58-Reid & Knipping (2010)

Proof in Mathematics Education
David A. Reid and Christine Knipping - 978-94-6091-246-7

Downloaded from Brill.com11/09/2020 12:11:29PM
via free access
via free access
Proof in Mathematics Education
Research, Learning and Teaching
David A. Reid
Christine Knipping
Acadia University, Wolfville, Canada
SENSE PUBLISHERS
ROTTERDAM/BOSTON/TAIPEI

via free access
A C.I.P. record for this book is available from the Library of Congress.
ISBN: 978-94-6091-244-3 (paperback)

ISBN: 978-94-6091-245-0 (hardback)
ISBN: 978-94-6091-246-7 (e-book)
Published by: Sense Publishers,

P.O. Box 21858,
3001 AW Rotterdam,
The Netherlands
http://www.sensepublishers.com
Printed on acid-free paper
All Rights Reserved © 2010 Sense Publishers
No part of this work may be reproduced, stored in a retrieval system, or transmitted in any
form or by any means, electronic, mechanical, photocopying, microfilming, recording or
otherwise, without written permission from the Publisher, with the exception of any material
supplied specifically for the purpose of being entered and executed on a computer system,
for exclusive use by the purchaser of the work.

via free access
TABLE OF CONTENTS
Introduction........................................................................................................... xiii
Part I: What is Proof?
1. History of Proof ................................................................................................... 3

The Standard View ............................................................................................ 3
Other Views of the History of Proof................................................................ 10
Summary ......................................................................................................... 24
2. Usages of “Proof ” and “Proving” ..................................................................... 25

Everyday Usages ............................................................................................. 25
Scientific Usages ............................................................................................. 26
Mathematical Usages ...................................................................................... 26
Usages in Mathematics Education Research ................................................... 27
“Demonstration” and “Proof ” in Other Languages......................................... 32
Summary ......................................................................................................... 33
3. Researcher Perspectives..................................................................................... 35
Philosophies of Mathematics........................................................................... 37
Research Based on an a Priorist Philosophy of Mathematics......................... 39
Research Based on an Infallibilist Philosophy of Mathematics....................... 40
Research Based on a Quasi-Empiricist Philosophy of Mathematics ............... 46
Research Based on a Social-Constructivist Philosophy of Mathematics......... 48
Summary ......................................................................................................... 52
Balacheff’s Epistemologies of Proof ............................................................... 53
Diverse or Comprehensive Perspectives?........................................................ 54
Part 2: Important Research Foci, Past and Present
4. Empirical Results.............................................................................................. 59
Key Studies...................................................................................................... 59
Many Students Accept Examples as Verification ............................................ 59
Many Students Do Not Accept Deductive Proofs as Verification ................... 62
Many Students Do Not Accept Counterexamples as Refutation ..................... 63
Students Accept Flawed Deductive Proofs as Verification.............................. 64
Students’ Criteria for the Acceptance of Arguments ....................................... 65
Students Offer Empirical Arguments to Verify................................................ 67
Most Students Cannot Write Correct Proofs.................................................... 68
Ideas for Research ........................................................................................... 70
5. The Role of Proof ............................................................................................. 73

The Roles of Proof in Mathematics ................................................................. 73

via free access
TABLE OF CONTENTS
What is Proving in School Mathematics? ...................................................... 79

The Roles of Proof for Students..................................................................... 80
Possible Roles of Proof in Teaching .............................................................. 81
Ideas for Research.......................................................................................... 82
6. Types of Reasoning ......................................................................................... 83

Deductive Reasoning ..................................................................................... 84
Inductive Reasoning ...................................................................................... 88
Abductive Reasoning..................................................................................... 99
Reasoning by Analogy ..................................................................................110
Other Kinds of Reasoning ........................................................................... 123
Summary...................................................................................................... 126
Ideas for research ......................................................................................... 127
7. Classifying Proofs and Arguments ................................................................ 129

Proofs and Arguments Described According to the Representations
Involved ....................................................................................................... 130
Other Classifications of Proofs and Arguments ........................................... 142
Ideas for Research........................................................................................ 151
8. Argumentation............................................................................................... 153
Argumentation Versus Proof........................................................................ 155
Argumentation in Accord with Proof........................................................... 158
Argumentation According to Krummheuer ................................................. 161
Summary...................................................................................................... 163
Argumentation in Japan ............................................................................... 164
9. Teaching Experiments ................................................................................... 165

Fawcett......................................................................................................... 165
The Debate Approach .................................................................................. 169
Expecting Explanations ............................................................................... 172
Italy.............................................................................................................. 173
Summary...................................................................................................... 175
Part 3: Processes of Reasoning and Argumentation
10. Argumentation Structures.............................................................................. 179

Toulmin’s Functional Model and Argumentation Structures ....................... 179
The Source-Structure ................................................................................... 180
The Reservoir-Structure............................................................................... 185
The Spiral-Structure..................................................................................... 187
The Gathering-Structure .............................................................................. 189
Ideas for Research ....................................................................................... 191
vi

via free access
TABLE OF CONTENTS
11. Patterns of Reasoning .................................................................................... 193

Deduce-Conjecture-Test cycle ..................................................................... 194
Proof Analysis.............................................................................................. 198
Scientific Verification .................................................................................. 201
Surrender...................................................................................................... 202
Exception and Monster Barring ................................................................... 204
Summary...................................................................................................... 207
Part 4: Conclusions
12. Implications for Teaching...............................................................................211

What is Proof and what is it for? ..................................................................211
Formality...................................................................................................... 212
Results from New Math and Two-Column Proof Teaching ......................... 215
Starting where Students and Teachers are.................................................... 217
Teaching Experiments.................................................................................. 218
Summary...................................................................................................... 219
13. Directions for Research ................................................................................. 221

Teaching Proof............................................................................................. 221
Students’ Understandings of Proof............................................................... 223
Conceptional Issues ..................................................................................... 224
Conclusion ................................................................................................... 225
References............................................................................................................ 227
Author Index ........................................................................................................ 241
Subject Index........................................................................................................ 245
vii

via free access
via free access
LIST OF FIGURES
Figure 1: A misleading diagram showing that 8×8 = 5×13 ................................. 14

Figure 2: Word usage in three ESM papers ......................................................... 29
Figure 3: Three dimensions of description for researcher perspectives............... 36
Figure 4: Diagram from Fischbein, 1982, p. 18................................................... 40
Figure 5: Diagram that formed the basis for the triangle angle sum proof
in Fawcett, 1938 ................................................................................... 42
Figure 6: Reid’s PRISM model............................................................................ 58
Figure 7: The Count the Squares problem ........................................................... 86
Figure 8: Drawing used by Bill and John when proving that the sum of two
odd numbers is even............................................................................. 86
Figure 9: Will’s Count the Squares pattern.......................................................... 94
Figure 10: Diagram used in French classroom for Pythagorean Theorem
proof................................................................................................... 104
Figure 11: Handshake diagram for six people .................................................... 105
Figure 12: Sofia’s diagram .................................................................................. 106
Figure 13: The relationship between analogy, generalisation and
specialisation.......................................................................................113
Figure 14: The Arithmagon problem (as posed by Reid, 1995b)..........................115
Figure 15: Triangles showing geometric properties drawn by Wayne
while exploring ...................................................................................116
Figure 16: A tetrahedron with equal masses at its vertices from Hanna and
Jahnke, 2002b..................................................................................... 122
Figure 17: Toulmin model ................................................................................... 180
Figure 18: Overall argumentation structure in the proving process in
Mr. Lüders’ class ............................................................................... 182
Figure 19: Proof diagram from Lüders’ class ...................................................... 182
Figure 20: Argumentation stream AS-4 from Lüders’ class ................................ 183
Figure 21: Proof diagram from Nissen’s class..................................................... 184
Figure 22: The source-structure in Nissen’s class ............................................... 185
Figure 23: The reservoir-structure in Pascal’s class............................................. 186
Figure 24: Diagram used in Pascal’s class for Pythagorean Theorem
proof................................................................................................... 186
Figure 25: Overall argumentation structure in Dupont’s class............................. 187
Figure 26: Argumentation structure in James’ class for the rhombus
proving process .................................................................................. 188
Figure 27: Argumentation structure in James’ class for the side-side-side
proving process .................................................................................. 190
Figure 28: The Deduce-Conjecture-Test cycle .................................................... 194
Figure 29: The first Arithmagon puzzle............................................................... 194
Figure 30: Sandy’s first guess: 5.......................................................................... 195
Figure 31: Sandy’s solution to the first puzzle..................................................... 195
ix

via free access
LIST OF FIGURES
Figure 32: Sandy’s symbols for the unknowns and givens in a general
Arithmagon ........................................................................................ 196
Figure 33: Sandy’s puzzle.................................................................................... 199
Figure 34: Sandy’s reasoning, including Proof Analysis ..................................... 201
Figure 35: Scientific Verification......................................................................... 202
Figure 36: Surrender............................................................................................ 202
Figure 37: Exception and Monster Barring ......................................................... 205

via free access
LIST OF PROOFS
Proof 1: Elements Book I Proposition 1............................................................... 5

Proof 2: Figure reconstructed on the basis of Liu Hui’s commentary on
Jiuzhang Suanshu ................................................................................ 12
Proof 3: Elements Book I Proposition 4............................................................. 17
Proof 4: Elements Book IX Proposition 20 ........................................................ 19
Proof 5: Proof that the sum of the interior angles of a triangle is
180 degrees .......................................................................................... 44
Proof 6: Textbook proof from Chazan, 1993, p. 365.......................................... 96
Proof 7: Empirical argument from Chazan, 1993, p. 366................................... 97
Proof 8: The sum of the first n integers (by MI) ................................................ 99
Proof 9: The perpendicular bisectors of a triangle meet in a point....................119
Proof 10: The medians meet in a single point, physics proof ........................... 122
Proof 11: The Goldbach conjecture ................................................................... 131
Proof 12: Product of negatives .......................................................................... 132
Proof 13: Try it with 15...................................................................................... 132
Proof 14: The diagonals of a rhombus are congruent ........................................ 133
Proof 15: There are 14 possibilities and all fit ................................................... 134
Proof 16: Not all prime numbers are odd........................................................... 134
Proof 17: Numeric Gauss proof ......................................................................... 134
Proof 18: Divisibility by Nine............................................................................ 135
Proof 19: Action proof of the commutativity of multiplication ......................... 136
Proof 20: Behold!............................................................................................... 137
Proof 21: Schorle proof...................................................................................... 138
Proof 22: Two-column proof for the triangle angle sum.................................... 139
Proof 23: Symbolic Gauss proof ........................................................................ 140
Proof 24: Infinitude of primes............................................................................ 140
Proof 25: The product of two diagonal matrices is diagonal.............................. 140
Proof 26: Proof of an algebraic identity ............................................................. 141
Proof 27: A formal proof.................................................................................... 142
Proof 28: A transformational proof .................................................................... 148
xi

via free access
LIST OF TABLES
Table 1: References in Euclid’s proof of Prop. I.1 ............................................... 6

Table 2: Chronology of people and publications mentioned in the text ............... 8
Table 3: Summary of research perspectives ....................................................... 53
Table 4: Balacheff’s epistemologies of proof..................................................... 54
Table 5: Key studies reviewed ........................................................................... 60
Table 6: Results showing that many students accept examples as
verification. .......................................................................................... 61
Table 7: Summary of results showing that many students and teachers
do not accept deductive proofs as verification ..................................... 63
Table 8: Summary of results showing students accept flawed deductive
proofs as verification............................................................................ 65
Table 9: Factors influencing acceptance of arguments....................................... 66
Table 10: Frequencies of ratings for the familiar and unfamiliar
deductive verifications ......................................................................... 66
Table 11: Students’ use of empirical arguments................................................... 68
Table 12: Students able to write correct proofs .................................................... 69
Table 13: Students who were able to construct a valid proof
for TIMSS item K18 ............................................................................ 69
Table 14: Results according to the type and truth value of the statements........... 70
Table 15: Reasoning dichotomies based on certainty .......................................... 90
Table 16: Types of abductive reasoning used in examples................................. 103
Table 17: Bill’s analogy ......................................................................................112
Table 18: Ben’s analogy......................................................................................117
Table 19: Some links between triangles and tetrahedra ..................................... 120
Table 20: Kinds of reasoning ............................................................................. 126
Table 21: Kinds of reasoning and roles.............................................................. 127
Table 22: Comparison of Balacheff and the preformalists’ classifications of
arguments ........................................................................................... 129
Table 23: Classification of arguments ................................................................ 131
Table 24: Overview of categories ...................................................................... 143
Table 25: Comparison of proof classification systems....................................... 144
Table 26: Harel and Sowder’s proof schemes ................................................... 149
Table 27: Notions of argumentation................................................................... 163
xii

via free access
INTRODUCTION
Research on teaching and learning proof and proving has expanded in recent
decades. This reflects the growth of mathematics education research in general, but
also an increased emphasis on proof in mathematics education. This development
is a welcome one for those interested in the topic, but also poses a challenge,
especially to teachers and new scholars. It has become more and more difficult to
get an overview of the field and to identify the key concepts used in research on
proof and proving.
When we met, Christine was working on her doctoral dissertation (Knipping,
2003b). She commented on the difficulty she had making sense of the existing
research. David understood this feeling having had the same struggle when
working on his doctoral dissertation (Reid, 1995b). In the interval the amount of
research to be read and understood had increased, but the relationship between the
work of different researchers was no more apparent. Key terms were used differ-
ently by different authors, disparate theoretical assumptions were made, phenomena
were classified in incompatible ways, all without comment. We wished that
some synthesis of the literature existed that would explain the discrepancies and
make the links we found missing. And having achieved a better understanding
ourselves of the literature as a result of our efforts, we wondered if we could
attempt such a synthesis, perhaps in a journal article. In our discussions it soon
became clear that a longer piece of writing would be needed, and this book is the
outcome.
This book is intended to help teachers, researchers and students to overcome the
difficulty of getting an overview of research on proof and proving. It reviews the
key findings and concepts in research on proof and proving, and embeds them in a
contextual frame that allows the reader to make sense of the sometimes contradictory
statements found in the literature.
The first part provides this frame. It begins with an outline of the history of
proof in mathematics, both as it is usually presented and as it is interpreted by
scholars who take a wider view. Then the various uses of the words “proof ” and
“proving” in everyday life, science, mathematics and mathematics education are
described and compared. Finally, the various perspectives taken by researchers in
the field are outlined and placed into a structure that allows for comparison.
The second part reviews current research. First, basic findings from empirical
research are summarised. Then important theoretical constructs and classification
systems are discussed in several chapters organised around the themes of the role
of proof, reasoning, types of proof, and argumentation. Finally, several teaching
experiments are described.
The third part focusses on two larger frameworks for examining proving and
argumentation. The first is argumentation processes which are social processes that
occur in classrooms (and elsewhere) through which knowledge changes status.
A method of describing and analysing argumentation processes is outlined which
xiii

via free access
INTRODUCTION
reveals differences in the role of argumentation in different contexts. The second is

reasoning processes, which are internal psychological processes which link together
different ways of reasoning about mathematics. Distinctive patterns in reasoning
processes are described and related to the goals of teaching proof.
The final part includes concluding comments, first on the implications research
on proof has for teaching, and then on questions that require further research.
Throughout there is an emphasis on exploring the multiple perspectives different
researchers bring to the study of proof in mathematics education. These perspect-
ives are seen as being not only inevitable in a field where international attention is
brought to problems that often have significant local elements, but also enriching
as a diversity of perspectives offers opportunities to make sense of phenomena that
might be seen in a limited way from a single perspective. Hence, we do not attempt
a combining of perspectives, as Harel and Fuller (2009) have done. While it can be
confusing to encounter multiple perspectives, we believe it is also very valuable.
We hope this book will decrease the confusion and increase the rewards of its
readers in their further explorations of proof in mathematics education.
xiv

via free access
PART I
WHAT IS PROOF?
Reading any research literature can be a challenge at first because most authors
make assumptions about what the reader already knows about the field. This is
necessary as there is never space to include all the background underlying a
publication. In the research literature on teaching and learning proof, assumptions
are often made about the historical context of proof in mathematics, the meanings
of words like “proof ” and “proving” and about the theoretical perspective of the
author, which is often assumed to be shared by the reader. In Part 1 we consider
these three sets of assumptions and provide a guide to what assumptions might be
made by authors in the field. Unfortunately, but perhaps necessarily, there is not a
single uniform set of assumptions all researchers on proof share. Hence, we will
describe a range of possibilities, without being able to state definitively what
assumptions a given publication is based on. Given an outline of the possibilities,
however, a reader should be able to pick up on the clues in a publication and
identify the assumptions being made.
Chapter 1 concerns the history of mathematics, and presents an outline of the
“standard” history of proof in mathematics, familiarity with which is often assumed
when proof is discussed. We also present several alternatives to key elements in the
standard history, that some authors refer to in their work.
Chapter 2 discusses the uses of the words “proof ” and “proving” in mathematics,
mathematics education, logic, science and everyday life. Authors sometimes write
from more than one of these perspectives, which means that their terminology can
shift meaning from one paragraph to the next. Being aware of the possible contexts
and the meanings for “proof ” and “proving” associated with them will help readers
find their way through this shifting terrain.
Chapter 3 explores the theoretical perspectives of researchers on proof in mathe-
matics education. From within a given perspective, it seems a natural way of seeing
things, and so authors often do not comment on the perspective they take. However,
for communication with the larger community some awareness of these perspectives
is necessary, and for a reader new to the field understanding that different pers-
pectives exists will aid in making sense of what sometimes seem to be contradictory
statements.

via free access
via free access
CHAPTER 1
HISTORY OF PROOF
Before embarking on a discussion of proof and proving in mathematics education,

a look back at proof and proving in the history of mathematics is in order. This will
provide the necessary background to some of the issues we will discuss in later
chapters, and introduce some important concepts related to the nature of proof, and
the acceptance of proofs.
A student of modern mathematics might be confused at this point, as her or his
experience of mathematical proof in school might have suggested that mathematics
grows by an accumulation of knowledge, so although there are now new theorems
that have been proven since the time of the Greeks, the word “proof ” in the context
of mathematics has meant the same thing since the time of Euclid, at least. However,
as Wilder (1981) points out: “‘proof ’ in mathematics is a culturally determined,
relative matter. What constitutes proof for one generation, fails to meet the standards
of the next or some later generation” (p. 40). By exploring this variation we can
discover other ways of perceiving proof, and other ways of proving.
As you read this chapter you may want to reflect on this question:
– Does the history of proof in mathematics have direct implications for the
teaching and learning of proof? If so, how?
THE STANDARD VIEW
When one reads a history of mathematics (e.g., Anglin, 1994; R. Jones, 1997; Kleiner,
1991; Kline, 1962), one is likely to encounter a version of the history of proof we
call the “standard” view. When the history of mathematics is mentioned in research
on teaching and learning proof, it is usually the standard view that is assumed, and
so it has had significant impacts on proof teaching and research. In this section we
will summarise the standard view. In the next section we will introduce some
critiques and alternatives.
The First Proofs

According to the standard view, proofs originated with the Greeks, specifically
with Thales (c. 600 BCE). Prior to that time mathematics was done without proofs.
A number of theorems are associated with Thales, not because he discovered them,
but because he proved them:
All these theorems were known to the Egyptians and Mesopotamians. The
reason they are associated with Thales is that he was the first person to offer
proofs for them. This was an essential difference between pre-Greek and Greek

via free access
CHAPTER 1
mathematics: the Greeks established the logical connections among their results,
deducing the theorems from a small set of starting assumptions or axioms.
(Anglin, 1994, p. 14)
A number of authors have speculated on the reason the Greeks began to insist on
proving mathematical statements. Some (e.g., Hannaford, 1998, p. 181; Kleiner,
1991, p. 293) have claimed that the democratic nature of Athenian society created a
context in which logical argument was valued. Others have noted that the existence
of a leisure class meant that there were individuals who had time for philosophical
and mathematical activity without any immediate practical application (e.g., Kline,
1962, p. 45). Kleiner (1991, p. 293) and Arsac (2007, p. 31) also mention the
problem of the incommensurability of the side and diagonal of a square, and
Kleiner adds the need to teach mathematics, as motivations for an emphasis on
proof. Hanna and Barbeau (2002) see the motivation for proving as arising from
the nature of the entities studied in mathematics:
For the early Egyptians, Babylonians, and Chinese, the weight of observational
evidence was enough to justify mathematical statements. But classical Greek
mathematicians found this way of determining mathematical truth or false-
hood less than satisfactory. They saw that, unlike other sciences, mathematics
often deals with entities that are infinite in extent or number, such as the set
of all natural numbers, or are abstractions, such as triangles or circles. When
dealing with such entities, mathematics needs to make absolute statements,
that is, statements that apply to every instance without exception. (p. 36)
Whatever the reason, the origin of mathematical proofs is credited to the Greeks,
whose innovation then spread to other cultures.
The Legacy of Logic

The standard view tells us that rather than basing mathematics on observation and
experiment, the Greeks based it on logic. Not only did they make use of logical
arguments, they reflected on their reasoning and their methods. Plato (427–349
BCE) argued for a restriction on the tools that could be used in geometry, and
Aristotle (384–322 BCE) formulated the methods appropriate to mathematics:
In the Posterior Analytics, Aristotle formulates what we call the deductive
method. It was adopted by Euclid and has always been an essential charac-
teristic of mathematics. This method consists of starting with propositions
called axioms and then proving propositions called theorems. Each statement
in the proof has to be justified either by an axiom or by a previously proved
theorem or by a principle of logic. (Anglin, 1994, p. 63)
For example, consider Proof 1, which is the first proof in Heath’s (1956) translation
of Euclid’s Elements (c. 300 BCE). The text consists of several parts: The proposition
to be proven (line 2), a description of what has to be proven given a specific case
(the segment AB in lines 3–4), a construction (lines 5–12), a proof that the object
constructed is what it is meant to be (lines 13–20), and finally a statement asserting
that what has been done is what was required (lines 21–23).

