, represent,
and solve mathematical problems” (“strategic competence”) and “the
capacity for logical thought, reflection, explanation, and justification”
(“adaptive reasoning”).
• Productive disposition [entails] a habitual inclination to see mathematics
as sensible, useful, and worthwhile, coupled with a belief in diligence and
one’s efficacy.”

The Need for Fostering the Four Strands in an Intertwined Manner. Rejecting the
skills-first and the concepts-first views, Kilpatrick et al. (2001) further concluded
that instruction should foster these four strands of mathematical proficiency
in a simultaneous or an intertwined manner. Next, I briefly explain why their
development must be done in an integrated (a simultaneous or, perhaps, an
iterative) manner.
Conceptual understanding is necessary for computational fluency and
strategic mathematical thinking and important to a productive disposition.
Note that the definition of computational proficiency implies adaptive, rather
than mere routine, expertise. Although computational efficiently can be achieved
without understanding, conceptual understanding can facilitate knowing when
 The Development of Adaptive Expertise and Flexibility 27

it is appropriate to apply a skill and is critical to modifying its application on

new tasks or problems.
Conceptual understanding is also necessary in thoughtfully formulating,
representing, and solving problems. It can also support adaptive reasoning.
For instance, it is easier to reason logically about familiar or comprehensible
situations than unfamiliar or incomprehensible ones. It is also much easier to
justify in a convincing manner a procedure one understands than one does not.
Finally, other than for the self-deluded, actually understanding mathematics
is normally needed to view it as sensible. Furthermore, children who understand
mathematics and view it as basically sensible are more likely to keep working at
a problem or the topic and to develop self-confidence than those who do not.

IMPLICATIONS FOR ONGOING REFORM EFFORTS

There is wide, though by no means unanimous, agreement that a traditional,
content-oriented skills approach does not adequately foster any aspect of
mathematical proficiency, including computational fluency. There is also wide
agreement that the same is true of a process-oriented, problem-solving approach,
particularly in regard to computational fluency. However, there remains a
deep division between those who support the NCTM (2000) standards (the
content and process-oriented investigative approach) and those who support
an alternative approach to reform (the content and somewhat process-oriented
conceptual approach). Next, I assess the latter’s and then the former’s chances
of achieving all aspects of mathematical proficiency.

The California Standards

The instructional goals and approach outlined in the California Public
Schools: Kindergarten Through Grade Twelve (Curriculum Development and
Supplemental Materials Commission, 1999), California Standards for short, was
developed, in part, as a reaction to the perceived shortcomings of the NCTM
(1989, 1991, 1995) standards. For instance, some critics have claimed that
instruction based on the NCTM (1989, 1991, 1995) standards is “fuzzy math.”
It is not clear, though, why encouraging students to understand mathematics,
use this understanding to invent solution procedures for challenging problems,
justify and defend their procedures and solutions, and critically analyze others’
procedures and solutions is fuzzy math. In any case, proponents of the California
Standards argued for a more direct and rigorous approach.
A Moderate, Conceptual Approach. The goals laid out in chapter 1 of the
California Standards are clearly consistent with those of the goal of achieving
mathematical proficiency (conceptual understanding, computational proficiency,
strategic mathematical thinking, and a productive disposition). The three-
phase instructional model recommended in chapter 4 to achieve these goals
appears most akin to the conceptual approach (meaningful learning of skills
and concepts). The mathematics content standards and test examples found in
28 The Development of Arithmetic Concepts and Skills