via free access
HISTORY OF PROOF
Proof 1: Elements Book I Proposition 1
Proposition 1.
On a given finite straight line to construct an equilateral triangle.
Let AB be the given finite straight line. Thus it is required to construct an equilateral
triangle on the straight line AB.
5 With centre A and distance AB let the

circle BCD be described; [Post. 3]
again, with centre B and distance BA
let the circle ACE be described; [Post. 3]
and from the point C, in which the
10 circles cut one another, to the points
A, B let the straight lines CA, CB be
joined. [Post. 1]
Now, since the point A is the centre of

the circle CDB, AC is equal to AB.
15 [Def. 15] Again, since the point B is the centre of the circle CAE, BC is equal to BA.
[Def. 15] But CA was also proved equal to AB; therefore each of the straight lines CA,
CB is equal to AB.
And things which are equal to the same thing are also equal to one another; [C.N. 1]
therefore CA is also equal to CB. Therefore the three straight lines CA, AB, BC are equal
20 to one another.
Therefore the triangle ABC is equilateral; and it has been constructed on the given finite
straight line AB.
(Being) what it was required to do.
Heath, 1956, Vol. 1, pp. 241–242, line numbers adjusted
Both the steps of the construction and the proof are justified by references to
common notions, postulates and definitions that are stated earlier in the Elements (see
Table 1 for those referred to in Proof 1). Euclid’s “common notions” and “postulates”
are assumptions that are to be accepted without justification. Nowadays they would
usually both be called “axioms”. Euclid’s distinction between them is that common
notions apply outside of geometry, while his postulates are specific to geometry.
Euclid’s Elements is a structured presentation of the mathematics of that time.
He did not discover any of the theorems he presented, but he did present them as
part of a larger structure. The Elements provided the model for proof in mathematics,
and in other domains, for centuries.
Euclid’s contribution was the logical organisation of the Elements – its
axiomatic structure in which everything is carefully deduced from a small
number of definitions and assumptions. This structure served as a model for
Aquinas’s Summa Contra Gentiles, for Newton’s Principia, and for Spinoza’s
Ethics. The Elements has been the most influential textbook in history. (Anglin,
1994, p. 81)

via free access
CHAPTER 1
Table 1. References in Euclid’s proof of Prop. I.1
Common Notion 1 Things which are equal to the same thing are also equal to one
another.
Postulate 1 Let the following be postulated: To draw a straight line from
any point to any point.
Postulate 3 To describe a circle with any centre and distance.
Definition 15 A circle is a plane figure contained by one line such that all the
straight lines falling upon it from one point among those lying
within the figure are equal to one another.
Euclid’s axiomatic approach is not only of historical importance. It has become

a central motif in mathematics. Modern mathematical structures are often
systems based on definition and axioms and relying on rules of inference. ...
In such a system, a proposition is considered true if it can be derived from the
axioms in a finite number of logical steps using the permitted rules of inference.
(Hanna & Barbeau, 2002, p. 38)
Descartes (1596–1650) found the inspiration for his philosophical method in the
Elements. What impressed him was the way Euclid based geometry on a few
axioms and proceeded to deduce further statements about which one could have
absolute confidence.
Those long chains of reasoning, each of them simple and easy, that geometri-
cians commonly use to attain their most difficult demonstrations, have given
me an occasion for imagining that all the things that can fall within human
knowledge follow one another in the same way and that, provided only that one
abstain from accepting anything as true that is not true, and that one always
maintains the order to be followed in deducing the one from the other, there is
nothing so far distant that one cannot finally reach nor so hidden that one
cannot discover. (Descartes, 1637/1993, p. 11)
For thinkers from Aristotle and Descartes to the present day, the deductive
method is associated with certainty.
Euclid regarded his starting assumptions not as mere hypotheses, but as
truths. He intended to instantiate the ideal described by Aristotle as the
beginning of the Posterior Analytics: sure basic knowledge is obtained by the
rigorous deduction of the consequences of basic truths. These truths are either
definitions or existence assertions. (Anglin, 1994, p. 82)
Of all those who have already searched for truth in the sciences, only the
mathematicians were able to find demonstrations, that is, certain and evident
reasons. (Descartes, 1637/1993, p. 11)

via free access
HISTORY OF PROOF
The Restoration of Rigour

In the standard view European mathematics since the Renaissance is a continuation
of the work of the Greeks. True, there were times when the discovery of new
theorems and methods overtook the task of rigourously proving them, for example,
when Newton and Leibniz introduced the calculus in the seventeenth century. Their
justifications for their methods were criticised as not up to the standard of Euclid
by, for example, Berkeley. This difficulty was addressed finally in the nineteenth
century, by Cauchy and others (see Kleiner, 1991, for an outline of this history and
references to other sources).
At the same time criticisms of lack of rigour in analysis were being addressed,
an important development occurred in the history of geometry: the invention of
non-Euclidean geometries. For Descartes and Kant, Euclidean geometry was an
example of knowledge that was undeniably true. Its foundation was the nature of
space itself. When Lobachevsky, Bolyai and Gauss announced that it was possible
to construct a geometry in which one of Euclid’s postulates (the famous parallel
postulate) is false, it became possible to question the truth of Euclidean geometry.
Of course, there was strong temptation to assume there was something wrong with
non-Euclidean geometry, that there was a contradiction somewhere. In 1871 Klein
eliminated this possibility by proving that if there is a contradiction in the new non-
Euclidean geometries then there is also a contradiction in Euclidean geometry. This
created the need for a new approach to securing the foundations of geometry and
the rest of mathematics, as it was not longer possible to convincingly claim that
Euclidean geometry was the true geometry of space (Many histories of mathematics
tell this story. Kline, 1972, is thorough.).
This method of establishing the lack of contradiction (the ‘consistency’) of
one mathematical system by showing it is just as consistent as another system,
was applied not only to geometry. Hilbert showed that the various geometries
were as consistent as basic arithmetic, raising the question of the consistency of
arithmetic. The next step was to try to establish arithmetic on the basis of set
theory, and then to establish set theory on the basis of logic (the logicist approach
of Russell, Whitehead, etc.). In the late nineteenth and early twentieth centuries
this led to a new focus on axiomatisation and the axiomatisation of set theory (by
Frege, Russell and others), geometry (by Hilbert) and arithmetic (by Peano and
others).
The next step was to replace traditional mathematical statements with purely
formal statements that could themselves be the objects of calculation. This was
the objective of Hilbert’s formalism. Mathematics, from a formalist perspective,
is the manipulation of symbols, without any reference to any meaning or
interpretation.
Mathematical proof will consist of this process: the assertion of some
formula; the assertion that this formula implies another; the assertion of the
second formula. A sequence of such steps in which the asserted formulas or
the implications are proceeding axioms or conclusions will constitute the
proof of a theorem. Also, substitution of one symbol for another or a group of

via free access
CHAPTER 1
symbols is a permissible operation. Thus formulas are derived by applying

the rules for manipulating the symbols of previously established formulas.
(Kline, 1972, p. 1205)
Some key authors and publications in the history of mathematics, and the history of
thought more generally are listed in Table 2, to provide an overview of the standard
version of the history of proof as we have presented it here. Note that there is a
considerable gap between the work of Aristotle and Euclid and the European
philosophers and mathematicians who found those works significant. Authors and
works who are often not discussed when relating the standard history of proof are
omitted here, resulting in this gap. We are not claiming that no mathematics
happened in this period, only that it is usually ignored when the standard version is
presented. Some Chinese works we will discuss later are included as a hint that the
standard version may be incomplete.
Table 2. Chronology of people and publications mentioned in the text
600 BCE Thales, first historically recorded proofs

427–349 BCE Plato
384–322 BCE Aristotle (335 BCE Posterior Analytics)
300 BCE Euclid’s Elements
250 BCE–100 Jiuzhang Suanshu (exact date of composition uncertain, and parts may
CE be much older)
0
200
263 Liu Hui’s commentary on the Jiuzhang Suanshu
300–600 Theoretical phase in Chinese mathematics
800
1000
1200
1258–1264 Thomas Aquinas’s Summa Contra Gentiles composed
1400
1500–1557 Tartaglia
1596–1650 Descartes (1637 Method)
1632–77 Spinoza (1677 Ethics published posthumously)
1642–1727 Newton (1687 Philosophiæ Naturalis Principia Mathematica)
1646–1716 Leibniz
1685–1753 Berkeley (1734 The Analyst)
1724–1804 Kant
1789–1857 Cauchy
1830 Bolyai and Lobachevsky publish first treatises on non-Euclidean geometry
1845–1918 Cantor
1848–1925 Frege
1849–1925 Klein (1871 Proof that a contradiction in non-Euclidean geometry
implies a contradiction in Euclidean geometry)
1858–1932 Peano
1862–1943 Hilbert (1899 Foundations of Geometry)
1872–1970 Russell (1910–1913 Principia Mathematica, with Whitehead)

via free access
HISTORY OF PROOF
Degrees of formality. It is unlikely that the reader will have encountered many
proofs that meet the formalist definition of mathematical proof (an example is
Proof 27 in Chapter 7). However many proofs are called “formal” that do not meet
the formalist definition. This can be confusing when trying to interpret a statement
like “students of mathematics should understand formal proof ” (Moore, 1990, p. 57).
Lakatos (1978) distinguishes between three different degrees of formality in
proofs: pre-formal, formal, and post-formal. Lakatos uses “formal” in the same sense
as the formalists: A sequence of symbols that makes it possible to “mechanically
decide of any given alleged proof if it really was a proof or not” (p. 62). By “pre-
formal” he means a proof which is accepted as such by mathematicians, convincing,
but not a formal proof.
[In a pre-formal proof] there are no postulates, no well-defined underlying
logic, there does not seem to be any feasible way to formalize this reasoning.
What we were doing was intuitively showing that the theorem was true. This
is a very common way of establishing mathematical facts, as mathematicians
now say. The Greeks called this process deikmyne and I shall call it thought
experiment. (pp. 64–65)
A similar description has been adopted by a group of mathematics education
researchers we refer to as “preformalists”. We will discuss their work in Chapters 3
and 7. Note, however, that Lakatos refers specifically to informal proofs acceptable
to mathematicians, and the preformalists include proofs that could be made
acceptable. To avoid confusion we will follow the preformalists’ usage. We will use
“semi-formal” to refer to informal proofs acceptable to mathematicians, instead of
Lakatos’s term “pre-formal” and use the word “preformal” (without a hyphen) to
refer to the proofs discussed by the preformalists.
Lakatos’s “post-formal” proofs are proofs about formal proofs. For example, the
proof of the Duality Principle in Projective Geometry “Although projective
geometry is a fully axiomatized system, we cannot specify the axioms and rules
used to prove the Principle of Duality, as the meta-theory involved is informal.”
(p. 68). Other examples include the consistency and completeness proofs of formal
systems such as Gödel’s proof of his Incompleteness Theorem.
Most proofs are either preformal or semi-formal. However, the influence of the
formalists has made proofs in general more formal, and within the sub-disciplines
of mathematical foundations and computer science formal proofs are the norm.
In the early years of the twentieth century the mathematical community began to
have confidence that the formal structures they were developing would, for mathe-
matics at least, achieve what Leibniz had dreamed of in the eighteenth century, “an
exhaustive collection of logical forms of reasoning—a calculus ratiocinator—which
would permit any possible deductions from initial principles” (Kline, 1980, p. 183).
The standard history of proof suggests that this has, in fact, been achieved.
It is believed that every mathematical text can be formalised. Indeed, it is
believed that every mathematical text can be formalised within a single formal
language. This language is the language of formal set theory. (Davis & Hersh,
1981, p. 136)

via free access
CHAPTER 1
By formalising mathematics, it was possible to revise the proofs of the past to new
standards of rigour inspired by, but improving upon, Euclid’s Elements.
During the period from about 1821 to 1908 ... mathematicians restored and
surpassed the standards of rigour which had been established during the
period of classical Greek mathematics. (R. Jones, 1996)
Formalist work at the foundations of mathematics inspired the Bourbaki group in
France to apply axiomatic approaches to algebra and analysis, which in turn inspired
some of the reforms of the New Math curriculum reforms of the 1960s. This brings
us to the present day. In the standard view today’s proofs are direct descendants of
the proofs of Euclid, although today’s proofs make more use of symbols to make
formalisation easier. Like Euclid’s proofs they start from axioms and lead to results
that are “true” within the structure defined by the axioms.
OTHER VIEWS OF THE HISTORY OF PROOF
Not everyone accepts the standard view of the history of proof, and alternative
viewpoints have emerged that challenge many claims of the standard view. These
challenges are usually based on historical evidence and sociological analyses. In
this section we will discuss the challenges to several claims in standard view of the
history of proof, including these:
– Proof began in Greece and was limited to cultures with an intellectual connection
to Greece, primarily those in Europe. “For the early Egyptians, Babylonians,
and Chinese, the weight of observational evidence was enough to justify mathe-
matical statements” (Hanna & Barbeau, 2002, p. 36). If one accepts that “the
deductive method. ... has always been an essential characteristic of mathematics”
(Anglin, 1994, p. 63) then one must conclude that what the early Egyptians,
Babylonians, and Chinese did was not mathematics.
– In Euclid’s Elements “everything is carefully deduced from a small number of
definitions and assumptions” (Anglin, 1994, p. 81). Euclid’s proofs are models
of mathematical rigour.
– The work of Russell, Frege, etc. re-established mathematics on firm foundations.
In principle, every mathematical proof can be reduced to a sequence of formal
statements, in which each statement follows from previous statements according
to the rules of symbolic logic.
– Proofs transmit truth from established axioms to the theorems they prove. The
purpose of proofs is to make this connection from the axioms to the theorems.
We will consider each of these beliefs in turn, but first, it is important to note a
feature of the history of mathematics that makes any discussion of specific practices
problematic.
The history of mathematics is spread over a wide time period and a wide range of
cultures, and in many cases the data available is far from ideal. In the cases of Greek
and Chinese mathematics, the original sources are lost, and most of what we know
about them comes from sources written a thousand years after the originals. In the
case of the Greek texts the copies we have come from Arab sources that had their
own rich mathematical culture, which may have influenced the transcription and
10

via free access
HISTORY OF PROOF
translation of the texts they had available. We can see this process in more recent
cases; for example, in Heath’s translation of Euclid’s Elements from Greek to English
his footnotes indicate places where he chose to translate passages into formulations
more accessible to contemporary readers, but differing from the original Greek text
(this occurs often; see, for example, the footnotes to Book I Proposition 4, Heath,
1956, p. 248). Aside from purposeful changes to the texts, there are also accidental
changes and missing sections that have forced later translators, transcribers, and
historians to interpolate material that might not be the same as in the original. With
Egyptian and Mesopotamian sources, we are a bit better off in that some original
papyri and cuneiform tablets have survived, but the interpretation of them is a
challenge for experts as the original languages fell into disuse and had to be
reconstructed.
In addition, we have no way of knowing if the mathematical texts we have
from these cultures are representative of their mathematical practices. If one plucked
a book about mathematics at random from all those printed in the twenty-first
century, the likelihood is that one would end up with a school book, as many
more school books are printed than university texts or specialists’ monographs.
Such a school book would hardly represent the present level of development of
mathematics.
The lack of original data is one problem facing historians of mathematics, but
the diversity of the data is another. Anyone attempting to look at the whole picture
is forced to work from secondary or even tertiary sources, as the number of
academics with strong backgrounds in both mathematics and history, and able to
read Arabic, traditional Chinese, ancient Egyptian, classical Greek, Sanskrit and
Sumerian (to name only a few of the languages in which important mathematical
texts have been written) is probably very small. As Smoryński (2008) comments:
The further removed from the primary, the less reliable the source: errors are
made and propagated in copying; editing and summarising can omit relevant
details, and replace facts by interpretations; and speculations can become
established fact even though there is no evidence supporting the “fact”.
(p. 11)
These considerations alone should make one cautious of accepting the standard
view of the history of proof (or any view of the history of proof), and there are also
some other reasons to be wary.
Proof in China
The standard view of the history of proof claims that proof originated in Greece,
and that while Chinese mathematics includes many significant discoveries, the
Chinese did not prove. “Mikami [1913] considered the greatest deficiency in old
Chinese mathematical thought was the absence of the idea of rigorous proof ”
(Needham, 1959, p. 151). Since the 1960s, however, Western scholars have been
aware that there is at least one work of ancient Chinese mathematics in which
proofs play an important role:
11

via free access
CHAPTER 1
Proof 2: Figure reconstructed on the basis of Liu Hui’s commentary on Jiuzhang Suanshu
A right triangle has sides of 8 steps and 15 steps. What is the diameter of its inscribed circle?
Compute the Xian from the Gou

and the Gu, then add the three
together and divide this sum into
twice the product of the Gou and
the Gu.
[Find c from a and b (using the

Area
Pythagorean theorem), then add
= Twice the
a + b + c and divide by 2ab (the
product of the
area of the two figures to the
Gou and the Gu
right)]
= 2ab
adapted from Siu, 1993, p. 350 and Martzloff, 1989, p. 150,

whose reconstructions are based on that of Li Huang (?–1812)
[While] most Chinese mathematical works contain no justifications ... there is

one major exception, namely a set of Chinese argumentative discourses which
has been handed down to us from the first millennium AD. We are essentially
referring to the commentaries and sub-commentaries of the Jiuzhang
Suanshu, the key work which inaugurated Chinese mathematics and served as
a reference for it over a long period of its history. (Martzloff, 1997, p. 69)
The Jiuzhang Suanshu (Computational Prescriptions in Nine Chapters) is the oldest
Chinese mathematical text known to us, having been compiled beginning in 200
BCE (Martzloff, 1997, p. 124). The proofs in the commentaries by Liu Hui were
made at the end of the third century, at the beginning of a significant period in
Chinese mathematics:
From the third to the sixth century, Chinese mathematics entered its theoretical
phase. For the first time, it seems, importance was attached to proofs in their
own right, to the extent that trouble was taken to record these in writing.
Approximate values for the number π were then derived by computation and
reasoning, rather than simply via an empirical process. (p. 14)
12

via free access
HISTORY OF PROOF
It must be said, however, that Liu Hui’s proofs are not like Euclid’s proofs. For one
thing, Euclid expressed his proofs primarily through words, but Chinese mathe-
maticians made extensive use of diagrams. For example, consider Proof 2, which
shows a visual proof based on the text of Liu Hui’s commentary on problem 16 of
Chapter 9 of the Jiuzhang Suanshu. The original diagrams are, unfortunately, lost.
Siu (1993) suggests a reconstruction of the figures, which we have changed slightly
to make the argument clearer. In modern terms, the theorem proven is:
Given a right triangle with sides a, b, c, the diameter α of the inscribed circle
is 2ab / (a + b + c).
Note that the solution is given in general terms (the words “Gou” and “Gu” are
used to refer to the shorter and longer legs of the triangle, instead of using the
numbers given in the problem) and that it is assumed that the reader knows how to
calculate the length of the hypotenuse (the Xian) from the lengths of the legs (i.e.,
that the reader is familiar with the “Pythagorean” theorem).
This practice of basing proofs on visually convincing diagrams continued when
Euclid’s Elements was translated into Chinese after it was introduced by Jesuit
missionaries. For example, Mei Wending (1633–1721) made changes to Euclid’s
diagrams when he incorporated parts of the Elements into his Jihe bubian (Comple-
ments of Geometry).
He modified the figures to make them immediately readable, although Euclid
operated in the opposite direction, thus making it necessary to resort to deductive
reasoning. (Martzloff, 1997, p. 113)
This preference for readable figures over verbal descriptions is one reason why
Chinese proofs are still not accepted as proofs by some historians of mathematics
(Siu, 1993, p. 345).
The use of diagrams is sometimes rejected entirely and misleading diagrams are
given to support a claim that basing proofs on diagrams is not reliable.
The prevailing attitude is that pictures are really no more than heuristic
devices; they are psychologically suggestive and pedagogically important —
but they prove nothing. (Brown, 1999, p. 25)
Philosophers and mathematicians have long worried about diagrams in mathe-
matical reasoning — and rightly so. They can indeed be highly misleading.
(p. 43)
One such misleading diagram is shown in Figure 1.
There are also other reasons beyond the use of diagrams for the perception of
the rarity of proofs in Chinese mathematics. Perhaps the most important is the role
of Chinese mathematical texts as textbooks in established schools.
Under the Sui dynasty (518–617), and above all under the Tang dynasty
(618–907), mathematics was officially taught at the guozixue (School for the
Sons of the State), based on a set of contemporary or ancient textbooks as
written support. (Martzloff, 1997, p. 15)
13

via free access
CHAPTER 1
Figure 1. A misleading diagram showing that 8×8 = 5×13.
Indeed, to a first approximation, a majority of Chinese mathematical works

may be better represented as pedagogical tools, in other words, didactic aids
used to teach numerical computation, together with prescriptive texts (for
example, user manuals). (p. 47)
As with problem based textbooks today, the intent was that the students should
work out the proofs for themselves. In his commentary on Jiuzhang Suanshu, Liu
Hui “associates in the same sentence two famous passages from the Lunyu
(Confucian Analects) which both suggest an idea of the same order:”
I told him what had gone before and he understood what followed: [I]
showed him a corner [i.e. an aspect of the question] and he replied with the
other three.
Here the author indicates his wish not to disclose all the details of his
reasoning to the student.... Consequently, instead of giving the details of his
own thought processes, he often merely indicates that the solution of a given
problem is analogous to that of some other problem, or that “the remainder
follows in the same way” (Martzloff, 1997, p. 70, bracketed insertions in
quotation from original)
Anyone who has studied from a contemporary university mathematics textbook
will be familiar with hints like “the remainder follows in the same way.” They
often appear in exercises in which new theorems are stated without their proofs,
which are left to be discovered by the students. Herbst (2002b) describes how such
exercises came to be included in high school geometry textbooks in the late
nineteenth century.
A characteristic of Chinese culture that may also have affected the nature of
Chinese proofs was the emphasis in Chinese literary style on conciseness.
Such a knowledge excluded literary ornaments and excessively long passages
such as those found in Euclidean theorems and proofs. In particular, syllogisms
and other logical forms were especially unacceptable for they involved numerous
14

via free access
HISTORY OF PROOF
repetitions and redundancies: they were contradictory with the canons of

Chinese literary redaction which, in the case of technical subjects, valued
conciseness above all. (Martzloff, 1997, p. 112)
It is also worth remembering that in Western countries there have been times when
mathematicians, far from publishing proofs of their theorems, kept their results
secret (for example, Tartaglia). The same phenomenon occurred in China:
Last but not least, assuming that an inventor had succeeded in creating novel
procedures, it is not certain that he would have been inclined ipso facto to
reveal the secrets of their creation; in fact, the existence of rivalry between
calendarist astronomers is known. (Martzloff, 1997, p. 49)
In summary, the belief that the practice of proving mathematical results began in
Greece and spread from there to other, primarily European, cultures, is a myth.
Proving was also a part of Chinese mathematics from the third century, and possibly
earlier, but in a different style than Euclid’s. That many historians based a belief
that proof was not a part of Chinese mathematics on their lack of awareness of Liu
Hui’s commentaries is a useful reminder that absence of evidence is not the same
thing as evidence of absence. This might suggest that we approach with caution the
belief that proof was not a part of mathematics in other cultures where significant
mathematical discoveries were made (for example, Egypt, Mesopotamia, and India).
As an illustration of this point we will now briefly consider proof in India before
moving on the to the claim that Euclid’s proofs are models of rigour.
Proof in India
The standard history of mathematics makes claims about proof in India that are
similar to those made about China. For example, in describing Hindu mathematics
in the period 200–1200 CE Kline (1972) writes:
There is much good procedure and technical facility, but no evidence that
they considered proof at all. They had rules, but apparently no logical scruples.
(p. 190)
Joseph (1992, 1994) critiques this claim, pointing out that, as in China, proofs
(“upapatti”) were often included in commentaries on mathematical texts, even if they
were not a part of the texts themselves. There is, however, an important difference
between upapatti and Euclid’s proofs:
The upapattis of Indian mathematics are presented in a precise language,
displaying the steps of the argument and indicating the general principles which
are employed. In this sense they are no different from the “proofs” found in
modern mathematics. But what is peculiar to the upapattis is that while presen-
ting the argument in an “informal” manner (which is common in many mathe-
matical discourses anyway), they make no reference whatsoever to any fixed
set of axioms or link the given argument to “formal deductions” performable
with the aid of such axioms. (1992, p. 194; see also 1994, p. 200)
15

via free access
CHAPTER 1
It is this lack of reference to axioms that distinguishes the upapattis of India

from Euclid’s proofs. This is related, as Joseph notes, to a focus on rigour and
certainty that is important to the Euclidean approach as opposed to an emphasis on
understanding and clarity that is present in the Indian tradition. We will encounter
similar differences in the role of proof in subsequent chapters (especially Chapter 5).
Now, however, we will turn to the question of whether Euclid’s proofs are as
rigourous and his theorems are established with certainty as the standard history of
proof claims.
The Euclid Myth

In the standard view Euclid’s proofs are taken to be models of mathematical rigour,
that establish theorems with certainty. While there is no denying that Euclid’s
Elements has had a profound influence on Western mathematics, the claim of
certainty has been questioned:
What is the Euclid myth? It is the belief that the books of Euclid contain
truths about the universe which are clear and indubitable. Starting from self-
evident truths, and proceeding by rigorous proof, Euclid arrives at knowledge
which is certain, objective, and eternal. Even now, it seems that most educated
people believe in the Euclid myth. Up to the middle of the nineteenth century,
the myth was unchallenged. Everyone believed it. (Davis & Hersh, 1981,
p. 325)
There are two aspects to this “myth”. One is the assertion that Euclid’s proofs are
rigourous, and the second that the knowledge arrived at using the deductive method
is certain and objective. Here we will discuss the first of these aspects, rigour.
Later we will consider the second aspect in the context of twentieth century
mathematics.
Rigour in Euclid’s proofs. Do the proofs in the Elements live up to the claims
sometimes made about them, that “everything is carefully deduced from a small
number of definitions and assumptions” (Anglin, 1994, p. 81)? In fact, they do not.
Euclid’s proofs make use of assumptions that are never stated, some involve
reference to physical manipulations (as in Liu Hui’s proofs) and some use specific
cases to justify general conclusions.
Since Euclid still has popularity, and even with mathematicians, a reputation
for rigour in virtue of which his circumlocution and longwindedness are
condoned, it may be worth while to point out, to begin with, a few of the
errors in his first twenty-six propositions. (Russell, 1903/1937, p. 404)
Recall, for example, the first proof in Book I, the construction of an equilateral
triangle (see Proof 1, on page 5). Each step in the construction (lines 5–12)
indicates the postulate that states that such a construction is possible, and each step
in the proof (lines 13–22) indicates which definition, postulate or common notion
justifies that step in the argument.
16

via free access
HISTORY OF PROOF
Proof 3: Elements Book I Proposition 4
Proposition 4.
If two triangles have the two sides equal to two sides respectively, and have the angles
contained by the equal straight lines equal, they will also have the base equal to the
base, the triangle will be equal to the triangle, and the remaining angles will be equal
5 to the remaining angles respectively, namely those which the equal sides subtend.
Let ABC, DEF be

two triangles having
the two sides AB, AC
equal to the two
10 sides DE, DF
respectively, namely
AB to DE and AC to
DF, and the angle
BAC equal to the
15 angle EDF. I say that the base BC is also equal to the base EF, the triangle ABC will be
equal to the triangle DEF, and the remaining angles will be equal to the remaining
angles respectively, namely those which the equal sides subtend, that is, the angle ABC
to the angle DEF, and the angle ACB to the angle DFE.
For, if the triangle ABC be applied to the triangle DEF, and if the point A be placed on
20 the point D and the straight line AB on DE, then the point B will also coincide with E,
because AB is equal to DE.
Again, AB coinciding with DE, the straight line AC will also coincide with DF, because
the angle BAC is equal to the angle EDF; hence the point C will also coincide with the
point F, because AC is again equal to DF. But B also coincided with E; hence the base
25 BC will coincide with the base EF.
[For if, when B coincides with E and C with F, the base BC does not coincide with the
base EF, two straight lines will enclose a space: which is impossible. Therefore the base
BC will coincide with EF] and will be equal to it. [C.N. 4]
Thus the whole triangle ABC will coincide with the whole triangle DEF, and will be
30 equal to it. And the remaining angles will also coincide with the remaining angles and
will be equal to them, the angle ABC to the angle DEF, and the angle ACB to the angle
DFE.
Therefore etc.
(Being) what it was required to prove.
Heath, 1956, Vol. 1, pp. 247–248, line numbers adjusted
Notice, in line 9, the mention of “the point C, in which the circles cut one another.”
From the diagram it is clear that there are two such points. That Euclid seems to
claim that there is only one is perhaps a minor flaw. He is only trying to prove that
it is possible to construct an equilateral triangle; that the construction might
produce two does not make it invalid.
More significantly, Euclid does not provide a common notion, postulate, or
definition to let us know when we can actually construct a “point C, in which the
circles cut one another.” While these circles intersect, it has not been established
17

via free access
CHAPTER 1
under what conditions the intersections will exist, or that these circles satisfy these
unstated conditions. Of course, from the diagram it is clear that C exists, but the
claim made in the Euclid myth is that Euclid reasons “from a small number of
definitions and assumptions” not from diagrams.
We turn now to a proof that, like Liu Hui’s proofs, makes use of physical
manipulations. It is Euclid’s proof of Proposition 4 of Book 1 (see Proof 3). One
thing that is interesting about the proof (lines 19–32) is the lack of references to
common notions, postulates or definitions. In fact, the only such reference (in line 28)
is thought to be a later interpolation (Heath, 1956, p. 249). This is not surprising
when one considers that the whole argument depends on the idea of picking up one
triangle and putting it on top of the other one. The phrase “if the triangle ABC be
applied to the triangle DEF ” (line 19) suggests that ABC be moved so that it
coincides with DEF.
This way of reasoning is not what Euclid is supposed to have done, but it is
quite similar to a way of reasoning used by Liu Hui:
Thus, the argumentation inevitably depends on methods. For example: ...
Recourse to non-linguistic means of communication. This is necessary
because, according to the adage of the Yijing cited by the commentator [Liu
Hui], “not all thoughts can be adequately expressed in words” ... In place of a
discourse, the reader is asked to put together jigsaw pieces, to look at a figure
or to undertake calculations which themselves constitute the sole justification
of the matter at hand. In each of these cases language is purely auxiliary to
such procedures. (Martzloff, 1997, pp. 71–72)
In Euclid’s proof we are asked to make ABC coincide with DEF in our imaginations,
and then to note the correspondences Euclid points out. This is easy to do, and
quite convincing, but it is not the deductive method as described by Aristotle.
To be fair, Euclid did not reason in this way very often, and it is not, in fact,
possible to deduce this proposition from his common notions, postulates and defini-
tions, so he had to depart from the deductive method, or change his postulates.
Hilbert took the latter approach in his Grundlagen der Geometrie (Foundations of
Geometry, 1899/1921) and added this proposition as an axiom.
Reasoning from visual evidence was a mainstay of Chinese and Hindu mathe-
matics, but fell out of fashion in Greece. There is evidence, however, that it was the
basis for Greek mathematics as well for some time. Euclid’s proof of Proposition
I.4 is part of this evidence. And, as Martzloff (1997) points out:
The Greek technical term meaning “to prove” is the verb δείκνυμι. Euclid
uses this at the end of each of his proofs. Originally this verb had the precise
meaning of “to point out,” “to show” or “to make visible.” Thus it appears
that the Chinese proofs of Liu Hui and Li Chunfeng were similar in nature to
the first known historical proofs, an example of which is given by Plato
(well-known dialogue in which Socrates asks a slave how to double the area
of a square); moreover, visual elements remained an essential component of
proofs in China for a long time, while in Greece these were abandoned at an
early stage although figurative references were maintained. (pp. 72–73)
18

via free access
HISTORY OF PROOF
Proof 4: Elements Book IX Proposition 20
PROPOSITION 20.
Prime numbers are more than any assigned multitude of prime numbers.
Let A, B, C be the assigned prime
numbers; I say that there are more
5 prime numbers than A, B, C.
For let the least number measured by

A, B, C be taken, and let it be DE; let
the unit DF be added to DE. Then EF is either prime or not.
First, let it be prime; then the prime numbers A, B, C, EF have been found which are
10 more than A, B, C.
Next, let EF not be prime; therefore it is measured by some prime number. [VII. 31]
Let it be measured by the prime number G. I say that G is not the same with any of the
numbers A, B, C.
For, if possible, let it be so. Now A, B, C measure DE; therefore G also will measure
15 DE. But it also measures EF. Therefore G, being a number, will measure the remainder,
the unit DF: which is absurd.
Therefore G is not the same with any one of the numbers A, B, C. And by hypothesis it
is prime.
Therefore the prime numbers A, B, C, G have been found which are more than the
20 assigned multitude of A, B, C. Q. E. D.
Heath, 1956, Vol. 2, p. 413, line numbers adjusted
Martzloff’s comment on the abandonment of visual elements by the Greeks is part

of the standard view of history. Netz (1998, 1999), however, has suggested that the
visual element was not so much abandoned as hidden by the Greeks. He finds
evidence for this in the abundance of points that are undefined except by the diagrams
accompanying the proofs, and by the lettering of the diagrams which suggests that
the proof was created before it was written down.
Netz describes a three stage process:
These three stages are:
(i) drawing a diagram;
(ii) a dress rehearsal in front of the diagram, in which the diagram is dressed,
i.e. letters are inserted
(iii) a full production, writing down the proof. (1998, p. 36)
If Netz is correct, Greek proofs were basically visual and oral, with the written
proof being a record of what came before. We have been equating Greek proofs
with written texts, not because they were, but because all the evidence that survived
was the written text.
We now turn to Euclid’s famous proof of the infinitude of primes (see Proof 4).
In the Elements it has a different form from that usually given in textbooks of
number theory. For an example of a modern version, see Proof 24 in Chapter 7.
19

via free access
CHAPTER 1
There are two things of interest about this proof. First, the diagram is not needed,
but it is there anyway. This is true in general of Euclid’s proofs in what we would now
call number theory. Netz (1998) explains the presence of diagrams in these contexts.
He notes that the Greek word “diagramma” refers to more than “a diagram.”
It means something much closer to “a proposition” or “a proof” (see Knorr,
1975, pp. 69–75). This is a notorious fact about Greek practice: it is generally
difficult to tell whether the authors speak about drawing a figure or proving
an assertion, and this is because the same words are used for both. And this
again is because the diagram is the proof, it is the essence of the proof for the
Greek, the metonym of the proof. (Netz, 1998, pp. 37–38)
So while Greek proofs are often taken as the model of modern discursive proofs,
for the Greeks themselves they were fundamentally, essentially, associated with
pictures. This explains why diagrams were included in proofs like Proof 4 which
from a modern perspective do not need diagrams. To the Greeks, if it did not have a
diagram, it was not a proof. The presence of diagrams where they are not needed
may also be related to the origins of proofs as visual arguments, noted above.
The second thing of interest about this proof is that it does not, strictly speaking,
establish that there are an infinite number of prime numbers. What it shows is that
if there are three prime numbers, then there must be four prime numbers. This
specific case is used to stand for all cases, a technique which is also common in
Chinese proofs.
Thus, the argumentation inevitably depends on methods. For example: ...
Passage from the particular to the general, based on a specific, well-chosen
example. (Martzloff, 1997, p. 71)
This method of proving, known as using a generic example (see Chapter 7), is
unavoidable if one does not have a method of representing unspecified numbers
symbolically. Euclid had one such method, representing a number as the length of a
line segment, but he did not have a method for representing an unspecified number
of numbers as he had to do in this proof.
To summarise, Euclid’s proofs do not rigourously use deductive reasoning to
derive propositions from axioms (common notions, postulates), definitions, and
previously established propositions. They use implicit axioms, non-verbal arguments,
and generic examples. This undermines the claim that they establish the proposi-
tions they prove with certainty. Nonetheless, they have been the model and measure
of proofs in the Western mathematical tradition for thousands of years. In Chapter 5
we will revisit the role of proving in mathematics and explore some reasons why
the flaws in Euclid’s proofs were not considered serious (or even noticed) until the
beginning of the twentieth century, and why they are still being offered as model
mathematical proofs (e.g., by Hanna & Barbeau, 2002).
The Twentieth Century: Formalism to the Rescue?

The standard view of the history of proof acknowledges that there were some diffi-
culties with less than rigourous proofs at times, especially in the seventeenth century.
20

via free access
HISTORY OF PROOF
But the severity of these difficulties is sometimes minimised, and the success of the
efforts to solve them overrated. According to the standard view, even if Euclid’s
proofs are flawed, they were a step in the right direction, and since the work of the
formalists in the early twentieth century, mathematics has once again been placed
on firm foundations. Now, in principle, any mathematical proof can be expressed in
purely formal statements, which can then be checked mechanically, with no chance
of error due to missing assumptions, unclear definitions, use of diagrams, or logical
mistakes. But this is never done, and not only for pragmatic reasons.
An ordinary page of mathematical exposition may occasionally consist entirely
of mathematical symbols. To a casual eye, it may seem that there is little
difference between such a page of ordinary mathematical text and a text in a
formal language. But there is a crucial difference which becomes unmistakable
when one reads the text. Any steps which are purely mechanical may be
omitted from an ordinary mathematical text. It is sufficient to give the starting
point and the final result. The steps that are included in such a text are those
that are not purely mechanical — that involve some constructive idea, the intro-
duction of some new element into the calculation. To read a mathematical
text with understanding, one must supply the new idea which justifies the
steps that are written down. (Davis & Hersh, 1981, p. 139)
The missing steps would first have to be supplied before the proof could be
formalised. This would have to be done by an expert in the field, and even if an
expert could be found with the patience for such a task, there would be no guarantee
that the translation of the proof into a formal language would be free of error. The
problem of checking the correctness of the proof becomes the problem of checking
the correctness of the translation into formal language, and that is not formalisable.
The actual situation is this. On the one side, we have real mathematics, with
proofs which are established by “consensus of the qualified.” A real proof is
not checkable by a machine, or even by any mathematician not privy to the
gestalt, the mode of thought of the particular field of mathematics in which
the proof is located. Even to the “qualified reader,” there are normally
differences of opinion as to whether a real proof (i.e., one that is actually
spoken or written down) is complete and correct. These doubts are resolved
by communication and explanation, never by transcribing the proof into first-
order predicate calculus. Once a proof is “accepted,” the results of the proof
are regarded as true (with very high probability). It may take generations to
detect an error in a proof. If a theorem is widely known and used, its proof
frequently studied, if alternate proofs are invented, if it has known applications
and generalisations and is analogous to known results in related areas, then it
comes to be regarded as “rock bottom.” In this way, of course, all arithmetic
and Euclidean geometry are rock bottom.
On the other side, to be distinguished from real mathematics, we have “meta-
mathematics” or “first-order logic.” As an activity, this is indeed part of real
mathematics. But as to its content, it portrays a structure of proofs which are
21

via free access
CHAPTER 1
indeed infallible “in principle.” We are thereby able to study mathematically

the consequences of an imagined ability to construct infallible proofs (Davis &
Hersh, 1981, pp. 354–355)
This leaves us with two sorts of mathematics: first-order logic in which it is possible
to begin with formal axioms and definitions and derive results using formalised
logical rules without fear of error, and what Davis and Hersh call “real mathematics”
in which we begin with axioms and definitions that are expressed in a mixture of
formal and informal language, and derive results using informal logical rules, with
an ever present possibility of error. In the first sort proofs are formal, while in the
second they are semi-formal. One thing these two sorts of mathematics share with
each other, and with the standard view of Euclid’s Elements, is starting from
axioms and definitions that are established beforehand, and on which the truth
(whether absolute or relative, certain or probable) rests. This aspect of the standard
view has also had its critics, as we will see in the next section.
Lakatos and the Retransmission of Falsity

Lakatos (1961, 1976) describes the process by which he claims mathematics is
discovered. In his work he criticises the Euclidean structure of definitions,
postulates, theorems and proofs and the formalist reduction of mathematics to
formal logic.
The history of mathematics and the logic of mathematical discovery, i.e. the
phylogenesis and the ontogenesis of mathematical thought, cannot be developed
without the criticism and ultimate rejection of formalism. (1976, p. 4)
We will concentrate here on Lakatos’s critique of one aspect of the standard view
of proof, the claim that proofs are based on axioms and definitions that are
established beforehand, that it is the (assumed) truth of the axioms that is the basis
for the claimed truth of theorems. As Lakatos notes, this basing of theorems on
axioms is reflected in the presentation of mathematics in textbooks and journals.
Euclidean methodology has developed a certain obligatory style of presentation.
I shall refer to this as ‘deductivist style’. This style starts with a painstakingly
stated list of axioms, lemmas and/or definitions. The axioms and definitions
frequently look artificial and mystifyingly complicated. One is never told how
these complications arose. The list of axioms and definitions is followed by
the carefully worded theorems. These are loaded with heavy-going conditions;
it seems impossible that anyone should ever have guessed them. The theorem
is followed by the proof. (p. 142)
According to Lakatos, the order in this presentation is almost entirely the reverse of
mathematical practice. Rather than beginning with axioms and definitions, he says,
mathematicians begin with conjectures. After a conjecture comes a proof, but a
proof is not a guarantee of the truth of the conjecture. Instead it is “a rough thought-
experiment or argument, decomposing the primitive conjecture into subconjectures
or lemmas” (p. 127). Proving is a means of analysing the conjecture, part of a
22

via free access
HISTORY OF PROOF
process he calls ‘proof-analysis’. The next stage in the process is the emergence of
counterexamples to the conjecture. These counterexamples can reveal problematic
definitions and hidden assumptions. Lakatos divides them into three types: The
first is a counterexample to some step of the proof, but not to the conjecture itself
(It is local but not global.). The second is a counterexample to some step of the
proof, and to the conjecture (It is both local and global.). The third type does not
contradict any step of the proof, and yet it is a counterexample to the conjecture
(It is global but not local.). Each type plays a different role in the proof-analysis
(p. 43). A first-type counterexample signals that there is a problem with the proof;
either a hidden assumption must be revealed, or a definition changed, or a new
proof produced. A second-type counterexample is the most important type for
proof-analysis. When a second type counterexample emerges, the next step is to re-
examine the proof to locate the step to which it is a local counterexample, the
“guilty lemma”.
This guilty lemma may have previously remained “hidden” or may have been
misidentified. Now it is made explicit, and built into the primitive conjecture
as a condition. The theorem — the improved conjecture — supersedes the
primitive conjecture with the new proof-generated concept as its paramount
new feature. (p. 127)
The process of proof-analysis is not primarily about proving the conjecture that
was its beginning, but rather improving the definitions and axioms on which it is
meant to be based. “Proof-generated concepts” are important original contributions
to mathematics. They account for the facts that “axioms and definitions frequently
look artificial and mystifyingly complicated” and that theorems “are loaded with
heavy-going conditions” (p. 142).
Counterexamples of the third type exist only if the proof analysis is invalid.
A proof-analysis is ‘rigorous’ or ‘valid’ and the corresponding mathematical
theorem true if, and only if, there is no ‘third-type’ counterexample to it. I call
this criterion the Principle of Retransmission of Falsity because it demands
that global counterexamples be also local: falsehood should be retransmitted
from the naive conjecture to the lemmas, from the consequent of the theorem
to its antecedent. (p. 47)
The Principle of Retransmission of Falsity is very important to Lakatos’s thinking,
and sums up what may be the most important critique in his work of the standard
view of proof. In the standard view, truth is transmitted from axioms to increasingly
complicated theorems. Lakatos claims that this is impossible, but more importantly,
that this does not reflect the way mathematics really works. Mathematics progresses
by the retransmission of falsity from conjectures to axioms and definitions. In this
way counterexamples to conjectures reveal problems with the axioms and definitions.
Many concepts in mathematics have existed since before the time of Euclid, but
these concepts are now much more sophisticated, because conjectures based on
them turned out to give rise to counterexamples which forced (because of the
Principle of Retransmission of Falsity) changes to be made to the concepts.
23

via free access
CHAPTER 1
According to Lakatos, Euclid’s presentation of geometry seriously distorted the

nature of mathematical discovery and the role of proof in it by inverting the order
of things and hiding the importance of conjectures and counterexamples.
One should not forget that while proof-analysis concludes with a theorem, the
Euclidean proof starts with it. In the Euclidean methodology there are no
conjectures, only theorems. (p. 107, footnote 3)
This Euclidean version of proof is an integral part of the standard view of the
history of proof, and so it has had significant impacts on the teaching of proof, which
we will discuss in later chapters.
SUMMARY
In this chapter we have summarised what we call the standard view of the history
of proof, and described some important limitations and flaws of this view. Most
notably:
While many sources claim that proof originated in Greece and was not a part
of the intellectual activity of other cultures, there is clear evidence of proving
in ancient China and India, and it is possible that proving was part of
mathematics elsewhere, in spite of the absence of evidence.
Euclid’s proofs are said to be models of rigour, however they make use of
unstated assumptions and evidence from diagrams.
It is believed that mathematical proofs are (or can be made) formal, and that
this means they are absolutely rigourous. In fact, formalisation of most proofs
is not possible, and proofs can only be checked by a “qualified reader”.
Proofs are said to transmit truth from established axioms to the theorems they
prove, but as Lakatos points out, the process can go the other way; proofs
allow us to locate hidden assumptions and flawed axioms by retransmitting
falsity from a conjecture with counterexamples to the underlying definitions
and axioms.
Euclid himself could never have imagined the consequences of his effort to
systematise the mathematics known in his day, and so it is unfair to blame him for
the confusion resulting from the standard view of proof. As his name keeps coming
up, however, it is convenient to use labels like Davis and Hersh’s “Euclid myth”
and Lakatos’s “Euclidean methodology” to describe this point of view. And as long
as we are clear that we are speaking of a particular perspective, held by many
people, even today, and not of a long dead mathematician, we would agree with
Lakatos that:
Euclid has been the evil genius particularly for the history of mathematics
and for the teaching of mathematics, both on the introductory and the creative
levels. (p. 140)
24

via free access
CHAPTER 2
USAGES OF “PROOF” AND “PROVING”
The words “proof ” and “proving” are used in everyday life, mathematics, and mathe-
matics education in a number of distinct ways, usually without comment. For resear-
chers in mathematics education this can lead to confusion and may be a serious
obstacle to future research (Balacheff, 2002/2004; Reid, 2005). Without trying to
establish the “right” usages of these words, we will outline here some frequent
ones and describe the differences between them.
As you read this chapter you may want to reflect on these questions:
– What does “proof ” mean to you?
– What should “proof ” mean to students in schools?
– How can you determine what an author means by “proof ”?
EVERYDAY USAGES
In everyday English, “proof ” and “proving” can refer to convincing someone of

something, or to testing something to see if it is correct.
Convincing
When we doubt a statement, we may ask, “Do you have any proof of that? Can you
prove it?” In these questions proof means evidence, and proving means convincing.
When Shakespeare’s Othello says, “Be sure of it; give me the ocular proof ” (Act III,
scene 1) he means that Iago must convince him of the truth of his accusation by
providing visible evidence. What counts as convincing evidence depends on context,
and may include physical force, verbal abuse, social pressure, or anything else that
persuades someone else. In the Sidney Harris cartoon captioned “You want proof ?
I’ll give you proof !” the humour comes from a shift in context, as one mathematician
is shown convincing another mathematician by punching him in the nose, which is
an everyday, but not a mathematical usage of “proof ” as convincing.
Testing
“Prove” is derived from the Latin verb probare, which means to test, to try. The
English verb “probe” still carries this meaning. Taking “prove” as meaning “con-
vince” when it means “test” can lead to odd interpretations of common expressions.
For example, the expression “the exception which proves the rule” is often taken in
the paradoxical sense of asserting that the presence of a counterexample establishes
25

via free access
CHAPTER 2
the general truth of a rule, which follows if “prove” is taken to mean providing
convincing evidence. However, the expression is not so paradoxical if “prove” is
being used to mean “test”. Then saying “the exception proves the rules” amounts
to suggesting that examining exceptions closely and reasoning out the way they
occur can lead to a clarification and improvement of the rule. This interpretation
is reminiscent of Lakatos’s (1976) process of proof-analysis in which counter-
examples and proving interact to improve theorems in mathematics (see Chapters 1
and 11).
The use of “prove” to mean “test, try” can also occur in the noun form; a
“proof ” can be a test or a trial. In some common phrases, “proof-read,” “proof of
the pudding,” “100 proof,” the word “proof ” is used in this way. Words like
“waterproof” and “fireproof” are also based on this meaning; they describe objects
that have been tested and found to be resistant.
SCIENTIFIC USAGES
When one reads an article about a scientific discovery, one might encounter the
words “proof ” and “proving” used to refer to convincing, but on the basis of
special types of evidence.
Experiments Prove Existence Of Atomic Chain ‘Anchors’
Atoms at the ends of self-assembled atomic chains act like anchors with
lower energy levels than the “links” in the chain, according to new measure-
ments by physicists at the National Institute of Standards and Technology
(NIST).
The first-ever proof of the formation of “end states” in atomic chains may
help scientists design nanostructures, such as electrical wires made “from the
atoms up,” with desired electrical properties. (NIST, 2005, italics added)
When scientists “prove” something they offer convincing evidence, but that evidence
must be of a special type appropriate to science.
MATHEMATICAL USAGES
Godino and Recio (1997, Recio & Godino, 2001) make a distinction between two
usages of the words “proof ” and “proving” in two areas of mathematics: foun-
dations of mathematics and mainstream mathematics. This distinction is similar to
the distinction made by Douek (1998) between “formal proofs” and “mathematical
proofs”, the distinction made by Davis and Hersh (1981) between metamathematics
and “real mathematics”, and our distinction between formal proofs and semi-formal
proofs which we mentioned in Chapter 1.
In foundations of mathematics, proofs give theorems “a universal and intemporal
validity”, “they rest on the validity of the logic rules used,” “the use of formal lang-
uages is required,” and proving is a way of coming to grips “with the theoretical
26

via free access
problem of organizing and structuring the system of mathematical knowledge”