chapter 2 are somewhat consistent with the conceptual approach. Specifically,

these standards focus on skills with only some references to understanding them
and inquiry competencies.
What’s Wrong with a Conceptual Approach? A conceptual approach is
based on the questionable assumption that an understanding of mathematics
can be imposed on children. Constructivists have cautioned that merely viewing
or even imitating a concrete model will not automatically or effectively help
most students to understand an analogous written procedure (e.g., Baroody,
1989; Baroody, with Coslick, 1998; Clements & McMillen, 1996; Fuson &
Burghardt, chap. 10, this volume; Miura & Okamoto, chap. 8, this volume; Seo
& Ginsburg, chap. 5, this volume). Efforts to impose understanding on students
are frequently unsuccessful because the instruction does not build on children’s
existing knowledge and, thus, does not promote connected knowledge either
by assimilation or integration. As Resnick’s (1982) study illustrates, children
frequently do not spontaneously see how the steps in a concrete model parallel
those in a written procedure (Fuson & Burghardt, chap. 10, this volume).
Unfortunately, concrete models illustrated in textbooks or by teachers often do
not highlight these links (cf. Fuson & Burghardt, chap. 10, this volume). To
make matters worse, concrete models for written procedures are often not related
explicitly or clearly to meaningful analogies (Baroody, with Coslick, 1998).
For example manipulative-based models for fraction multiplication are only
loosely linked, if linked at all, to familiar situations such as finding the area of a
rectangle. Thus, they may not understand even the concrete manipulativebased
model and may simply memorize it by rote, as they would an incomprehensible
written procedure (e.g., Baroody, 1989). All too often, then, efforts to impose
concrete models on students do not enhance either conceptual understanding or
the thoughtful and flexible use of procedures (i.e., computational proficiency).
Moreover, a conceptual approach does little in the way of socializing children
in the ways of mathematical inquiry—that is developing process competencies
such as problem solving, reasoning, representing, and communicating. Little
or no effort is made to encourage students to use conceptual understanding to
propose their own informal models, look for patterns, and devise short cuts
that lead to efficient symbolic procedures (cf. Kieren, 1992). Though students
may engage in some inductive reasoning and conjecture proposing and testing
if involved in guided discovery learning activities, such experiences are often so
highly structured they do not promote debate and the need to justify procedures
or solution. Finally, problem solving in the conceptual approach usually and
merely involves applying a teacher-imposed procedure to similar problems
examined in class.

The Need for the Investigative Approach

A carefully implemented investigative approach may be the best way to help
all students achieve mathematical proficiency. This moderate approach is more
likely than other approaches, including the conceptual approach, to foster
 The Development of Adaptive Expertise and Flexibility 29

genuine conceptual understanding. As this understanding underlies the other

aspects of mathematical proficiency, the investigative approach is more likely
than other approaches to promote computational fluency, strategic mathematical
thinking, and a productive disposition. In terms of the focus of this book, the
investigative approach may be the best means for fostering integrated knowledge
of procedures and concepts or adaptive expertise.

CONCLUSION
Mathematics instruction cannot be significantly improved, as Brownell (1935)
argued, by turning away from more complex methods of instruction because
teachers are not sufficiently prepared to implement them. In order to develop
truly professional teachers who are capable of implementing the investigative
approach, pre- and in-service teacher training needs to foster adaptive expertise
with mathematics, mathematical psychology, and effective pedagogy (Ball, 1998;
Baroody, in press; Baroody, with Coslick, 1998; Campbell & White, 1997;
Even & Lappan, 1994). One reason Chinese schools are relatively effective in
fostering mathematical achievement is that their teachers have a relatively deep
understanding of the mathematics they teach (Ma, 1999). As Lee S.Shulman
observed in the foreword of Ma’s (1999) book, although Chinese teachers have
studied far less mathematics than U.S. teachers, “What they know they know
more profoundly, more flexibly, more adaptively” (p. xi). In essence, despite
taking fewer mathematics courses in school, Chinese teachers are afforded
the opportunity to construct adaptive expertise for the mathematics that they
teach. A profound understanding of how children’s mathematical learning and
thinking develops (e.g., knowing both the supporting and motivating concepts
of a procedure) is also necessary to effectively plan a sequence of worthwhile
tasks, create cognitive conflict, and otherwise promote the meaningful learning
(assimilation and integration) of concepts and procedures (e.g., Baroody, 1987a;
Baroody, with Coslick, 1998; Ginsburg, 1977; Ginsburg et al., 1998). A broad
understanding of teaching practices is needed to thoughtfully and flexibly devise
or adapt worthwhile tasks, orchestrate classroom discourse, and otherwise
create the interesting, meaningful, thought-provoking instruction that promotes
adaptive expertise and mathematical proficiency (e.g., NCTM, 1991). In brief,
unlike some calls for reform that focus exclusively on improving teachers’
mathematical (content) knowledge or attracting those with mathematical
expertise, implementing the reforms envisioned by the NCTM will require
teachers who have a powerful mathematical, psychological, and pedagogical
framework. Only with adaptive expertise in all of these areas will teachers
be highly effective at fostering the mathematical proficiency of their students,
including their mathematical adaptive expertise.

Acknowledgments
Preparation of this chapter was supported, in part, by a grant (“Early Arithmetic
Development”) from the University of Illinois Research Board, a Faculty