(Godino & Recio, 1997, pp. 315–316). Hanna (1983) defines formal proofs in this
way:
The term rigorous proof or formal proof ... is understood here to mean a proof
in mathematics or logic which satisfies two conditions of explicitness. First,
every definition, assumption, and rule of inference appealed to in the proof
has been, or could be, explicitly stated; in other words, the proof is carried out
within the frame of reference of a specific known axiomatic system. Second,
every step in the chain of deductions which constitutes the proof is set out
explicitly. (p. 3)
In mainstream professional mathematics, theorems do not have the “character of
absolute and necessary truths”, the validity of proofs is “‘judged by qualified
judges’ (Hersh, 1993, p. 389)”, “proofs are deductive but not formal,” and proving
is a way “to solve new problems, to increase the knowledge body, and, secondarily,
to organize and found the whole system of mathematics” (Godino & Recio, 1997,
pp. 316–317).
Mathematicians working in the foundational domain of proof theory recognise
this distinction, and use “formal proof” to refer to the proofs they study, and
“social proof ” to refer to the proofs of mainstream mathematicians. In Chapter 1
we suggested the adjective “semi-formal” to refer to these mainstream proofs. Davis
and Hersh (1981) note that although an “ideal” mainstream mathematician might
claim his semi-formal proofs meet the same criteria as those of mathematicians
working on foundations, when pressed, he would admit the differences:
What you do is, you write down the axioms of your theory in a formal
language with a given list of symbols or alphabet. Then you write down the
hypothesis in the same symbolism. Then you show that you can transform the
hypothesis step by step, using the rules of logic, till you get the conclusion.
That’s a proof. ... Oh, of course no one ever really does it. It would take
forever! ... [A proof is really] an argument that convinces someone who
knows the subject. (pp. 39–40)
USAGES IN MATHEMATICS EDUCATION RESEARCH
Many researchers in mathematics education use the words “proof ” and “proving”,
in a number of distinct ways. Most use the words in different ways within the same
paper.
For example, consider this sentence:
Even when students seem to understand the function of proof in the mathe-
matics classroom ... and to recognise that proofs must be general, they still
frequently fail to employ an accepted method of proving to convince themselves
of the truth of a new conjecture, preferring instead to rely on pragmatic methods
and more data. (Hoyles & Küchemann, 2002, p. 194, references removed for
clarity, italics added)
27

via free access
CHAPTER 2
The use of the singular form “proof ” in the first line instead of the plural “proofs”
suggests that the word is being used to mean a concept or category. In the second
line the plural is used, suggesting that a set of objects is meant. Finally in the third
line, the verb “proving” is used. The fact that “method of proving” seems to include
“accepted methods” as well as “pragmatic methods” suggests that “proving” is used
to mean something different than “constructing a proof” in this case.
This suggests a starting point for an investigation of the usage of the words
“proof ” and “proving” in research in mathematics education. Three categories of
usage can be distinguished on purely grammatical grounds:
1) The use of “proof ” in the singular, without an article to refer to a concept.
2) The use of “proof ” with an article or in the plural to refer to an object.
3) The use of the verb “prove” to refer to an action or process.
Note that “proving” is a difficult case, as it can be a form of the verb “prove” but
also a noun: “Jim is proving the theorem” or “Jim’s proving of the theorem”.
Considering word usage in mathematics education research even at the surface
level of the forms of words reveals some striking differences. For example,
consider the frequency of the use of the words “proof ”, “proofs”, “prove” (including
“proves” “proven” and “proved”), “proving”, words beginning with “argu+”
(“argument”, “arguing”, “argue”, etc.) and “reasoning”. In Figure 2 the frequency
of the usage of these words in three papers published in Educational Studies in
Mathematics is shown. The left hand column shows an example of an author
(Fischbein, 1999) who uses the verb “prove” more often than the nouns “proof ”
and “proofs”. In contrast, the right hand column shows an example of an author
(Uhlig, 2002) who uses the nouns much more than the verb. It is clear from the
centre column that Hanna (2000) uses the word “proof ” much more than “prove”,
but it is not clear whether she means a concept or an object when she writes
“proof ”. A closer look at the article clarifies this. Hanna’s use of “proof ” breaks
down into four categories:
– “proofs” in the plural form, 32 occurrences, 18%
– “proof ” preceded by “a”, 17 occurrences, 10%
– “proof ” preceded by “the”, 7 occurrences, 4%
– other uses of “proof ”, 122 occurrences, 71%
The final category still contains a few uses of “proof ” to refer to an object (for
example when it is preceded by an adjective, e.g., “an explanatory proof”), but
most uses of the word refer to a concept.
In the following we will go into more detail about the ways mathematics
education researchers use “proof ” and “proving” to refer to a concept, an object, or a
process. Note, however, that we do not claim that any researcher’s usages fall neatly
into a single category, nor that the usages we describe here are themselves disjoint
categories. As the quote from Hoyles and Küchemann at the start of this section
indicates, several usages can occur in a single paragraph. And while the three ESM
articles analysed in Figure 2 show the predominance of some usages over others,
almost all usages appear in all three articles.
28

via free access
Figure 2. Word usage in three ESM papers
Proof as a Concept
The use of the word “proof ” to refer to a concept is usually clear from the context
or from syntactical considerations, but once one knows that the word is intended to
refer to a concept, does one know to what concept it is meant to refer? Unfortunately,
no. Researchers in mathematics education have a wide range of perspectives on
proof which make it difficult to know what concept they might mean by the word
“proof ”. In the next chapter we will describe some researchers’ perspectives, but
as many researchers do not provide enough clues in their writing to definitively
identify their perspectives, we can only leave the reader with the advice to be wary.
Proof as an Object
There are a number of different objects “proofs” can refer to in mathematics education
research, and those objects can be distinguished by their forms or by their function.
The two most common usages are to refer to texts, usually written texts, of a certain
form, or to refer to arguments, spoken or written, with the function of convincing.
Proof-Texts.
The majority of the [high attaining 14 and 15 year old] students were unable
to construct valid proofs in [the domain of number and algebra]. (Healy &
Hoyles, 2000, p. 425)
29

via free access
CHAPTER 2
In mathematics schoolbooks and journals, one encounters some texts under the
heading “proof ”. Such proof-texts are characterised by a particular form and style.
The proof-texts of schoolbooks are different from the proof-texts of professional
mathematics journals, but there is sufficient unity in the styles to justify the use
of the same term for both. Writing proof-texts is a goal of recent reform documents:
“High school students should be able to present mathematical arguments in written
forms that would be acceptable to professional mathematicians” (National Council
of Teachers of Mathematics [NCTM], 2000, p. 58). That students cannot do this is
what is meant when researchers such as Duval (1990), Senk (1985) and Healy and
Hoyles (2000) conclude that students do not understand proof.
“Proving” can refer to writing a proof-text (e.g., Douek, 1998) but not everyone
who uses “proof ” to refer to proof-texts uses “proving” in this way.
Convincing arguments. The use of “proof ” to refer to a convincing argument by

mathematics education researchers is essentially a return to the everyday usage we
noted above. But the audience to be convinced can vary. For example, for Mason,
Burton, and Stacey (1982) a proof is an argument that convinces an enemy, for
Davis and Hersh (1981) it is an argument that convinces a mathematician who
knows the subject and for Volmink (1990) it is an argument that convinces a
reasonable sceptic. In all cases, it is not the argument itself that makes it a proof,
but rather that fact that it convinces someone. As Manin points out, in this often cited
quotation, “A proof becomes a proof after the social act of ‘accepting it as a proof’.
This is true of mathematics as it is in physics, linguistics, and biology.” (1977,
p. 48). Using proof this way necessarily means whether a given proof-text is a
proof or not can vary. “Proof is that which compels belief. That means that proof is
different in different eras, and indeed, that it is different for different people at any
one time.” (MacKernan, 1996, p. 14).
If “proof ” is used to mean a convincing argument, then “proving” usually refers
to convincing someone of something. This usage is compatible with some ways of
using “proof ” is used to refer to a reasoning process or a social discourse (see
below).
Proof as a Process
“Proof ” can refer to a psychological process of reasoning, or to a social process, a
certain kind of discourse. In both cases, (as with proof as an object) what process is
being referred to can be determined both by the form of the process, and by its
function.
Deductive reasoning. “In fact, ‘proof’ is just ‘reasoning’, but careful, critical
reasoning looking closely for gaps and exceptions” (Hersh, 2009, p. 19). When
“proof ” refers to deductive reasoning it is being used to refer to a psychological
process which takes on a certain form. Because psychological processes are not
directly observable, specifying this form is difficult. It can be loosely described as
a chain or tree of connected statements beginning from some that are taken as true
30

via free access
and proceeding to a conclusion according to a few logical rules (modus ponens,

etc.) occasionally supplemented by special rules unique to mathematics (e.g., the
principle of mathematical induction; see Chapter 6 for details). Some authors (e.g.,
Reid 1995b) use the verb form “proving” to refer to reasoning deductively, but use
the noun form “proof ” in another way.
Reasoning for a purpose. For some researchers, “proof ” refers to a reasoning

process, but not necessarily a deductive one. For Harel and Sowder, for example,
“the emphasis is on the student’s thinking rather than on what he or she writes”
(Harel & Sowder, 1998, p. 276) but it is not the nature of the reasoning process that
is important, but instead the function that it serves.
By “proving” we mean the process employed by an individual to remove or
create doubts about the truth of an observation. (p. 241)
In Harel and Sowder’s case, that function is verification of the truth of a statement.
They are careful to distinguish between “proving”, “proof ” and what they call
“proof schemes” but all are related to the purpose of verifying. “Proving” is a
mental act, “the process of removing or instilling doubts about an assertion”
(Harel, 2007, p. 65). “Proof ” is “a particular statement one offers to ascertain for
oneself or convince others” (p. 66), a proof-text. A “proof scheme” is a characteristic
“way of thinking associated with the proving act” (p. 66).
Another function of proof as a process of reasoning is understanding. Raman
seems to use “proof ” in this way, when referring to “the private aspect of proof”:
I distinguish between a private and a public aspect of proof, the private being
that which engenders understanding and provides a sense of why a claim is
true. The public aspect is the formal argument with sufficient rigor for a
particular mathematical setting which gives a sense that the claim is true.
(Raman, 2002, p. 3; see also 2003, p. 320)
Raman’s “public aspect of proof” seems to refer to proof as an object, either proof-
texts or convincing arguments.
Discourse defined by function. Knipping (2004, p. 73) discusses “collective proving

processes” or “argumentations” which are “collective processes in which students
and teacher develop the proof together.” The “collective proving process” she
refers to are embedded in a “proving discourse”. This discourse is a social process
whose function is producing reasons for the truth of a statement.
Balacheff (1988b) seems to use “proof ” in the same sense when he writes:
Le passage de l’explication à la preuve fait référence à un processus social
par lequel un discours assurant la validité d’une proposition change de statut
en étant acceptée par une communauté [The transition from explanation to
proof refers to a social process by which a discourse asserting the truth of a
proposition, gives it the status of being accepted by the community.] (p. 29)
Consider, for example, a class attempting to decide the truth of the proposition
“The centre of gravity of a triangle is at the intersection of the medians”. Conjectures
31

via free access
CHAPTER 2
are voiced and reasons given. Through a process of social negotiation (probably
guided in significant ways by the teacher) an argument is produced that verifies the
truth of the proposition. Note that the criteria for accepting an argument depend on
the class (including the teacher) or more generally on the community. Arguments
might be accepted by some community that would be rejected by others.
Discourse defined by form: Deductive discourse. When deductive reasoning is

expressed in a social context it becomes a method of arguing. Above we noted that
the kind of arguments accepted by a community is community dependent. Some
communities, notably communities of mathematicians, insist on a deductive basis
for acceptable arguments. In such communities if “proving” refers to a collective
process, it will only be used to describe processes with a deductive basis.
To return to the example of the centre of gravity of a triangle, a possible argument
would make use of cut-out triangles of various sizes and angles and empirical
testing of the locations of their centres of gravity. Such an approach might be
acceptable in a physics classroom, but not in a mathematics classroom. In the
mathematics classroom the argument would have to proceed deductively, perhaps
by establishing that each median divides the triangle into two equal areas, that the
three medians meet in a single point, and finally that any line through this point
will divide the triangle into equal areas (However, see Proof 10 in Chapter 6 for a
proof that bridges these two contexts).
“DEMONSTRATION” AND “PROOF ” IN OTHER LANGUAGES
As the research literature on proof in mathematics education includes significant

contributions in languages other than English it is worth being aware of some
issues related to word usage in other languages.
In older English language texts on proof and proving in mathematics education
(e.g., Fawcett, 1938) one encounters the word “demonstration” used to refer to mathe-
matical proofs. This word is now rarely used in this sense, more often being used to
refer to a political protest or a presentation intended to show how something works.
In Romance languages, however, cognate words (e.g., démonstration, dimostrazione,
demostración) continue to be used, and Balacheff, for example, makes a distinction
between “démonstration” and “preuve” (the French cognate of “proof ”).
Nous appelons preuve une explication acceptée par une communauté donnée
à un moment donné. ... Au sien de la communauté mathématique ne peuvent
être acceptées pour preuve que des explications adoptant une forme particulière.
Elles sont une suite d’énoncés organisée suivant des règles déterminées: un
énoncé est connu comme étant vrai ou bien est déduit de ceux qui le précèdent
à l’aide d’une règle de déduction prise dans un ensemble de règles bien
défini. Nous appelons démonstrations ces preuves. (1987, p. 148)
[We call proof an explanation accepted by a given community at a given
moment... Within the mathematical community only explanations adopting a
particular form can be accepted as proofs. They are an organized succession
32

via free access
of statements following specified rules: a statement is known to be true or is

deduced from those which precede it using a deductive rule taken from a well
defined set of rules. We call such proofs “démonstrations”. ]
When writing in English Balacheff attempts to preserve this distinction by translating
“démonstration” as “mathematical proof”.
By “proof ” we mean a discourse whose aim is to establish the truth of a
conjecture (in French: Preuve), not necessarily a mathematical proof (in
French: Démonstration) (Balacheff, 1991b, p. 109, Note 2)
But most English writers do not use “proof ” and “mathematical proof” in the same
way as Balacheff does, and many authors writing in French, Italian and Spanish
do not make the same distinction between “preuve”, “prova” and “prueba” and
“démonstration”, “dimostrazione”, and “demostración”.
In German the situation is like that of English. “Proof ” can be translated as
“Beweis” and the German word “Demonstration” means approximately the same
thing as the English word “demonstration”.
SUMMARY
The words “proof” and “proving” can be used in a number of ways, even in an
academic discipline like mathematics education where the exact meanings of these
words would seem to be important. As Herbst and Balacheff (2009) note, ignoring
these multiple usages can lead to a deadlock in efforts to communicate. But the
answer is not to insist on one “correct” usage.
If the field is in a deadlock as regards to what we mean by “proof,” we
contend this is so partly because of the insistence on a comprehensive notion
of proof that can serve as referent for every use of the word. ... We have
argued that to make it operational for understanding and appraising the mathe-
matics of classrooms we need at least three meanings for the word. (p. 62)
We have identified a number of usages in this chapter:
– A concept of proof
– Proof-texts
– Convincing arguments
– Deductive reasoning
– Personal verification
– Personal understanding
– A social discourse to verify
– A deductive social discourse
These usages are not disjoint categories, nor does a researcher’s use of “proof ” or
“proving” in one way in one context guarantee that her or his next usage will be the
same. However, being aware that there are different usages is an important step to
being able to decipher mathematics education research.
In our writing we will attempt to use more precise words to say what we
mean, reserving the word “proof ” primarily to refer to a concept. However, to
avoid unnecessary repetitions we may use “proof ” and “proving” in one of their
33

via free access
CHAPTER 2
other senses when the meaning is clear. Similarly, when quoting others we will
clarify how these words are being used if possible and necessary. If the meaning is
sufficiently clear from the context, or if the meaning is so unclear we cannot
determine what it is, we will not attempt to suggest how the author is using “proof ”
and “proving”.
Word usages can also offer important hints towards larger issues. In the next
chapter we will use three of these usages, proof-texts, reasoning, and discourse, to
distinguish between theoretical perspectives in mathematics education research.
34

via free access
CHAPTER 3
RESEARCHER PERSPECTIVES
There are a number of substantial ways in which researchers in mathematics

education differ in their perspectives on learning, teaching, proof, and proving, which
result in their pursuing distinct research agendas, and in some cases arriving at
contradictory results.
For example, perspectives can differ in the relationship between mathematical
proof and empirical observations. This can lead to contradictory assertions that
proof is completely autonomous of empirical argumentation, or that proof is a guide
to empirical exploration:
A geometrical fact, a theorem ... is acceptable only because it is systematised
within a theory, with a complete autonomy from any verification or argumenta-
tion at an empirical level. (Mariotti, 1997, section 1.3)
Mathematical proof should not be seen as a turning way from observation
and measurement, but rather ... it should be seen as a guide to the intelligent
exploration of phenomena. (Hanna & Jahnke, 2002b, p. 38)
Perspectives can also differ in the relative valuing of the form of a proof and the
meaning of its contents. These can range from an extreme of basing proof entirely
on form, through an intermediate position that states that proofs based on form alone
are less convincing, to another extreme which focusses on proofs that are indepen-
dent of form:
La forme d’une proposition est sa valeur de vérité. Dire qu’un raisonnement
ne dépend que de sa forme, revient donc en fait à dire que les propositions
elles-mêmes ne sont pas prises en compte : on les nomme seulement comme
support pour des valeurs de vérité. [The form of a proposition is its truth
value. To say that reasoning depends only on form, amounts to saying that the
propositions themselves are not taken into account: one names them only as
support for truth values.] (Duval, 1990, p. 198)
Mathematicians freely admit that a proof may lack conviction when it is
shown to be valid by virtue of its form alone, without regard to its content.
(Hanna, 1990, p. 8)
We mean by a preformal proof a chain of correct, but not formally represented
conclusions which refer to valid, non-formal premises. (Blum & Kirsch, 1991,
p. 187)
This diversity of perspectives can be confusing to a reader examining the research
literature on proof and proving, and has implications for teaching proof as
35

via free access
CHAPTER 3
different perspectives lead to different recommendations. In this chapter we will

explore this diversity, describing a number of researcher perspectives and offer-
ing examples of each. We will also attempt to identify researchers with similar
perspectives as an aid to the interpretation of the research results we will discuss
in Part 2.
– Which of these perspectives feels “right” to you?
– Are there common threads that run through all researcher’ perspectives?
– How do researchers whose work you are familiar with fit onto these perspectives?
Do some seem to require a broadening or elaboration of the descriptions?
– How can being aware of a researchers’ perspective change your reading of their
research?
We have selected three dimensions in which researcher perspectives can be differ-
entiated (see Figure 3). We do not claim this is the only possible approach; in fact
we will describe an alternative approach at the end of this chapter. However, in
examining the work of leading researchers in the field, especially those who make
their perspective clear, we have found these three dimensions to be useful in differen-
tiating perspectives and describing key features of them.
One dimension is the philosophy of mathematics assumed. There are many ways
to categorise and describe philosophies of mathematics. Perhaps the most well
known (including Platonism, intuitionism and formalism) focusses on the ontological
status of the objects of study in mathematics. As we are focussing here on proof
we will concentrate instead on philosophies of mathematics that consider the truth
assumed for the premises of proofs and the degree of validity and fallibility assumed
for the rules of inference employed. For example, philosophies might differ in
considering the premises (axioms, postulates, etc.) to be arbitrary or a reflection on
some underlying truth in nature of the mind. Similarly the rules of inference
allowed might be seen as arbitrary or necessary. In the next section we will
describe four of these philosophies.
Figure 3. Three dimensions of description for researcher perspectives: Meaning

of “proof ”, Philosophy, Breadth of category “proof ”.
36

via free access
Another dimension is the meaning given to the words “proof ” and “proving”. Of
the possible meanings we mentioned in Chapter 2, we will focus here on proofs as
proof-texts, proving as a process of reasoning, and proving as a social discourse.
We have found these three meanings to differentiate usefully between perspectives.
Our final descriptive dimension is the narrowness or the breadth of the category
“proof ” within the researchers’ perspective. A narrow view of proof requires that
all proofs have three characteristics: they must be deductive, convincing, and at
least semi-formal. A more broad view relaxes one of these requirements, and a very
broad view relaxes two of them. For example, a researcher who considers proofs
that are deductive and convincing, whatever their degree of formality, has a broad
category for proof, and a researcher who considers any convincing argument a
proof has a very broad category. We should note that if researchers are interested in
mathematical proof (viewed narrowly) as a part of a more broad category of
proofs, then we consider them to have a broad view.
PHILOSOPHIES OF MATHEMATICS
At one extreme among the philosophies of mathematics we will consider here

stands a claim that the axioms and definitions of mathematics refer to real objects
in the world, that have the properties claimed for them, and that can have no other
properties. This is coupled with a claim that deductive rules of inference transmit
the truth of the axioms and definitions to the conclusions deduced from them. We
call this philosophical position “a priorism” in reference to the work of Kant,
who asserted the “apodictic [established on incontrovertible evidence] certainty of
all geometric principles, and the possibility of their construction a priori.” (Kant,
1781/1927, p. 19).
As noted in Chapter 1, after the discovery of non-Euclidean geometries, most
mathematicians dropped the a priorist claim that the axioms and definitions of
mathematics refer to real objects in the world and shifted instead to an “infallibilist”
philosophy in which axioms are only hypotheses, but the claim that the rules of
inference preserve truth is retained.
The deductive method provides the warrant for the assertion of mathematical
knowledge. The grounds for claiming that mathematics (and logic) provide
absolutely certain knowledge, that is truth, are therefore as follows. First of all,
the basic statements used in proofs are taken to be true. Mathematical axioms
are assumed to be true, for the purposes of developing that system under
consideration, mathematical definitions are true by fiat, and logical axioms are
accepted as true. Secondly the logical rules of inference preserve truth, that is
they allow nothing but truths to be deduced from truths. On the basis of these
two facts, every statement in a deductive proof, including its conclusion, is true.
Thus, since mathematical theorems are all established by means of deductive
proofs, they are all certain truths. (Ernest, 1991, pp. 7–8; see also 1998, p. 14)
Ernest calls this philosophy of mathematics “absolutist” but we feel his labelling
could be misleading. The claim that mathematics provides absolutely certain know-
ledge is a limited claim, limited to asserting that the conclusions of proofs are no
37

via free access
CHAPTER 3
less true than the premises, without making a claim that the truth of the premises is
anything more than an assumption. This view asserts only that “the basic statements
used in proofs are taken to be true” which is different than stating that “the basic
statements used in proofs are true” (as in a priorism). The claim is not that
theorems are absolutely true, but only that theorems are absolutely true within the
framework of an axiom system.
There are several philosophies of mathematics that deny the infallibilist claim
that a proof guarantees the truth of its conclusion, presuming its premises are true.
Most of these can be traced back to the work of Imre Lakatos.
Lakatos first presented his historical analysis of the role of proof in mathematics
in his doctoral thesis (1961), then in a series of articles in the British Journal for
Philosophy of Science (1963–64). After his death Worrall and Zahar edited his
articles and published them in book form as Proofs and Refutations (Lakatos, 1976),
which was translated into French by Balacheff and Laborde (Lakatos, 1984). Through
these publications Lakatos has influenced the thinking of many researchers in mathe-
matics education.
Lakatos argues against an infalliblist position and in favour of a more sceptical
position that is called “fallibilist” or “quasi-empiricist”.
For more than two thousand years there has been an argument between dog-
matists and sceptics. The dogmatists hold that – by the power of our human
intellect and/or senses – we can attain truth and know that we have attained it.
The sceptics on the other hand either hold that we cannot attain truth at all
(unless with the help of mystical experience), or that we cannot know if we
can attain it or that we have attained it. (Lakatos, 1976, pp. 4–5)
Lakatos’s scepticism goes beyond that of the infallibilists. He claims that not only
are the truths of mathematics not a priori certainties about the world, they are also
not certainties within an axiom system, because the axiom systems used in mathe-
matics are not fixed. The deductive method does not transmit truth (or assumed
truth) from the assumptions to the conclusions. Rather it retransmits falsity from
refuted theorems back to the axioms and definitions on which they are based (see
Chapter 2). This changes the role of proofs.
Proofs, even though they may not prove, certainly do help to improve our
conjecture. ... Our method improves by proving. This intrinsic unity between
the ‘logic of discovery’ and the ‘logic of justification’ is the most important
aspect of the method... (p. 37)
According to Lakatos, proof is part of proof-analysis, a cycle of proofs and refu-
tations. Each refutation or counterexample to a theorem (thought to have been
established by a proof ) opens a phase of criticism of the theorem, its proof, and the
definitions and assumptions on which they are based.
Informal, quasi-empirical, mathematics does not grow through a monotonous
increase of the number of indubitably established theorems but through the
incessant improvement of guesses by speculation and criticism, by the logic
of proofs and refutations. (p. 5)
38

via free access
The result of this criticism is an improved theorem, definition, postulate and/or

proof.
‘Certainty is never achieved’; ‘foundations’ are never found – but the ‘cunning
of reason’ turns each increase in rigour into an increase in content, in the
scope of mathematics. (p. 56)
Lakatos still believes in the “cunning of reason”, that the deductive method
accurately retransmits falsity. The axioms and definitions are not fixed and can be
doubted and re-negotiated, but the deductive method itself is infallible. In fact, this
is a requirement of Lakatos’s method of proof-analysis by proofs and refutations.
Lakatos’s belief that the deductive method itself is infallible can be questioned,
and is questioned by those who hold a “social-constructivist” philosophy of mathe-
matics. They would assert that the deductive method is a social construction, and
what counts as valid deductive argument varies from community to community. The
standards of present day mathematicians differ from the standards of mathematic-
cians of the past, and variations occur even between sub-specialities of mathematics.
This philosophy of mathematics was named by Ernest (1991) although one can see
evidence of it in earlier writings.
RESEARCH BASED ON AN A PRIORIST PHILOSOPHY OF MATHEMATICS
It seems unlikely that researchers in mathematics education would include an a

priorist philosophy of mathematics within their research perspective, but occasionally
one wonders. For example, Fischbein and Kedem (1982) state that “a formal proof
of a mathematical statement confers on it the attribute of a priori universal validity”
(p. 128) and they concluded that because students were not completely convinced
by written proofs, they did not understand proof.
This perspective also includes a narrow category of proof requiring that proofs
convince, that they be deductive and that they be at least semi-formal. From the
statement above the importance of establishing certainty is clear. Elsewhere Fischbein
(1982) discusses proof in more detail, and notes the necessity of deductive reasoning.
With reference to mathematics the way of proving is different: the statement
we consider must be the logical, necessary conclusion of some other previously
accepted statements. ... The universality of the truth expressed by the theorem
is guaranteed by the universal validity of the logical rules used in the proof.
(p. 15)
His second sentence repeats the claim that proofs confer universal validity on
theorems, but as reference is made only to logical rules and not to axioms that are
also a priori true, perhaps this perspective belongs to infallibilism rather than a
priorism. From Fischbein and Kedem’s research (in which they asked students to
comment on written proofs) it seems that “proof ” means proof-texts, which must
be semi-formal. However, Fischbein is also aware of the importance of preformal
presentations, as an aid to developing the “intuition” he believes underlies under-
standing of proof.
39

via free access
CHAPTER 3
Figure 4. Diagram from Fischbein, 1982, p. 18.
A new “basis of belief ”, a new intuitive approach, must be elaborated which

will enable the pupil not only to understand a formal proof but also to believe
(fully, sympathetically, intuitively) in the a priori universality of the theorem
guaranteed by the respective proof. (Fischbein, 1982, p. 17)
Fischbein suggests that an argument like the following for the sum of the angles of
a triangle being equal to two right angles (or 180˚) can provide the basis for the
development of an intuition of necessity.
Let us now draw a segment AB and the perpendiculars MA and NB to the
segment. The angles MAB and NBA are right angles. We can “create” a triangle
by “inclining” MA and NB. So, it can be seen that the angle APB “accumu-
lates” what is “lost” by the angles MAB and NBA when “inclining” MA and
NB. [Figure 4]
Of course this is not mathematical language. It is rather a story about lines
and angles, but a story which can catch the spirit, which can impose itself as
intrinsically true. The same story can be translated into the form of a mathe-
matical proof. Consequently the formal necessity and the intrinsic necessity
will coincide. (p. 17)
While Fischbein values this argument for the development of the intuition of necessity,
he does not count it as a proof, as he requires a greater level of formality.
RESEARCH BASED ON AN INFALLIBILIST PHILOSOPHY OF MATHEMATICS
As we noted above, the key features of an infallibilist philosophy of mathematics

are that: “logical axioms are accepted as true [and they] preserve truth; that is, they
allow nothing but truths to be deduced from truths” (Ernest, 1991, pp. 7–8; 1998,
p. 14). In this section we will describe three research perspectives that incorporate
this infallibilist philosophy; those of Fawcett, the preformalists, and Duval.
Fawcett’s Research Perspective

We will begin by considering one of the first researchers to make a thorough study
of the teaching of proof. Harold Fawcett (1938) reports the results of a teaching
40

via free access
experiment focused on proof in the thirteenth yearbook of the National Council of

Teachers of Mathematics [NCTM].
Fawcett’s understanding of the nature of proof can be inferred from what he
says about students’ understanding of proof:
For purposes of this study it is assumed that a pupil understands the nature of
deductive proof when he understands:
1. The place and significance of undefined concepts in proving any conclusion.
2. The necessity of clearly defined terms and their effect on the conclusion.
3. The necessity of assumptions or unproved propositions.
4. That no demonstration proves anything that is not implied by the assump-
tions. (p. 10)
It is further assumed that a pupil who understands these things will also
understand that the conclusions thus established can have universal validity
only if the definitions and assumptions which imply these conclusions have
universal validity. The conclusions are “true” only to the extent that the funda-
mental bases from which they were derived are “true.” Truth is relative and
not absolute. (p. 11)
Fawcett’s truth is “relative and not absolute” so it fits better with infalliblism than a
priorism. Most importantly, Fawcett’s view includes a belief that the truth of the
definitions and assumptions is transmitted infallibly to the conclusions through the
logical rules of inference.
Fawcett also believes that the main reason to teach mathematical proof in schools
is as preparation to apply the infallible methods of deductive reasoning not only in
mathematics, but also in other areas.
The concept of proof is one concerning which the pupil should have a growing
and increasing understanding. It is a concept which not only pervades his
work in mathematics but is also involved in all situations where conclusions
are to be reached and decisions to be made. Mathematics has a unique
contribution to make in the development of this concept (p. 120)
These quotations suggest that Fawcett has a narrow view of proof. He expects
proofs to involve the three characteristics of being deductive, being convincing and
being at least semi-formal. His focus on proofs as a model for deductive reasoning
and on the use of proving to convince others (“in all situations where conclusions
are to be reached and decisions to be made”) makes the first two characteristics
clear. And the stress placed on explicit definitions in the teaching he describes
reveals a desire for a certain level of formality.
To determine whether Fawcett thinks of proof in terms of proof-texts, deductive
reasoning processes or social discourses it is necessary to reflect on the nature of
the teaching he describes. Consider, for example, how the class proved that the
angle sum of a triangle is 180˚. They began by examining the diagram shown in
Figure 5. The teacher asked them what happens to angle p as line h' rotates about P.
The students quickly came to the conclusion that angle p is decreased by the angle
41

via free access
CHAPTER 3
Figure 5. Diagram that formed the basis for the triangle angle sum
proof in Fawcett, 1938.
of rotation p' and that the new angle t formed by the intersection of h'' with h has
the same measure as p'. This led to the conclusion that the sum of the measures of
angles r, RPT and t is 180˚.
After a further study of this figure, including the changes that occur as h' makes
one complete revolution about P, the pupils were asked to draw any triangle
and see what they could discover about the sum of the angles of that particular
triangle. All of them felt that the sum was 180˚ but no one knew, at first, just
how to proceed to demonstrate this fact. However, after thoughtful study
followed by a suggestion from the teacher that reference to the diagram of the
preceding discussion might prove helpful, there was increasing evidence from
all parts of the room that discoveries were being made, and before the class
period was over seventeen of the twenty-four pupils present had worked out
an “acceptable” proof for the theorem concerning the sum of the angles of a
triangle. (p. 60)
The students’ proofs resembled Proof 22 (see Chapter 7) except that only four
students specified that the line drawn be parallel to one side of the triangle, and
only one specified that it should pass through a vertex. A discussion of the
weaknesses of the proofs ensued.
These proofs are called “acceptable” because in each of them the basic
mathematical ideas are acceptable. Their weakness lies in lack of precise and
accurate statement. All of them, however, served as excellent illustrations of
the way in which unrecognized assumptions can creep into one’s thinking and
definitely affect conclusions. (pp. 60–61)
The students wanted to make their implicit assumptions explicit, and their need
to provide a justification for the construction of the auxiliary line through a
vertex and parallel to a side brought them to propose more and more precise
formulations of the postulate “Through a given point not on a given line it is
possible to draw one and only one line parallel to the given line”. The students were
told of the importance of the postulate in Euclid’s Elements after they had disco-
vered the need for it and formulated it precisely for themselves. The wording of
the theorem itself was also discussed and after the formulation “The sum of the
interior angles of any triangle is 180˚” was agreed on the students entered it into
their notebooks.
42

via free access
Fawcett does not include a proof-text in his description of this, or any other,
classroom activity. This suggests that he does not see proofs as primarily written
texts. As in this case, the classroom activities he reports are mostly discussions,
although there was also time for individual work. Both individual deductive reasoning
processes and social discourses are mentioned as being of importance. In Fawcett’s
case it is perhaps not possible to decide which of these is of the greatest importance,
however, if we remember that for Fawcett the transfer of the “concept of proof” to
other situations outside of mathematics is of great value, and that such situations
are likely to be ones of social discourse, claiming that Fawcett saw proof mainly as
social discourse seems reasonable.
The “Preformalist” Research Perspective

Blum and Kirsch (1979, 1989, 1991) argue for the use of “preformal” or “inhaltlich-
anschaulich” proving in schools, within which they include the “action proofs”
advocated by Semadeni (1984). These researchers, along with Wittmann and Müller
(1988, 1990), share a common research perspective that we call “preformalist” and
which can be traced back to Branford (1908), who advocated the use of “intuitional”
proofs.
According to Branford:
[An intuitional proof] establishes general and accurate truths, but appeals
implicitly to postulates of sense-experience whenever necessary: founds the
truth on an independent basis of its own by direct appeal to first principles
(p. 97)
This research perspective is based on an infallibilist (or perhaps in some cases a
priorist) philosophy of mathematics. Branford states that an intuitional proof
“establishes general and accurate truths” (p. 97) and Semadeni believes that “action
proofs ... are valid mathematical proofs, that is, provide absolute certainty” (1984,
p. 34). Blum and Kirsch use “preformal proof” to mean “a chain of correct, but not
formally represented conclusions, which refer to valid, non-formal premises” (1991,
p. 187, original italics removed).
While the preformalists share the infallibilist philosophy of mathematics with
Fawcett, they focus on proof-texts rather than discourse, and have a broader category
of proof, as we will see below.
Blum and Kirsch (1991) and Semadeni (1984) provide several texts as examples
of preformal or action proofs. This suggests that they use “proof ” and “proving” to
refer to proof-texts. The fact that their focus is on proofs that can be shown by a
teacher to students also suggests that they are referring to proof-texts, although in
the case of action proofs, the idea of “texts” must be interpreted broadly to include
physical actions on objects.
As the name “preformalist” suggests, the preformalists relax the requirement
that proofs be at least semi-formal. However they retain expectations that deductive
reasoning be used and that proofs should be convincing. Because preformal proofs
“if formalised, ... have to correspond to correct formal-mathematical arguments”
43

via free access
CHAPTER 3
Proof 5: Proof that the sum of the interior angles of a triangle is 180 degrees
Hoyles, 1997, p. 12
(Blum & Kirsch, 1991, p. 187), that is to say they can be written as semi-formal
proofs, it is clear that they must involve deductive reasoning. And one would expect
that proofs that are based on intuitive or obvious bases would be just as convincing
(if not more so) as traditional formal proofs. “One may think of an action proof as an
idealised, simplified version of a recommended way in which children can convince
themselves of the validity of a statement” (Semadeni, 1984, p. 32).
Proof 5 is an example of a preformal proof that the angle sum of a triangle is 180˚.
It describes a physical action that could be done, or at least imagined, and makes
reference “to postulates of sense-experience” (Branford, 1908, p. 97). The line “If
you walk all the way around the edge of the triangle you end up facing the way you
began” makes reference to the sense experience of walking around an object.
Duval’s Research Perspective

Raymond Duval’s research focusses on the distinction between argumentation and
proof. From his perspective, argumentation is directed towards convincing, while
proof is associated with deductive reasoning, establishing truth, and writing proof-
texts.
Surgissant dans toute situation d’interaction sociale où il faut persuader un
interlocuteur ou réfuter une thèse, elle [argumentation] est un raisonnement
ordonné à des fins de communication. Aussi se trouve-t-elle toujours opposée
aux démarches de démonstration et au raisonnement déductif : « la logique de
l’argumentation ne peut être que non formelle » (J. M. Borel, 1983, p. 20).
Mais cette opposition ne va pas sans ambiguïté sur le caractère de raisonnement
propre à l’argumentation. D’une part, puisqu’elle subordonne les questions de
validité aux stratégies d’action sur les représentations d’un interlocuteur,
l’argumentation n’apparaît pas comme un raisonnement véritable : sa portée
s’y trouve limitée au probable ou au vraisemblable. (1990, p. 195)
[Emerging in any situation of social interaction where it is necessary to
persuade an interlocutor or to refute a thesis, argumentation is reasoning
oriented to the goal of communication. Also it is always opposed to methods
of mathematical proof and to deductive reasoning: “the logic of argumentation
44

via free access
can be only nonformal” (J. M. Borel, 1983, p. 20). But this opposition does
not remove the ambiguity concerning the character of reasoning suitable for
argumentation. On the one hand, since it subordinates questions of validity to
strategies of action about representations of an interlocutor, argumentation
does not seem to be certain reasoning: its range is limited to the probable or
the plausible.]
This passage allows us to see that Duval’s perspective is based on an infallibilist
philosophy of mathematics. Specifically, the description: “l’argumentation n’apparaît
pas comme un raisonnement véritable : sa portée s’y trouve limitée au probable ou
au vraisemblable” suggests that its opposite, mathematical proof and deductive
reasoning, does provide certainty. His position is not a priorist because, as we will
see, he separates the form and content of propositions and considers mathematical
proof to operate only on the form. The a priorist position claims that mathematical
propositions state truths about the real world, which must mean that their content is
essential to their truth value, which Duval would deny.
Duval’s perspective involves a narrow view of proof. The form of a proof is
very important to Duval, as he considers form as being the sole determinant of truth,
and logical, deductive, relations between propositions as the essence of proofs.
En effet, on ne reconnaît aux propositions que deux aspects: leur contenu, ou
leur sens, et leur valeur de vérité. En définissant les relations logiques entre
les propositions comme des fonctions de vérité de ces propositions, on
élimine leur contenu. Dans cette perspective les propositions ne se distin-
guent réellement que par leur valeur de vérité. La forme d’une proposition est
sa valeur de vérité. Dire qu’un raisonnement ne dépend que de sa forme,
revient donc en fait à dire que les propositions elles-mêmes ne sont pas prises
en compte : on les nomme seulement comme support pour des valeurs de
vérité. (pp. 197–198)
[Indeed, one recognizes in propositions only two aspects: their content or
their meaning, and their truth value. By defining logical relations between
propositions as truth functions of these propositions, one eliminates their
content. From this point of view propositions are characterized only by
their truth value. The form of a proposition is its truth value. To say that
reasoning depends only on form, amounts to saying that the propositions
themselves are not taken into account: one names them only as support for
truth values.]
Although Duval distinguishes between argumentation (aimed at convincing) and
proof (based on deductive reasoning), his view of proof also includes the expectation
that proofs should be convincing. One could argue that because Duval considers
proof in the broader context of comparison with argumentation he has a broader
view than if he had ignored argumentation entirely, which is true, but his focus on
separating arguments from proof leads us to consider his view of proof as narrow.
For Duval to accept a text as a proof it must have all three characteristics, being at
least semi-formal, deductive and convincing.
45

via free access
CHAPTER 3
While Duval often refers to reasoning, in his discussions he describes proof-texts

more than reasoning as a cognitive process. This is a consequence of his equating
deductive reasoning with the manipulation of the forms of propositions.
Rappelons que celui-ci [raisonnement déductif ] consiste en une opération de
substitution, portant sur des propositions, dans laquelle celles-ci interviennent
d’abord selon leur statut opératoire et non directement selon leur contenu.
(p. 204)
[Let us recall that deductive reasoning consists of an operation of substitution,
applied to propositions, in which they are involved primarily according to their
operational status and not directly according to their content.]
Hence, we see Duval as considering “proof ” to mean a proof-text. Both Duval’s
emphasis on substitution in the quote above and the removal of content in favour of
form is reminiscent of the formalist project (see Chapter 1) which is perhaps not
surprising given that Duval is working in a time and place much influenced by
Bourbaki. Proof 22 (in Chapter 7) is an example of the kind of proof-text Duval
would consider a proof.
RESEARCH BASED ON A QUASI-EMPIRICIST PHILOSOPHY OF MATHEMATICS
In this section we will describe a research perspective incorporating a quasi-

empirical philosophy of mathematics, which stay very close to Lakatos’s original
conception. In the next section we will consider social-constructivist perspectives
that extend Lakatos’s ideas.
Balacheff’s Research Perspective (pre-1990)

As Nicolas Balacheff was one of the translators of Preuves et réfutations, it should
come as no surprise that Lakatos’s ideas are an important element in Balacheff’s
research perspective. But research perspectives change over time. Prior to about
1990 Balacheff’s research perspective seems to have included a quasi-empirical
philosophy of mathematics quite close to that outlined in Lakatos’s work. After
1990 his philosophy is more social-constructivist. Hence we will consider
Balacheff’s research perspective twice, once in this section and once in the next.
In the 1980s Balacheff was quite explicit about the importance of Lakatos’s
ideas in his work.
La thèse développée par Lakatos est que la pratique des mathématiques,
l’activité des mathématiciens et donc les processus qui soustendent la décou-
verte en mathématique, relèvent d’une dialectique preuve-réfutation. Il décrit
un modèle des rapports entre la génération de preuves et leur contradiction à
l’aide de contre-exemples qui va au-delà de la dualité vrai/faux.
C’est en nous plaçant dans ce cadre que nous analysons les conduites observées
chez les élèves. (1982, p. 277)
46

via free access
[The thesis developed by Lakatos is that the practice of mathematics, the

activity of mathematicians and thus the processes which supports discovery
in mathematics, involves a dialectic of proofs and refutations. It describes a
model of the relationship between the generation of evidence and contra-
dictions using counterexamples which goes beyond the true/false duality. It is
from within this framework that we analyse the observed conduct of students.]
He also notes that there is an underlying rationality, the deductive method, involved
in Lakatos’s cycle of proofs and refutations.
Les mathématiciens auxquels s’intéresse Lakatos semblent bien tous adhérer
au même fond de rationalité. [The mathematicians in which Lakatos is
interested all seem to stick to the same basic rationality.] (1987, p. 167)
Balacheff’s work at this time involved developing teaching situations in which
students encountered conflicts between their conjectures and alternate conjectures
or counterexamples, and observing the students’ responses to these situations. He
did not expect that the students’ arguments would meet the standards of professional
mathematics, but that encountering counterexamples and refutations would improve
their arguments, as Lakatos had suggested counterexamples and refutations improve
the arguments of mathematicians.
The students were always involved in communicating their explanations to
others and responding to questions and criticisms. This suggests that for Balacheff,
at this time, proof was a mode of discourse.
Le passage de l’explication à la preuve fait référence à un processus social
par lequel un discours assurant la validité d’une proposition change de statut
en étant acceptée par une communauté [The transition from explanation to
proof refers to a social process by which a discourse asserting the truth of a
proposition, gives it the status of being accepted by the community.] (1988b,
p. 29)
Balacheff had a broad view of proof. His main interest was in proof as convincing
argument, including proofs that were informal or not deductive. Within this broad
view are included mathematical proofs (démonstrations) which are distinguished
from other proofs by their form and deductive character.
Nous appelons preuve une explication acceptée par une communauté donnée
à un moment donné. ... Au sien de la communauté mathématique ne peuvent
être acceptées pour preuve que des explications adoptant une forme parti-
culière. Elles sont une suite d’énoncés organisée suivant des règles déterminées:
un énoncé est connu comme étant vrai ou bien est déduit de ceux qui le
précèdent à l’aide d’une règle de déduction prise dans un ensemble de règles
bien défini. Nous appelons démonstrations ces preuves. (1987, p. 148)
[We call proof an explanation accepted by a given community at a given
moment... Within the mathematical community only explanations adopting a
particular form can be accepted as proofs. They are an organized succession
47

via free access
CHAPTER 3
of statements following specified rules: a statement is known to be true or is

deduced from those which precede it using a deductive rule taken from a well
defined set of rules. We call such proofs “démonstrations”. ]
As proof for Balacheff is a discourse, he would probably see the proving in
Fawcett’s class (described above) as acceptable from his perspective. In fact, the
process of identifying and refining assumptions fits very well into a quasi-empiricist
perspective, and if Fawcett had lived at another time he might have had much in
common with Balacheff.
RESEARCH BASED ON A SOCIAL-CONSTRUCTIVIST PHILOSOPHY

OF MATHEMATICS
Recall that a social-constructivist philosophy of mathematics differs from quasi-

empiricism in that it calls into question the special place of the deductive method.
Social-constructivists assert that the deductive method is a social construction, and
what counts as valid deductive argument varies from community to community
(Ernest, 1991).
Balacheff’s Research Perspective (post-1990)

In Balacheff’s (1991a) reflections on his teaching experiments, we can perhaps see
the beginnings of a shift in a researcher’s perspective. Balacheff notes that, in
planning his experiments, “we thought that genuine mathematical proving processes
would be observed” (p. 181). Above we suggested that this expectation was a
reasonable one based on a quasi-empirical philosophy of mathematics. However,
Balacheff found that for the most part “genuine mathematical proving processes”
were not observed when the students encountered counterexamples or alternative
arguments. Instead the students made use of a wide range of methods to persuade
their peers that some arguments were better than others, almost never making reference
to aspects of the arguments that mathematicians would feel were relevant.
In Balacheff’s post-1990 papers we see evidence of a more social-constructivist
philosophy, which stresses the negotiation of the rules of argument. It differs from
the infallibilist and quasi-empiricist philosophies in that it does not assert that
because the deductive method permits the infallible transmission of truth or retrans-
mission of falsity, it will emerge naturally in situations of doubt or contradiction.
So, in order to teach mathematical proof successfully, the major problem
seems to be that of how to negotiate the acceptance by the students of new
rules (p. 189)
It is important to note that these “new rules” are not taken to be pre-given, absolute
or infallible. They are part of mathematics, which is a socially constructed body of
knowledge.
As a social behaviour it [argumentation] is an open process, in other words it
allows the use of any kind of means; whereas, for mathematical proofs, we
48

via free access
have to fit the requirement for the use of some knowledge taken from a
common body of knowledge on which people (mathematicians) agree. (p. 189)
Post-1990 Balacheff continues to use “proof ” “preuve” and “démonstration” to refer
to social discourses. Mathematical proofs (démonstrations) must meet different
standards for social acceptance than other proofs, but they remain discourses.
By “proof ” we mean a discourse whose aim is to establish the truth of a
conjecture (in French: Preuve), not necessarily a mathematical proof. (in French:
Démonstration) (1991b, p. 109, Note 2)
Balacheff continues to have a broad view of proof, including any discourse that a
community finds convincing, but noting that within the mathematical community,
only semi-formal and deductive proofs are acceptable.
Mariotti’s Research Perspective

Maria Alessandra Mariotti (2006) articulates her “epistemological perspective”
which is social-constructivist. She sees the concept of proof evolving over time,
and integrating two elements: logical validity and an explanatory role. She notes
that in different historical periods one or the other of these aspects has been stronger.
A key idea in her thinking is that “it is not possible to grasp the sense of a
mathematical proof without linking it to the other two elements: A statement and
overall a theory” (p. 184). The reference theory is the framework in which the proof
makes sense, including the rules of logic, the definitions used, and the postulates
assumed. Mariotti introduced this concept of theory in a paper with her colleagues
Bartolini Bussi, Boero, Ferri and Garuti (1997), and Bartolini Bussi describes the
importance of the reference theory to proof:
A ‘rigorous’ proof exists only within a reference theory, that states what is
postulated within the classroom and which are the rules for reasoning. (2000,
para. 6)
What counts as proof depends on what postulates are accepted as true and what
rules of inference are accepted as valid. The “rules for reasoning” that are accepted
might change over time and from community to community, and the reference
theory of which they are a part must be understood for a theorem and its proof to
be understood.
In particular, as far as theorems are concerned, it is worth reminding (Mariotti
et al., 1997) that any mathematical theorem is characterised by a statement
and a proof and that the relationship between statement and proof makes
sense within a particular theoretical context, i.e. a system of shared principles
and inference rules. Historic – epistemological analysis highlights important
aspects of this complex link and shows how it has evolved over the centuries.
The fact that the reference theory often remains implicit leads one to forget or
at least to underevaluate its role in the construction of the meaning of proof.
(Mariotti, 2000, p. 29)
49

via free access
CHAPTER 3
The existence of a reference theory as a system of shared principles and

deduction rules is needed if we are to speak of proof in a mathematical sense.
Principles and deduction rules are so intimately interrelated so that what
characterises a Mathematical Theorem is the system of statement, proof and
theory. (Mariotti et al., 1997, p. 182)
This position differs from the infallibilist position in that an infallibilist would
claim that there is only one reference theory, deductive logic.
Mariotti and her colleagues are also clear about the importance of convincing
and deduction to proof. Both are involved, but deduction is essential.
A deductive approach is deeply rooted in the practice of justification (de Villiers,
1990; Hanna, 1990). Proving consists in providing both logically enchained
arguments which are referred to a particular theory, and an argumentation
which can remove doubts about the truth of a statement. This twofold meaning
of proof is unavoidable and pedagogically consistent (Hanna, 1990). (Mariotti,
2000, pp. 30–31)
It is also important, in this narrow view of proof, that proofs be semi-formal and
the formalisation of informal arguments is an important aspect of teaching proof.
Bartolini Bussi notes that “the translation of the set of arguments into a logical
chain is a matter of social construction” (2000). She describes teaching experiments
in which she has participated and how this challenge has been addressed in them.
In these experiments proofs are clearly part of the discourse of the classroom, although
a final exercise for the students is to try to produce a written text corresponding to
what went on in the discussions.
Hanna and Jahnke’s Research Perspective

Gila Hanna has contributed a great deal to research on proof, from her monograph
Rigorous Proof in Mathematics Education (1983) to her more recent work with
Hans Niels Jahnke on proof ideas from physics (e.g., Hanna & Jahnke, 1993,
2002a). Throughout she has reminded researchers in mathematics education that
discussions of proof should acknowledge the role of proof in the actual practice of
mathematicians.
Her perspective includes a social constructivist philosophy of mathematics, in
which there is not a single valid set of rules for reasoning (as an infallibilist would
claim).
It is clear that any mathematical truth arrived at through a proof or series of
proofs is contingent truth, rather than absolute truth, in the sense that its validity
hinges upon other assumed mathematical truths and rules of reasoning. (Hanna,
1996, p. 32)
She describes the system of rules of inference and assumptions on which any theorem
and proof must be based in language similar to that of Mariotti and Bartolini Bussi.
One cannot really discuss proof without discussing theories. The very notion
of proof is tied to the notion of theory (and here we use “theory,” of course,
50

via free access
in the sense of “systematic structure”). Every proof is based upon an under-

standing, explicit or implicit, of what can be taken as given and what kinds of
argument are acceptable, and these questions can be answered only in the
context of a theory. (Hanna & Jahnke, 2002a, p. 2)
In another context, and a bit earlier, her philosophy seems to have been more
Lakatosian.
Rather than transferring truth from the conditions to the proved theorem, the
proof actually has the effect of calling the conditions themselves into question
(Hanna & Jahnke, 1993, p. 427)
One of Hanna’s contributions is the distinction between formal proofs, convincing
proofs (proofs that prove), and explanatory proofs. Hanna’s view of proof is
broader than some in that she recognises that in mathematical practice there are
formal proofs that are not convincing:
Mathematicians freely admit that a proof may lack conviction when it is
shown to be valid by virtue of its form alone, without regard to its content.
(Hanna, 1990, p. 8)
And beyond proofs that are convincing there are proofs that explain as well:
Thus it is not its ability to convince which distinguishes the explanatory proof,
convincing though we would expect it to be. (p. 12)
All of these types of proofs must be deductive, in the sense of employing only the
rules of inference accepted by mathematicians.
A proof that explains and a proof that proves are both legitimate proofs. By
this I mean that both types of proofs meet the requirements for a mathematical
proof, and thus serve in equal measure to establish the validity of a mathe-
matical assertion. Both consist of statements that are either axioms themselves
or follow from previous statements (and thus eventually from axioms) as a
result of the correct application of rules of inference. (p. 9)
However, although Hanna’s description of “the requirements for a mathematical
proof” seems similar to what an infallibilist might write, she is fully aware that “the
correct application of rules of inference” in mathematical practice is quite different
from the picture presented in formal logic.
While a proof is considered a prerequisite for the publication of a theorem, it
need be neither rigorous nor complete. (1991, p. 59)
In her writings Hanna discusses proofs only as texts, not as discourses or reasoning
(See Figure 2 in Chapter 2).
Harel and Sowder’s Research Perspective

Guershon Harel and Larry Sowder are best known for their categorisation of
university students’ “proof-schemes” which are their methods of verifying statements
51

via free access
CHAPTER 3
in mathematics (Harel, 2007; Harel & Sowder 1996, 1998, 2007; Sowder & Harel,
1998, 2003). Harel and Sowder set out in some detail a “comprehensive perspective
on proof” (2007, p. 806) that guides their work. That perspective includes consi-
deration of historical and socio-cultural factors, suggesting that though they do not
name it as such, Harel and Sowder operate from a social-constructivist philosophy
of mathematics.
Proofs (and theorems) are a product of human activity. (Harel & Sowder,
1998, p. 237)
It is clear that Harel and Sowder focus on proof as reasoning, as opposed to discourse
or especially proof-texts.
The final categories – the analytical proof schemes – encompass mathe-
matical proof, although again the emphasis is on the student’s thinking rather
than on what he or she writes. (p. 276)
Harel and Sowder take a broad view of proof, their essential requirement being that
a proof be convincing.
Proofs are first and foremost convincing arguments. (p. 237)
Deduction is important in some proof schemes, including those that they would
identify with mathematical proof:
Key to the analytical proof scheme is the transformational proof scheme: the
creation and transformation of general mental images for a context, with the
transformations directed toward explanations, always with an element of
deduction. (p. 276)
Reid (1995ab, 1996a, 2002ab) has a similar research perspective, although he differs
from Harel and Sowder in the way in which the category “proof ” is broadened. For
Harel and Sowder a proof must be convincing but not necessarily deductive or
semi-formal. For Reid a proof must be deductive, but not necessarily convincing or
semi-formal.
SUMMARY
We summarise the research perspectives described above in Table 3. You will note
that there is a rough correlation between philosophy, meaning of proof and breadth
of proof as researchers with an infallibilist philosophy are more likely to use
“proof ” to mean proof-texts and to have a narrow category of proof. On the other
hand researchers with a social-constructivist philosophy are more likely to use
“proof ” to refer to reasoning or discourse and to have a broad category of proof.
But this correlation is far from perfect. Some notable exceptions are Fawcett, the
preformalists, Mariotti and Hanna. Note also that some combinations do not occur
in our table. Whether this is due to the limited number of researchers we were able
to describe in this way, or due to some combinations being practically impossible,
is a question that would require further research to answer.
52

via free access
Table 3. Summary of research perspectives
Meaning for
“proof” or
Researcher Philosophy “proving” Breadth of “proof”
Fischbein & A priorist or Proof-text Narrow
Kedem Infallibilist
Duval Infallibilist Proof-text Narrow
preformalists Infallibilist Proof-text Broad (informal proof possible)
Fawcett Infallibilist Discourse Narrow
Balacheff Quasi-empiricist Discourse Very Broad (informal and non-deductive
(pre-1991) proofs permitted)
Hanna Social- Proof-text Broad (unconvincing
constructivist proofs permitted)
Harel & Social- Reasoning Very Broad (informal and non-deductive
Sowder constructivist proofs permitted)
Reid Social- Reasoning Very Broad (informal and unconvincing
constructivist proofs permitted)
Mariotti Social- Discourse Narrow
constructivist
Balacheff Social- Discourse Very Broad (informal and non-deductive
(post 1991) constructivist proofs permitted)
BALACHEFF’S EPISTEMOLOGIES OF PROOF
The classification of perspectives we have described above is not definitive or

widely accepted. It represents our effort to describe differences among researchers’
basic assumptions that are not often made explicit and for which no generally
accepted framework of description exists. Another noteworthy effort has been
made by Balacheff (2002/2004, 2008) to describe differences in researchers’
perspectives, in terms of “epistemologies of proof”
Our epistemology of proof (the relationship we have with truth and
validity) first shapes our research framework, even before the choice of a
problématique (i.e., the choice of the relevant questions and research problems),
and the choice of a theoretical framework and its related methodology. (2008,
p. 502)
Balacheff outlines five epistemologies of proof, which we summarise in Table 4. A
comparison with Table 3 will show that Balacheff’s epistemologies make similar
distinctions to our classification of perspectives. The researchers Balacheff chooses
to exemplify each of his epistemologies have perspectives that differ in at least two
of our three dimensions. He includes one exemplar (Healy & Hoyles) who we do
53

via free access
CHAPTER 3
Table 4. Balacheff’s epistemologies of proof
Epistemology Exemplar
Proof is the basis of mathematics and mathematics should be Fawcett, 1938
viewed as a model for proving in other fields, and as the best
example of rationality.
Proof is entirely subjective and is viewed on a continuum from Harel & Sowder,
the more “idiosyncratic” to the more “objective”. 1998
Proof is the basis of mathematics and that is what make mathe- Healy & Hoyles, 1998
matics distinct from other fields.
Proof is “an indispensable tool of mathematics rather than at the Hanna & Jahnke,
very core of that science” 1996
(Hanna & Jahnke, 1996, pp. 877–879)
A proof is part of a triad which also includes a statement and a Mariotti (1997) &
theory, which is a system of shared principles and inference rules. other Italian
mathematics is distinct from other fields in that it makes reference researchers (e.g.,
to theories, but theories are not fixed but socially generated. Mariotti et al., 1997)
not discuss because we could not find sufficient information to make a classification.
Balacheff omits himself from his classification, which is unfortunate as one would
assume he would have special insight into his own epistemology.
Balacheff also discusses another way in which researchers differ in their
approaches to proof, the relationship seen between proof and language. Here he
distinguishes two approaches: a focus on the text as an object and an interpersonal
focus on proof as communication or discourse. Balacheff’s exemplars here are
Duval (1991) and Fawcett versus Pimm (1987) and Burton and Morgan (2000). We
interpret this as similar to our dimension distinguishing proof-texts, discourses, and
reasoning.
DIVERSE OR COMPREHENSIVE PERSPECTIVES?
We mentioned above that Harel and Sowder (2007) set out their “comprehensive
perspective of proof”. “A comprehensive perspective on the learning and teaching
of proofs is one that incorporates a broad range of factors: mathematical, historical-
epistemological, cognitive, sociological, and instructional” (p. 806). For each of
these factors they identify questions whose answers clarify the perspective. This
represents a third approach (along with ours and Balacheff’s) to distinguishing and
describing researchers’ perspectives. However, Harel and Sowder use it to describe
only their own perspective, and Harel and Fuller (2009) make use of the factors
and questions to combine, rather than to distinguish, the perspectives of the twenty
contributions to Stylianou, Blanton, and Knuth (2009). This suggests that they see
it as possible and desirable to unify proof research in mathematics education by
achieving a consensus.
We, however, would agree with Balacheff that this is not possible.
54

via free access
Is a consensus possible? By consensus I mean at best a common theoretical

framework, at least a glossary guaranteeing a minimal set of shared meanings.
The deadlock on the route towards achieving such a programme is our own
epistemology of mathematical proof. Epistemology means here the identifi-
cation of an object and the web of the relations we establish around it with
other objects, as well as problems, tasks and other possible activities involving
it. .... Indeed, researchers themselves cannot avoid involving in their work
their own epistemology of mathematical proof and, beyond it, their own episte-
mology of mathematics. (2008, p. 508)
In fact, we would argue that this is not even desirable. A diversity of perspectives
provides the mathematics education community the opportunity to observe complex
phenomena in multiple ways. Any single perspective can notice only a part of what
is, but multiple perspectives allow us to observe more. However, having observed,
it is necessary to share with others and so our perspectives can be neither implicit
(so that their effects on our observations are not noticed) nor mutually incompre-
hensible. But this raises a question: Is it possible to share ideas with another whose
perspective you do not share? Balacheff raises and answers this question in this
way:
Would it be possible ... to connect our research outcomes—and beyond them,
our problématiques and theoretical frameworks? My belief is that this should
be possible, provided that each of us try to locate his or her own approach
among the possible ones, and make the effort to propose an understanding of
his or her results from a different perspective. (p. 511)
Attempts to describe perspectives, as we have outlined in this chapter, address
these issues, leaving us sharing Balacheff’s optimism that multiple perspectives
will be a source of strength for mathematics education research in the future.
55

via free access
via free access
PART 2
IMPORTANT RESEARCH FOCI, PAST AND PRESENT
In these six chapters we review the body of research on teaching and learning
proof. The volume of work in this field makes it impossible to review it all, but we
have included the research that is widely cited, as well as work that has contributed
important theoretical constructs to the field.
We begin, in Chapter 4, by examining some key empirical studies that are cited
to support some generally accepted beliefs about the state of proof teaching and
learning:
– Many (perhaps most) students accept examples as verification.
– Many students do not accept deductive proofs as verification.
– Many students do not accept counterexamples as refutation.
– Students accept flawed deductive proofs as verification.
– Many students accept arguments on bases other than logical coherence.
– Students offer empirical arguments to verify.
– Most students cannot write correct proofs.
Chapters 5–7 are structured around a model for reasoning outlined by Reid
(1995b). Reid refers to this model as a model for the psychology of reasoning in
school mathematics. It includes four linked aspects of reasoning shown in Figure 6:
– The need addressed by the reasoning,
– The type of reasoning employed,
– The degree to which the reasoning is formulated,
– The formality of the text produced.
Reid’s model presumes that reasoning begins in response to a personal need to
reason. Empirically, he identified four needs giving rise to reasoning in problem
solving contexts among the students he studied: to explain, to explore, to verify and
to engage in a teacher game. Reid’s “needs” correspond to roles or functions of
proving in professional and school mathematics, and several roles in addition to the
ones Reid mentions have been discussed in the mathematics education research
literature. These are discussed in Chapter 5.
Given a need, some type of reasoning is employed in an effort to address that
need. Reid’s figure indicates (by segments joining ovals) which types of reasoning
he found students used to address each need. Empirically, he observed three types:
inductive, deductive and reasoning by analogy. He also observed a method of
avoiding reasoning by making reference to an authority. We discuss these types of
reasoning and others, such as abductive reasoning, in depth in Chapter 6.
Reid’s third and fourth aspects concern ways of classifying proofs and arguments,
which is the topic of Chapter 7.
“Formulation” addresses the degree to which the person reasoning is aware of
their own reasoning. This can range from a complete lack of awareness (indicated
by the “unformulated” oval) to complete awareness (“formulated”). The process of
57

via free access
PART 2
Figure 6. Reid’s PRISM model.
formulation is an important part of proving. It is required for the articulation of

conjectures, generalisations, and explanations, the absence of which can cause
confusion, and it is necessary for expressing proofs semi-formally. Reid’s third
aspect also includes two special cases, that he calls “mechanical deduction” and
“formulaic proof-making.” We will discuss mechanical deduction as a type of
reasoning in Chapter 6 and formulaic proofs in Chapter 7.
Reid’s fourth aspect, formality of proof texts, has been considered in Chapter 1,
in connection with formalism, and will be referred to Chapter 7 in connection to
classifying proofs.
The PRISM model is based on a psychological perspective and so social processes,
such as argumentation, are not included in it. In Chapter 8, we turn to the literature
related to argumentation and discuss the different ways that work has been connected
to proof. Finally, in Chapter 9, we describe key teaching experiments with impli-
cations for teaching proof.
58

via free access
CHAPTER 4
EMPIRICAL RESULTS
A number of studies have been conducted that have produced results concerning
students, proofs and proving. These studies have revealed a number of findings that
are generally accepted by the community of researchers.
– Many students do not accept deductive proofs as verification.
– Many students do not accept counterexamples as refutation.
– Students accept flawed deductive proofs as verification.
– Many students accept arguments on bases other than logical coherence.
– Most students cannot write correct proofs.
As you read this chapter, you may want to reflect on these questions:
– What psychological and social factors might account for these findings?
– What kind of teaching might address them?
KEY STUDIES
The studies we review here are those often cited by researchers on proof and
proving. Basic information about them is summarised in Table 5. The focus of each
study is classified according to whether the subjects were expected to read and
comment on proofs and other arguments, or to write proofs themselves.
It is interesting to note that the studies included in Table 5 span a long time
frame; however, the geographical range represented is limited. This reflects our
own linguistic limitations, but also preferences for certain kinds of studies in
specific sub-communities of educational research. Specifically, prior to the advent
of TIMSS and PISA, there was little interest outside the English speaking world in
large scale assessments like those of Reynolds, Senk and Healy and Hoyles.
It is also worth taking into consideration that the subjects are not all in the same
age group. Most are secondary school students, but some university students, future
elementary school teachers and inservice upper elementary and secondary school
teachers are also included. This means that the findings listed above that refer to
students might also apply to teachers.
MANY STUDENTS ACCEPT EXAMPLES AS VERIFICATION
Several studies have shown that students are willing to accept examples as verifi-
cation for statements in mathematics. These findings are summarised in Table 6. These
results suggest that somewhere between 20% and 80% of students and teachers
(depending on age and mathematical background) consider a set of examples to be
sufficient to verify a mathematical statement.
59

via free access
CHAPTER 4
Table 5. Key studies reviewed
Study Year Location Type Subjects Focus

Balacheff 1988 France Written 14 pairs of students (aged Writing
assessment 13–24)
Barkai, Tsamir, 2002 Israel Written 27 in-service upper elementary
Tirosh, & assessment school teachers
Dreyfus
Bell 1976 UK Written 32 students (aged 14–15) Writing
assessment
Chazan 1993 US Interview 17 secondary school students
(aged about 15)
Coe & 1994 UK Written 60 students in the final year of Writing
Ruthven assessment secondary school
Fischbein & 1982 Israel Interview 397 students (aged 15–17) Reading
Kedem
Galbraith 1981 Australia Interview over 170 students (aged 12–17) Reading,
writing
Healy & 2000 England Written 2459 high attaining students Reading,
Hoyles and Wales assessment (aged 14–15) writing
Knuth 2002 US Interview 16 upper secondary school Reading
teachers
Martin & 1989 US Written 101 pre-service elementary Reading
Harel assessment school teachers (aged 18–22)
Porteous 1990 UK Interview 50 secondary school students Writing
(aged 11–16)
Recio & 2001 Spain Written 429 beginning university Writing
Godino assessment students taking mathematics
courses
Reynolds* 1967 UK Written almost 2000 secondary school Reading,
assessment students (aged 11–17+) writing
plus 80
interviews
Senk 1985 US Written 1520 secondary school students Writing
assessment (aged 15–17)
TIMSS 1994- more than Written 3342 advanced mathematics Writing
1996 40 assessment students in their final year of
countries secondary school
Williams 1979 Canada Written Grade 11 students (aged 16) Reading
assessment
*reported in Lovell, 1971
60

via free access
EMPIRICAL RESULTS
Table 6. Results showing that many students accept examples as verification
Author % accepting examples as verification

Reynolds, 40% of first form (aged 11–12),
1967 35% of sixth form (aged 16+) non-mathematical stream,
20% of sixth form mathematical stream,
said “Yes” when asked if sixteen cases of even numbers written as the sum
of two primes were sufficient to verify this for all even numbers.
An additional 20% in the first form either checked a few more cases before
declaring the statement true, or checked the converse for a few cases.
Williams, 68% felt that examples were sufficient on one item (only 6.8% felt the
1979 need for a deductive proof ).
On another item the numbers were 54% and 14%.
Martin & 80% gave a high validity rating (3 or 4) to at least one empirical argument.
Harel, over 50% gave a 4 to at least one empirical argument.
1989 Less than 10% gave a 1 to all four empirical arguments.
Healy & 37%-28% did not explicitly recognise limits of examples
Hoyles, 3%-2% believed examples would get the best mark from a teacher
1998, 2000 (see below for discussion)
Knuth, 31% (5 of 16 teachers) considered an empirical argument to be a proof,
2002a and two others (12.5%) raised objections to it other than the use of
examples, and would have accepted the empirical argument if revised.
In Healy and Hoyles’ (1998, 2000) study:

Students were presented with mathematical conjectures and a range of arguments
in support of them; they were asked to make two selections from these
arguments—the argument that would be nearest to their own approaches and
the argument they believed would receive the best mark from their teachers. ...
Two conjectures were included, one familiar (Question A1) and the other
unfamiliar (Question A6) ... (2000, p. 399)
Six arguments were provided for Question A1 and four were provided for Question
A6. In each case one argument was empirical. Each argument was ascribed to a
fictional person for identification purposes. The empirical arguments were ascribed
to “Bonnie” (Question A1) and “Leon” (Question A6).
Their results suggest some factors that may influence students’ acceptance of
empirical arguments. They show that while students might use empirical evidence
themselves (see below), and feel it has value for convincing others (see Chapter 5),
they recognise that it is not acceptable as mathematical proof in school and that it
does not provide verification in general.
Only a tiny minority of students chose an argument consisting entirely of
examples as one that would receive the best mark (3% in A1, 2% in A6...),
although many chose such an argument as being nearest their own approach
(24% in A1 and 39% in A6...). This result suggests that most students were
61

via free access
CHAPTER 4
aware that empirical arguments had limitations; they knew more was expected
of them. (pp. 409–410)
The students were asked whether each of the arguments for each question has a
mistake, shows that the statement is always true or only shows that the statement is
true for some numbers. From the responses to these questions Healy and Hoyles
generated a rating of the validity the students gave the argument. “... although
around one third of students (37% and 28% for Bonnie’s and Leon’s arguments,
respectively) had no idea of the validity of these empirical arguments, more than
half gave completely correct evaluations (54% and 60%); that is, they knew that
these arguments had been proved to be true only in a subset of cases” (p. 411). In
combination with interview data, Healy and Hoyles concluded that students some-
times accept examples as verification if they already believed the result was true:
The data indicate that students were more likely to assess empirical arguments
as general—to believe them to be proofs—if they were already convinced of
the truth of the statement and so intuitively could extend the argument for them-
selves. When using this strategy was not possible, as in response to Question
A6, they assessed the limitations of the empirical argument correctly. (p. 412)
In summary, Healy and Hoyles conclude that “The majority were ... aware that
empirical arguments were not general—particularly if the statement to be proved
was not familiar—but they recognized that examples offered a powerful means
of gaining conviction about a statement’s truth” (2000, p. 425). This conclusion
suggests that the results of other studies that indicate that students and teachers
accept examples as verification may need to be examined carefully to see what
beliefs underlie this behaviour.
Chazan (1993) reports on a study which also reveals some possible explanations
for the fact that students prefer arguments based on examples. He interviewed
seventeen US secondary school students studying geometry using dynamic geometry
software, selected on the basis of the diversity of their responses to a questionnaire.
He shows that while students accept and use examples to verify, they recognise that
there are limitations to this method. The students he interviewed made it clear that
if they were verifying a statement about triangles using examples, they would be
sure to examine triangles of each type known to them, and possibly special cases
that they knew to be problematic (p. 370). Chazan also identifies three objections
student might make to the use of examples to verify (These were the same as the
objections suggested to them by their teachers.). The first is that a counterexample
might exist outside of the set of examples checked (pp. 370–371). The second is
that the examples checked might all be special in some (known or unknown) way
(p. 371). The third focussed on the limitations of measuring as a way to determine
properties of a figure (p. 371).
MANY STUDENTS DO NOT ACCEPT DEDUCTIVE PROOFS AS VERIFICATION
Another belief that seems to underlie some students’ and teachers’ behaviour is a
belief that deductive proofs do not provide verification in general. Table 7 summarises
62

via free access
EMPIRICAL RESULTS
Table 7. Summary of results showing that many students and teachers

do not accept deductive proofs as verification
Author % rejecting deduction as verification

Williams, 1979 20%
Fischbein & 63%–80% (depending on mathematics background)
Kedem, 1982
Martin & Harel, 26%–38% rated a general proof 1 or 2 (depending on item) (see Table 10)
1989
Healy & 38% (62% stated that no further checks were needed once a deductive
Hoyles, 2000 proof was given, which suggests that the remainder felt that further checks
might be needed.)
Knuth, 2002a 38% (All those interviewed stated that proofs establish the truth of a
conclusion, however six of the sixteen teachers also believed that even
after a proof was done a counterexample might be found.)
the results of studies that touch on this question. The behaviour reported by
Schoenfeld (1989) also seems related to this belief. He found that students could
produce a deductive proof of a general statement, but then made conjectures that
violated the statement they had proved (p. 340).
Chazan’s (1993) interview study explored some of the reasons for students’
disbelief in deductive proof as a method of verification. Their objections included:
– Counterexamples might still exist that are outside the cases covered by the proof
(this often included the next objection) (p. 372);
– The proof only proves the result for the diagram given (in many cases, the
diagram was considered to be general enough to prove the result for all triangles
of the same type, but not for all triangles of all types) (pp. 372–373);
– The assumptions used in the proof might be wrong (pp. 373–374);
– The proof is expressed in the singular, so it only applied to a single case (If it
specified “for every” at every step, it would be different) (p. 375);
– Misunderstanding what was given in the proof (p. 376).
Based on these objections, Chazan reports that the students mentioned a number
of classroom behaviours that seem counterproductive for their learning of geometry:
They looked for counterexamples to statements their teacher had proven (p. 381);
They did not believe the statements their teacher had proven (p. 381); and they saw
no reason to learn to do proofs (p. 382).
MANY STUDENTS DO NOT ACCEPT COUNTEREXAMPLES AS REFUTATION
A few studies have directly investigated whether students behave as if they believe
that a single counterexample refutes a general statement.
Galbraith (1981) reports that about 18% of the 12–17 year old Australian students
interviewed in his study felt that a single counterexample was insufficient to refute
a general statement. The statement was: “Every number in the list L [of numbers
less than 70 for which the sum of the digits is divisible by 7] can be found by
63

via free access
CHAPTER 4
adding 9 to the previous number. You start with 7” (p. 10, 17). A similar proportion
felt that a single counterexample was insufficient to refute a general statement
concerning quadrilaterals, but in some cases different students were involved,
“Consequently the number of pupils having problems accepting the significance of
counterexamples almost certainly exceeds the number identified from a particular
context” (p. 19). In the task where the students had to produce an argument of their
own, Galbraith observed some interesting cases in which the students came to an
incorrect conclusion and were then shown counterexamples. “In those cases very
strong prompting was required to prise the pupils away from their own deductions to
consider other numbers ... which refuted their claims. Some ... continued to revert
to their perceived principle in the face of contradictory data” (p. 21). Galbraith
concludes that “A consistent minority of pupils did not accept the meaning which the
world of mathematics ascribes to the presence of a counter-example” (p. 24).
However, Porteous (1990) found that of the 50 UK secondary school students he
interviewed, 48 rejected a false generalisation on the basis of a counterexample. It
should be noted that in Porteous’ study, the student had no vested interest in the
statement but in Galbraith’s study either the students had verified the statement
empirically beforehand (in the case of the divisibility task) or the counterexample
contradicted their prior experience of quadrilaterals.
Barkai, Tsamir, Tirosh, and Dreyfus (2002) report that in the case of the
statement “The sum of any four consecutive integers is divisible by four” all the
upper elementary school teachers participating in their study correctly noted that
this statement is false. 72% justified their claim by providing at least one counter-
example, but only 36% felt that their counterexample would be considered an
acceptable justification by the university course instructor, suggesting they did
not see a counterexample to a universal statement as a valid mathematical
proof (p. 2–60). And the fact that some of the teachers provided more than one
counterexample suggests that they do not believe that a single counterexample is
sufficient to refute a universal statement. On the other hand, for the matching
existence statement “There exist four consecutive integers whose sum is divisible
by four” 20% of the teachers judged this statement to be false, provided a counter-
example, and expected that their justification would be accepted by the university
course instructor (p. 2–62), showing that among these teachers misconceptions
concerning counterexamples included both underestimating and overestimating
their power.
STUDENTS ACCEPT FLAWED DEDUCTIVE PROOFS AS VERIFICATION
From Table 8 it is clear that a substantial number of students, future elementary

school teachers, and practising secondary school teachers accepted flawed deductive
arguments as verifications. The lower rate of acceptance in Healy and Hoyles’
study can be accounted for by noting that in their study the students had to select
only one of the four to six arguments offered as the one that would get the best
mark. This means that the students who chose flawed arguments believed they would
get a better mark than a correct deductive proof.
64

via free access
EMPIRICAL RESULTS
Table 8. Summary of results showing students accept flawed deductive

proofs as verification
Author % accepting incorrect deductive proofs

Martin & 52% rated a false proof of a familiar statement “High” (3 or 4)
Harel, 1989 38% rated a false proof of an unfamiliar statement “High” (see Table 10).
Healy & 40% on Question A1 and 24% on Question A6 chose an incorrect
Hoyles, 2000 deductive proof as the one their teacher would mark the best.
Knuth, 2002a 63% accepted an argument that proved the converse as a proof.
One might expect that upper secondary school teachers would be more successful
than secondary school students or elementary school teachers at recognising valid
and invalid proofs. Knuth’s (2002a) study shows that this is not the case. Knuth inter-
viewed 16 US upper secondary school teachers. He asked them questions explicitly
about what proof meant to them, and also asked them to compare and evaluate sets of
arguments that included deductive proofs, incorrect deductive proofs, proofs based
on “generic examples” (see Chapter 7) and empirical arguments based on examples.
Almost all the teachers could identify correct proofs from a set of arguments.
“93% of the ratings given to the arguments that constituted proofs were correct”
(p. 391). However:
a third of the ratings that the teachers gave to the nonproofs were ratings as
proofs. In fact, every teacher rated at least one of the eight nonproofs as a proof,
and 11 rated more than one as a proof. (p. 391)
As some of the “nonproofs” were empirical and at least one was a “particular case”,
i.e., a generic example, this high rate of acceptance of nonproofs does not translate
directly into a high rate of acceptance of logically flawed deductive proofs. However,
among the nonproofs was an argument that proved the converse, and 10 of the
16 teachers (63%) accepted it as a proof (p. 392). Some teachers admitted that they
were judging an argument based on the method it used, rather than its correctness.
For example, one teacher, who confessed to not really understanding proof by
[mathematical] induction ... nevertheless found such a proof convincing
because of its method: “I know that that is a valid way of proving things” ...
Similarly, another teacher commented, “I know that this ... is one I’ve seen
used before, and I assume it’s a good way to do it” ... Thus in both of these
cases, the teachers were convinced that the argument was a proof because of
the method employed rather than because of an understanding of the method
itself. (p. 395)
STUDENTS’ CRITERIA FOR THE ACCEPTANCE OF ARGUMENTS
The comments of the teachers interviewed by Knuth suggest that the form of an
argument might lead some teachers and students to accept flawed deductive
arguments because they look like deductive proofs. There is some evidence that this
65

via free access
CHAPTER 4
Table 9. Factors influencing acceptance of arguments
Author Factors influencing acceptance of arguments

Martin & Harel, 1989 Form, familiarity
Healy & Hoyles, 2000 Empirical reasoning, narrative form (for their own approach)
algebraic form (for best mark)
Knuth, 2002a Use of diagrams and examples to make the argument concrete,
familiarity of the methods used.
might be the case, but other factors may also influence students’ acceptance of
arguments. These are summarised in Table 9.
The study most often cited in support of the assertion that students accept
arguments on the basis of their form is that of Martin and Harel (1989). They
conducted a study in which 101 university students enrolled in mathematics courses
for prospective elementary school teachers were asked to rate “proofs” of two
statements, one familiar (“If the sum of the digits of a whole number is divisible by
3, then the number is divisible by 3”) and one new to them (“If a divides b, and b
divides c, then a divides c”) (p. 43). The “proofs” are described as being either
“inductive” (empirical) or “deductive”.
Four types of empirical verifications and three types of deductive verifications
were offered. The deductive verifications included a correct general symbolic proof,
an incorrect symbolic “ritualistic” argument, and a “particular proof ” in which the
structure of the general symbolic proof was followed, but specific numbers were
used instead of variables; that is, used as generic examples .
They comment: “Many students who correctly accepted a general-proof verifi-
cation did not reject a false proof verification; they were influenced by the appearance
of the argument—the ritualistic aspects of the proof—rather than the correctness of
the argument” (p. 49). In the case of the familiar statement this seems to have been
the case. Of the 75 who rated the general proof highly, slightly more than half (38)
also rated the ritualistic proof highly. From Table 10 it is clear that this is true of
those who rated the general proof low as well. In the case of the unfamiliar
statement the situation is different and it appears that factors other than the form of
Table 10. Frequencies of ratings for the familiar and unfamiliar deductive verifications
Statement/ Verification Rating Rating Rating Rating Rating “High”

1 2 3 4 (3 or 4)
Familiar/ General proof 3 23 22 53 75
Familiar/ False proof* 26 22 27 25 52
Familiar/ Particular proof 12 30 25 34 59
Unfamiliar/ General proof 14 24 35 28 63

Unfamiliar/ False proof* 33 29 31 7 38
Unfamiliar/ Particular proof 19 27 33 22 55
*One student did not rate this item.
66

via free access
EMPIRICAL RESULTS
the proof came into play. Of the 63 students who rated the general proof of the
unfamiliar statement highly, only a third (21) also rated the ritualistic proof highly.
This is still a considerable number, but noticeably less than in the case of the familiar
statement. This outcome is interesting to compare with the comment of Healy and
Hoyles (above) that students were more likely to accept examples as verification
for results they already knew to be true. It suggests the plausible hypothesis that
students are less critical in their assessment of arguments supporting conclusions
they know to be true than of arguments for unfamiliar conclusions.
STUDENTS OFFER EMPIRICAL ARGUMENTS TO VERIFY
Above it was noted that students accept empirical arguments, examples, as verifi-
cation, and so it is no surprise that they offer examples as verifications themselves.
Results that support this are summarised in Table 11. Note, however, that these
results indicate that there is variation in the approaches students choose, and that
their choices seem to depend on the context.
Porteous (1990) interviewed 50 UK secondary school students (ages 11–16) who
were asked to verify a number of statements. He reports that most used empirical
evidence to do so, only about 15% providing a deductive proof (p. 595). After the
students had verified a general statement (either empirically or with a deductive
proof ) they were asked if it was true also for a particular case. On almost all
occasions when a student had used empirical methods to verify a general statement
the student chose to check the particular case, indicating that while a student might
say that they believe a general statement applied to all numbers, their confidence is
not absolute. Porteous found that in only 12 of the 43 cases where a student used a
deductive proof to verify did the student then wish to check the particular case,
although an additional 7 seemed to use the additional example as an opportunity to
verify the steps of their proof on a specific case (a behaviour similar to that reported
by Vinner, 1983) (p. 596). When Porteous showed a deductive proof to those who
had used empirical methods to verify, most of the students (55%) continued to
check particular cases individually. This suggests that the students who had found
their own deductive proofs were more convinced by them than the students who
had used empirical methods and were then shown deductive proofs.
Healy and Hoyles (2000) suggest that students offer examples as verification,
but not because they think examples are better for verification than deductive proofs.
Rather they are simply not capable of producing deductive proofs: “These differences
not only showed that students were more likely to construct empirical arguments
than to choose them but also supported the suggestion that they were the best
arguments available to the students, and not necessarily that they were satisfied
with them as proofs” (p. 412).
Students’ use of examples in situations where a deductive proof might be expected
is at first glance disappointing. However, the issue might not be so simple, as the
work of Alcock (2004) suggests. She found that research mathematicians made use
of examples, before during and after proving a statement, for three purposes:
understanding the statement, generating an argument, and checking the argument.
67

via free access
CHAPTER 4
Table 11. Students’ use of empirical arguments
Author Use of empirical arguments

Bell, 1976 In a numerical context, 28% argued empirically (31% argued deductively,
and the remainder misunderstood the problem).
In a geometric context 40% argued empirically.
Galbraith, 1981 25% were satisfied to do a partial check in a context where a complete
check was possible.
Balacheff, Of 14 pairs, 11 used empirical methods (either naïve empiricism or
1988a crucial experiment, see Chapter 7).
Porteous, 1990 About 85% used empirical methods. (see discussion below).
Coe & Ruthven, 54 of 60 (90%) pieces of course work used only empirical methods.
1994
Barkai, Tsamir, 52% offered an empirical argument when asked to justify a universal
Tirosh, & statement.
Dreyfus, 2002
MOST STUDENTS CANNOT WRITE CORRECT PROOFS
Many publications on proof in the mathematics education literature begin by asser-

ting that there is a problem because most students cannot write a correct proof. A
number of studies are cited to support this assertion, the results of which are sum-
arised in Table 12. These studies also show that the ability to construct a valid proof
varies depending on age, the statement to be proven and nationality. Younger students
seem to be less able to construct proofs than older students. More research is required
to clarify the relationship between item type and success in constructing proofs.
The Third International Mathematics and Science Study (TIMSS) was conducted
in more than 40 countries in 1994–1996. One question (Item K18) on the test for
advanced mathematics students in their final year of secondary school asked the
students to write a proof (TIMSS, 1998c, p. 89). Ten countries tested a sample of
students sufficient to consider the results representative (Martin & Kelly, 1998, p. 32).
Table 13 shows data from nine of the ten countries with suitable samples (The data
from one country is inconsistent in the published reports and so is not included
here). Note, however that this is only one item on a test that was intended to give a
more general picture of achievement.
The columns “% able to construct a valid proof ” and “% writing a wrong or
partial answer” do not add up to 100% because not all students attempted the proof
item. Some of these ran out of time, but others may have skipped the item recogni-
sing that they would not be successful if they attempted it.
The range of success on this item is quite wide (from 21% in Germany to 65%
in Greece). The high rate of success in Greece is striking, but should be interpreted
carefully. It is possible that something very similar to the proof item is part of the
standard curriculum in Greece. Note that unlike all the other countries’ students,
the Greek students who successfully completed the proof all took the approach
68

via free access
EMPIRICAL RESULTS
Table 12. Students able to write correct proofs
Author % able to write correct proofs

Reynolds, 1967 14% in first form, 60% in fifth form and in sixth form but not taking
mathematics and almost 70% in sixth form taking mathematics, could
produce a correct proof by contradiction in a geometric context.
32% in first form and over 75% in sixth form could deduce a correct
conclusion from three non-mathematical statements
Bell, 1976 19% of attempts were deductive, but these were all proofs by exhaustion
or counterexample.
Senk, 1985 30% “master proofs”, 20% “can do some proofs of greater complexity”,
and 25% “can do only trivial proofs” (p. 453–454).
TIMSS, 1994– 35% could write a proof in a geometric context. (International average).
1996 21% to 65% (depending on country) were successful (see discussion,
below).
Healy & Hoyles, 22% were able to write a complete proof of a familiar conjecture;
2000 18% could write a partial proof.
3% were able to write a complete proof of an unfamiliar conjecture;
9% could write a partial proof.
Recio & Godino, 32.9% were able to write a correct proof for two statements, one
2001 arithmetic and one geometric.
Barkai, Tsamir, 23%–100% were able to write a correct proof (depending on item, see
Tirosh, & discussion below).
Dreyfus, 2002
Table 13. Students who were able to construct a valid proof for TIMSS item K18
Number of % able to % taking the % writing a

students construct a valid expected wrong or partial
Country sampled proof approach answer
Canada 807 34% 96% 61%
Cyprus 110 35% 94% 54%
Czech Republic 271 34% 93% 39%
France 259 53% 96% 29%
Germany 732 21% 86% 46%
Greece 109 65% 100% 6%
Lithuania 249 30% 93% 44%
Russia 445 44% 92% 36%
Switzerland 360 46% 89% 35%
Sources: TIMSS 1998ab
69

via free access
CHAPTER 4
expected by the test designers, suggesting prior experience with the proof or one
like it. Also note that very few Greek students attempted the item and got it
wrongor only partly right. Most were successful or they skipped the item, either
because they chose not to do it or because they ran out of time (20% did not
complete any items after K18). This suggests that for these students proof is an all
or nothing affair; either they know the proof and write it correctly or they recognise
that they do not know it and write nothing.
Barkai, Tsamir, Tirosh, and Dreyfus (2002) report on a study of in-service upper
elementary school teachers in Israel, that suggests that proving the existence of
something is easier than proving the non-existence of something. 27 teachers were
asked to judge whether six statements were true or false, and to provide justification
for their position. The statements, three true and three false, were related to divisi-
bility of sums of consecutive numbers. Three were universal statements and three
were existential statements (p. 2–58).
The teachers’ success in providing correct deductive proofs or counterexamples
varied by item. Table 14 summarises the results according to the type and truth
value of the statements. The teachers had the most difficulty writing proofs for
statements like #1 and #5 that require a deductive approach because they assert that
something is true for an infinite number of cases. The other statements can be proven
or disproven using examples or counterexamples, and the teachers had much greater
success with those statements. The only problems they had arose in judging whether
statements that are sometimes true and sometimes false (e.g., #3 and #6) are true
when presented in universal or existential forms. Of those who made correct judge-
ments of those statements, almost all could produce correct proofs.
Table 14. Results according to the type and truth value of the statements
Statement #1 #2 #3 #4 #5 #6
Type Universal Universal Universal Existential Existential Existential
Truth value True False False True False True
Correct Judgement 100% 100% 69% 100% 77% 68%
Correct proofs 41% 88% 69% 96% 23% 64%
True for what set? All n No n Some n All n No n Some n
1. The sum of any five consecutive integers is divisible by five.

2. The sum of any four consecutive integers is divisible by four.
3. The sum of any three consecutive integers is divisible by six.
4. There exist five consecutive integers whose sum is divisible by five.
5. There exist four consecutive integers whose sum is divisible by four.
6. There exist three consecutive integers whose sum is divisible by six.
IDEAS FOR RESEARCH
The results summarised above are interesting as a basis for research, as they establish
some accepted truths about students’ and teachers’ understandings of proofs and
proving. However, they also raise questions that deserve additional research. Here we
70

via free access
EMPIRICAL RESULTS
will outline some of these questions. Doubtless the reader will have noted other
points of personal interest.
Reynolds (1967) and Healy and Hoyles (2000) both conducted large scale
written assessments in the UK, but separated by a very eventful thirty years of
social and curriculum change. One might ask whether their results reveal significant
changes in students’ understandings of proofs and proving over that time period.
Healy and Hoyles focussed on high achieving 14–15 year olds, while Reynolds
looked at a wider age and achievement range, so one would expect that Healy and
Hoyles’ results should fall somewhere within the range of results obtained by
Reynolds. For accepting examples as verification this seems to be the case (Reynolds’
range is 20%-40%; Healy and Hoyles have 28%–37%). Healy and Holyes’ results
concerning proof writing also seem to fall into the range defined by Reynolds’ results.
In their study 22% were able to write a correct proof of a familiar conjecture, and
an additional 18% could write a partial proof. Reynolds’ range is from 14% in first
form to 70% in sixth form. But these results must be interpreted cautiously. Healy
and Hoyles used different tasks than Reynolds, and their criteria for a “correct”
proof might be different. A researcher interested in pursuing the question of change
over time would have to examine the assessments in detail. Note also that Healy
and Hoyles’ results for writing a proof of an unfamiliar conjecture (3% complete,
9% partial) fall outside of Reynolds’ range. This suggests that there has been a
decline in proof writing skill over time, but whether this is related to curriculum
change (as Healy and Hoyles suggest), social changes (the school population has
expanded in the time period being considered) or other factors requires additional
research to clarify.
Another question arising from the results summarised in this chapter concerns
national differences. Almost all the studies reported here (which are the studies
often cited in the literature) were conducted in English speaking countries. Only one
of these countries is represented in Table 13 (Canada). The other countries represented
in the TIMSS data showed success in writing proof that ranges from much higher
to somewhat lower. This is not surprising, as it is still true, as Bell (1976) notes,
that “Viewed internationally, the proof aspect of mathematics is probably the one
which shows the widest variation in approaches” (p. 23). Additional research on
national differences is necessary before results from the English speaking world
can be extrapolated to the whole world.
Further research is also needed on teachers and teacher preparation for teaching
proof. Three of the studies summarised here looked at teachers’ understanding of
proof, and generally there seems to be not much difference when compared to
students’ understanding of proof. If this is the case then the level of students’
understanding might be best improved by addressing teachers’ understanding of
proof itself, rather than exposing them to new methods of teaching about proofs
and proving.
As noted above, many students do not accept deductive proofs as verification.
However, the studies cited give a wide range of percentages of the populations
studied who behave in this way, from 20% in Williams’ study to 80% in Fischbein
and Kedem’s. And Chazan’s work suggests a variety of objections students might
71

via free access
CHAPTER 4
have to accepting deductive proofs, some of which are valid (the use of false
assumptions). Before concluding (as Fischbein & Kedem do) that students do not
understand the nature of proof because they do not accept deductive proofs as
verification more research is needed on the circumstances under which students
reject deductive proofs, and their reasons for doing so.
Students’ attitudes towards counterexamples also deserves additional study. For
example, Harel and Sowder (1998) report that students at two universities behaved
very differently with respect to counterexamples.
The first author’s students seldom used proof by counterexample ... and they
did not seem to be convinced by it .... The second author’s students, on the
other hand, often sought counterexamples first, as did Goetting’s subjects
(1995), although many of her interviewees were not certain whether a counter-
example gave a proof. Whether this seeming difference from the first author’s
students is a fact or an happenstance of either the particular interviewees or
the curricula at the different universities, we do not know. This, together with
Balacheff’s (1991[b]) finding that younger students (junior-high school students
in France) do react to counterexamples in various ways requires a further
look at college students’ conception of proof by counterexample. (p. 254)
72

via free access
CHAPTER 5
THE ROLE OF PROOF
In Chapter 2 we considered various meanings the words “proof ” and “proving”

have, in everyday life, science, mathematics and logic. Some of these meanings are
related to the role of proof, for example, when “proving” means “verifying” it is its
role that is dominant, and other aspects, such as the reasoning used, are less
important. In this chapter we will look at the roles proof can have in mathematics
as a way of outlining what the goal of teaching proof might be. We will then look
at the roles proof has for students to establish what the starting point for teaching
could be. Finally we will discuss the relationship between the role of proof and
teaching.
Different authors use different words to refer to roles, needs, function, purposes,
etc. of proof and proving. In the following we will follow the original authors’
usage when discussing their work without intending to imply that the different authors
are referring to different things.
As you read this chapter you may want to consider these questions:
– What significance might the different roles of proof have for teaching?
– In what way?
THE ROLES OF PROOF IN MATHEMATICS
Mathematicians have long recognised that there are different roles a proof might play.
Proof serves many purposes simultaneously. In being exposed to the scrutiny
and judgement of a new audience, the proof is subject to a constant process of
criticism and revalidation. Errors, ambiguities, and misunderstandings are
cleared up by constant exposure. Proof is respectability. Proof is the seal of
authority.
Proof, in its best instances, increases understanding by revealing the heart of
the matter. Proof suggests new mathematics. The novice who studies proofs
gets closer to the creation of new mathematics. Proof is mathematical power,
the electric voltage of the subject which vitalizes the static assertions of the
theorems.
Finally, proof is ritual, and a celebration of the power of pure reason. (Davis &
Hersh, 1981, p. 151)
As Barbin (1996) notes, in the seventeenth century the mathematicians Arnauld
and Nicole criticised the proofs of Euclid for verifying without explaining, for
“taking more care over certainty than with evidence, and of convincing the mind
73

via free access
CHAPTER 5
rather than enlightening it” (Arnauld & Nicole, 1965, p. 326, cited in Barbin,
1996, p. 199). In addition to verification and explanation, many other roles of proof
in mathematics have been suggested, including exploration/discovery, systematis-
ation, communication, gaining “theorem credits”, aesthetics, intellectual challenge,
construction of an empirical theory, clarification of a definition or the consequences
of an assumption, and incorporation of a fact into a new framework. Here we will
expand on the major roles of proof in mathematics, and describe the minor ones.
Verification
As de Villiers (1990) notes “Traditionally the function of proof has been seen almost
exclusively in terms of the verification (conviction or justification) of the correctness
of mathematical statements. The idea is that proof is used mainly to remove either
personal doubt and/or those of skeptics” (p. 17). This role for proof is reflected in
the oft quoted sequence:
Convince yourself, convince a friend, convince an enemy. (Mason, Burton &
Stacey, 1982, p. 95)
Note that here an important difference is indicated. Verifying (for oneself ) is a
different process than convincing (someone else). Nonetheless these two roles are
similar enough that they are almost always discussed together. As verification is the
traditional role ascribed to proof, it is not surprising that many authors have
mentioned in when discussing possible roles of proof. These include Fischbein and
Kedem (1982), Bell (1976), and many others.
Recall that for Descartes proofs in geometry were taken as the model for
reasoning because they verified.
Of all those who have already searched for truth in the sciences, only the
mathematicians were able to find demonstrations, that is, certain and evident
reasons. (Descartes, 1637/1993, p. 11)
Epistemic value. In considering the role of proof as verification, it is important to

recognise that while logically a statement can only be true or false, psychologically
it can take on one of many values, which Duval (1990, 2007) calls its “epistemic
value”. Epistemic value is a personal judgement of whether and how the proposition
is believed. It can take on values such as opinion, belief, certainty, principle, hypo-
thesis, etc. “La valeur épistémique est la valeur d’opinion, de croyance, de certitude,
de principe, d’hypothèse, etc.” (1990, p. 198). The range of epistemic values that are
possible and that correspond to “truth” depends on the discipline, situation or domain.
In mathematics, according to Duval, for a mathematical statement to be “true” is
has to fall in a very narrow range of epistemic values, but for a science fact to be
“true” it can fall into a wider range (e.g., observed).
Selon les disciplines, et aussi selon les situations, la gamme des valeurs épisté-
miques qui correspondent à la valeur “vrai” change : elle est par exemple très
réduite en mathématiques, elle l’est moins dans les disciplines expérimentales.
74

via free access
THE ROLE OF PROOF
[According to discipline, and also depending on the situation, the range of

epistemic values that correspond to the value “true” changes: for example it is
very small in mathematics, it is wider in the experimental subjects.] (p. 198)
Reasoning or argument, according to Duval, has the function of changing the
epistemic value of a statement, for example from conjecture to theorem. “Le raisonne-
ment joue sur des différences de valeur épistémique de certaines propositions
(hypothèses, suppositions, définitions, règles, principes, etc..) pour établir la vérité
d’autres propositions, et par suite pour en modifier la valeur épistémique initiale”
[Reasoning uses the differences in epistemic value of statements (hypotheses,
assumptions, definitions, rules, principles, etc.) to establish the truth of other state-
ments and so to modify their original epistemic value.] (p. 199).
Explanation
We have already mentioned Arnauld and Nicole’s criticisms of the proofs of Euclid
for verifying without explaining and “mathematicians routinely distinguish proofs
that merely demonstrate from proofs which explain” (Steiner, 1978, p. 135), which
suggests that explanation is an important role of proof in mathematics.
Steiner describes an explanatory proof in this way:
An explanatory proof makes reference to a characterizing property of an
entity or structure mentioned in the theorem, such that from the proof it is
evident that the result depends on the property. It must be evident, that is, that
if we substitute in the proof a different object of the same domain, the
theorem collapses; more, we should be able to see as we vary the object how
the theorem changes in response. (p. 143)
From this it is clear that not all proofs are explanatory, as not all proofs make this
reference to a characterising property. This is an important distinction, as all
proofs must be able to fulfil the role of verifying, but not all proofs fulfil the role
of explaining. However, proofs that explain are considered preferable to those
that merely verify. Thurston (1995) comments on his experience as a graduate
student discovering that what his colleagues wanted was not only verification:
“I thought what they sought was a collection of powerful proven theorems that
might be applied to answer further mathematical questions. But that’s only one
part of the story. More than knowledge, people want personal understanding.
(pp. 35–36).
The role of proof as explanation also becomes important in cases where other
evidence has effectively verified a conjecture, making it unlikely that a proof that
only verified what is already known will be satisfying. Instead what is sought is a
proof that explains:
We believe, in other words, that a proof would be a way of understanding why
the Riemann conjecture is true, which is something more than just knowing
from convincing heuristic reasoning that it is true. (Davis & Hersh, 1981,
p. 368)
75

via free access
CHAPTER 5
The role of proofs as explanation is also illustrated by the proofs chosen for Proofs
from the Book (Aigner & Ziegler, 2004) which includes the most revealing proofs
of a wide range of theorems. Both the nature of the proofs included and the very
fact that mathematicians are especially interested in such proofs shows the importance
of explanation as a role of proof in mathematics.
Exploration / Discovery
De Villiers (1990) asserts that proving is an important means of exploring in mathe-
matics.
Even within the context of such formal deductive processes as a priori axio-
matization and defining, proof can frequently lead to new results. To the
working mathematician proof is therefore not merely a means of a posteriori
verification, but often also a means of exploration, analysis, discovery and
invention. (p. 21)
Elsewhere de Villiers (1999) notes, “there are numerous examples in the history of
mathematics where new results were discovered or invented in a purely deductive
manner [e.g., non-Euclidean geometries]” (p. 5). A very simple example occurred
in a Masters course (taught by Reid). A group of students assigned the task of
verifying that the sum of two consecutive odd numbers is even as part of a class
presentation. The point of the exercise was to try to come up with new ways of
proving this fact, so simply asserting it as a special case of the known fact that the
sum of any two odd numbers is even was not allowed. Reid easily verified it using
a generic example (see Chapter 7) and by reasoning by recurrence (see Chapter 6).
He then wrote a straightforward algebraic proof:
The two numbers are 2n–1 and 2n+1.
2n – 1 + 2n + 1 = 2(2n) which is even.
IN FACT it is a multiple of FOUR!
His first two proofs had told him that the statement is true. They had verified it. By
the last one also told him something new, that the sum is always a multiple of 4.
Although he had not set out to discover anything, his final proof had allowed him
to discover something new.
Systematisation
The third role Bell (1976) ascribes to proof is systematisation, which he considers
to be “the most characteristically mathematical” (p. 24). It is “the organisation of
results into a deductive system of axioms, major concepts and theorems, and minor
results derived from these.” (p. 24). De Villiers (1990) also mentions systematis-
ation as a role of proof in mathematics and notes its importance in axiomatisation
and defining. He seems to refer to a process like that described by Lakatos (1976)
in which proofs provide the clarity necessary to identify inconstancies, leading to
the refining of the axioms and definitions the proof depends on.
76

via free access
THE ROLE OF PROOF
Communication
The publication of proofs in journals is an important, if not the primary way in
which new mathematics is communicated. Much university teaching also involves
the presentation of proofs with commentary to students as a way of communicating.
De Villiers (1990) quotes Volmink (1990) who writes:
Proof is a form of discourse, a means of communication among people doing
mathematics. (p. 8)
De Villiers goes on:
According to this view proof is a unique way of communicating mathematical
results between professional mathematicians, between lecturers and students,
between teachers and pupils, and among students and pupils themselves. (p. 22)
Getting “Theorem Credits”

Thurston (1995) mentions one other role of proofs beyond verification and expla-
nation: “in our credit driven system, [mathematicians] also want and need theorem
credits (p. 36). By “theorem credits” he means the social acknowledgement that
comes from publishing theorems, and in mathematics results must be published with
proofs to count.
We are driven by considerations of economics and status. Mathematicians,
like other academics, do a lot of judging and being judged. Starting with
grades, and continuing through letters of recommendation, hiring decisions,
promotion decisions, referees reports, invitations to speak, prizes...we are
involved in a fiercely competitive system. (p. 34, ellipses in original)
Other Roles of Proof in Mathematics

Several other roles that proofs play in mathematics have been identified, although
none has been advocated as being as significant to mathematics or to mathematics
education as those described above. Other roles include aesthetics, intellectual
challenge, construction of an empirical theory, clarification of a definition or the
consequences of an assumption, and incorporation of a fact into a new framework.
De Villiers (1990) briefly mentions the role of aesthetics: “This list of functions
is however by no means complete. For instance, we could easily add an aesthetic
function” (p. 23, emphasis removed). Rota (1997) states that both whole proofs and
a short step in a proof can be thought beautiful. He comments, “The most common
instance of beauty in mathematics is a brilliant step in an otherwise undistinguished
proof.” (p. 172). However, in trying to come to grips with what is meant by
“beauty” in mathematics, he relates it to “enlightenment” which we have treated
above under the heading of “explanation,” so perhaps the aesthetic role of a proof
may be another way of referring to its explanatory role.
De Villiers (1990) also mentions the role of personal self-realisation and later
expands on it.
77

via free access
CHAPTER 5
To mathematicians proof is an intellectual challenge that they find as appealing

as other people may find puzzles or other creative hobbies or endeavors. ... In
this sense, proof serves the function of self-realization and fulfillment. Proof
is therefore a testing ground for the intellectual stamina and ingenuity of the
mathematician (compare Davis & Hersh, 1983:369). To paraphrase Mallory’s
famous comment on his reason for climbing Mount Everest: We prove our
results because they’re there. Pushing this analogy even further: it is often not
the existence of the mountain that is in doubt (the truth of the result), but
whether (and how) one can conquer (prove) it! (de Villiers, 1999, p. 8)
Hanna and Jahnke (1996) suggest construction of an empirical theory as a role of
proof. For example, they describe Newton’s proof of Kepler’s laws on the basis of
his own law of gravity, as a proof that played an important role in the construction
of gravity as an empirical theory. Newton’s proof did not verify Kepler’s laws as
the law of gravity was a less certain foundation than the empirical evidence for
Kepler’s laws at that time. Instead, it might be said that the proof increased
confidence in the Newton’s law of gravity; “the law of gravity [acquired] credibility,
by being applied to Kepler’s laws.” (p. 895). In other words confidence in Kepler’s
laws led to increased confidence in Newton’s theory because his theory could be
used as a deductive basis for Kepler’s laws.
Hanna and Jahnke also suggest that a similar process of verifying assumptions
by proving established theorems can occur in purely mathematical contexts,
leading to exploration of the meaning of a definition or the consequences of an
assumption as a possible role of proof. This was the case with twentieth century
attempts to define axioms for set theory and arithmetic that were sufficient to prove
the theorems already accepted in those fields. A set of axioms and definitions was
acceptable if it could be shown to prove all the desired theorems (completeness)
without giving rise to any contradictions (consistency). But another way in which
proofs could be involved in exploration of the meaning of a definition occurs in
cases where a surprising result is proven, leading to a questioning, and possible
refining of a definition. Hanna and Jahnke mention the proof of the equivalence of
the sets of natural numbers and square numbers. “Rather than transferring truth
from the conditions to the proved theorem, the proof actually has the effect of
calling the conditions themselves into question” (p. 896). Lakatos’s (1976) process
of proof-analysis seems to us to be another case where proof (or in his case disproofs
by counterexamples) lead to questioning and refining of definitions and assumptions.
The role of proof in exploration of the meaning of a definition or the consequences
of an assumption is undoubtedly an important one, which should be acknowledged,
perhaps in combination with more traditional ways in which proof leads to new
discoveries.
Finally, Hanna and Jahnke (1996) mention incorporation of a well-known fact
into a new framework and thus viewing it from a fresh perspective as a role of
proof. They do not give any examples, but the fundamental theorem of calculus
could be one. By showing that integrals are anti-derivatives, the proof repositions
numerous facts about the derivatives of various functions as facts about functions
whose integrals are known.
78

via free access
THE ROLE OF PROOF
WHAT IS PROVING IN SCHOOL MATHEMATICS?
The comments in the previous section show that proof plays many roles in mathe-
matics. However, in school mathematics, as represented by the practices of most
teachers, the only role for proof seems to be verification.
With very few exceptions, teachers of mathematics seem to believe that a
proof for the mathematician provides absolute certainty and that it is therefore
the absolute authority in the establishment of validity of a conjecture. (de
Villiers, 1990, p. 18)
In describing what meaning they ascribed to the notion of proof in general ...
the majority of the teachers (11) stated, to varying degrees, that a proof is a
logical or deductive argument that demonstrates the truth of a premise. ...
Other teachers (6) ascribed a slightly more general meaning to proof, that of
proof as a convincing argument. For example, one teacher stated that proof is
“a convincing argument showing that something that is said to be true is
actually true” (KA). Overall, whether defining proof as a deductive argument
or as a convincing argument, teachers viewed proof as an argument that
conclusively demonstrates the truth of a statement. (Knuth, 2002b, p. 71)
This restricted role for proof may be influenced by the language of curriculum
documents. This definition from Alberta’s curriculum documents is typical:
Prove: to substantiate the validity of an operation, solution, formula or theorem
in general and to provide logical arguments for each step in the process
(Alberta Education, 1991, p. 5).
This meaning of proving is concerned with providing evidence, with substantiating
validity. Even in some so-called “reform” documents, the role of proof (or at least
deductive reasoning) is primarily verification.
A mathematician or a student who is doing mathematics often makes a
conjecture by generalizing from a pattern of observations made in particular
cases (inductive reasoning) and then tests the conjecture by constructing either
a logical verification or a counterexample (deductive reasoning). (NCTM,
1989, p. 143)
There may, however, be other roles for proof in classrooms. The NCTM’s (2000)
Principles and Standards advocates teaching proof to answer “Why does this
work?” not just “Does this always work?” (p. 58). De Villiers (1991b) reports the
results of a study in which he asked 205 prospective mathematics teachers why
one might prove a fact that is easily verified experimentally. While most (61%)
suggested verification as the role of proof in this case, other roles were also
suggested: systematisation (11%), explanation (7%) and developing logical thinking
(4%). The remainder provided no response or an unclassifiable response. Deve-
loping logical thinking is an interesting response as it is sometimes suggested as
the primary role of proof in schools:
The majority of the teachers (13) identified the development of logical
thinking or reasoning skills as a primary role proof plays in secondary school
79

via free access
CHAPTER 5
mathematics. Included within this category are teacher responses regarding a

role of proof being its applicability to the real world; the applicability role
was subsumed by the logical thinking category because the teachers discussed
logical thinking skills in terms of the value outside the domain of mathematics
as well as inside. (Knuth, 2002b, p. 78)
Herbst, Chen, Weiss and González (2009) also report teachers who see proof as
being taught primarily to develop logical thinking.
THE ROLES OF PROOF FOR STUDENTS
As we have seen above, proof has many roles in mathematics, but only verification
and the development of logical thinking seem to be significant ones from teachers’
perspectives. One might expect teachers’ views to be passed on to students, but
research has shown that this is not the case. Researchers have looked at the roles of
proof for students in several ways: In terms of the roles students ascribe to proof
when asked, in terms of the contexts in which students prove, and in terms of the
needs that are satisfied when students are shown a proof.
Healy and Hoyles’ (2000) survey (described in Chapter 4) included an open
ended question which asked the students to describe proof and its purposes. The
two roles mentioned most often were verification (50%) and explanation/commu-
nication (35%; Healy and Hoyles do not distinguish between explanation and
communication). Discovery/systematisation was mentioned by only 26 students
(1% of those sampled) (p. 417). They note, however that the interviews they
conducted suggest that some of the students whose responses could be classified as
“verification” might also view proof as explanation (p. 418). 28% of the students
gave no response, suggesting that the role of proof is unclear to them.
A number of authors have commented on what sense students might make of
being asked to write proofs, when they do not know what the role of proof might
be. Alibert (1988) comments that proof in school mathematics “is only a formal
exercise to be done for the teacher.” (p. 1–109). McCrone and Martin (2009) inter-
viewed students for whom “proofs were helpful for demonstrating some relationship
between components of a diagram or for showing the teacher that they understood
the content of the geometry theorems” (p. 218). Wheeler (1990) points out that to
students who are told that the role of proof is verification and who are asked to
prove statements they know to be true, proof reduces to “just a game because you
already know what the result is” (p. 3). Reid (1995b) calls this a “teacher-game”:
“A teacher-game is an activity that earns marks and acceptance, but is seen as
being otherwise useless.” (p. 17).
Reid (1995b) investigated what needs arose for secondary school and university
students during mathematical problem solving, and what kinds of reasoning they
used to satisfy those needs. He reported that students used deductive reasoning
(proving) spontaneously to explain, explore (discover), and verify. He also reported
some students engaging in a “teacher-game” with the interviewer in which the
students produced proofs or what Reid calls “formulaic” proofs that resemble
proofs but which are logically flawed.
80

via free access
THE ROLE OF PROOF
Chazan (1993) reports on US secondary school geometry students who preferred

deductive proofs to examples because they are more explanatory:
Nicole: This one [the deductive proof ] because it is more clear on why—why
it’s true
Int: Why is the two column proof better than measuring?
Nicole: Because I think it helps me to understand better why it’s true.
(p. 380)
Based on this, Chazan argues that instruction based on deduction as a means of
explanation might be more successful (p. 383).
In summary, students suggest verification and explanation/communication as
roles of proof when asked, they use proving to explain, explore, verify and to
play “teacher-games,” and they find proofs they are shown more explanatory than
examples. This suggests that students and teachers agree about the role of proof
(verification) when asked, but that students’ actions may indicate that other roles
are important to them.
POSSIBLE ROLES OF PROOF IN TEACHING
As noted above verification is the role of proof most often assumed in teaching.
However explanation and communication have also been proposed as possible, and
perhaps preferable, roles.
Fawcett (1938) reports a successful teaching experiment in which verification
was the role ascribed to proof. For Fawcett,
The concept of proof is ... involved in all situations where conclusions are to
be reached and decisions to be made. (p.120)
Fawcett’s teaching was focussed on geometry, but also included direct connections
to arguments in other contexts outside of mathematics. These contexts were those
in which there was a need to establish the truth of a statement and proof was seen
as a model for doing so. Fawcett reports success in teaching his students to prove,
but it would be a mistake to assume that the role he gave to proof (verification)
was an important factor in this success. Recall that in a large number of teaching
contexts (reported in Chapter 4) the role of proof was verification and the results
were unsatisfactory. However, Fawcett’s study provides an important counterexample
to claims that teaching based on proof as verification is necessarily unsuccessful.
The importance of illumination, understanding or explanation as the role of
proof in teaching has been emphasised by many authors (e.g., Chazan, 1993; de
Villiers, 1990, 1991ab; Hanna, 1989; Hersh, 1993). Few, however, have studied
teaching based on explanation as the role of proof. De Villiers (1991a) investi-
gated whether secondary school students, who had been convinced by multiple
examples of the truth of a statement, would feel a need for an explanation and
would accept a proof as an explanation. He reported a positive result. Since then
he has obtained similar results in dynamic geometry contexts (e.g., Mudaly & de
Villiers 2000).
81

via free access
CHAPTER 5
The studies cited above show that teaching proof based on the roles of verification
and explanation can be successful, however, more research is needed on this topic.
IDEAS FOR RESEARCH
Not many studies have examined the effects of giving a specific role to proof has
on students’ motivation to prove or their acceptance of proofs. One might speculate
teaching on the role of verification is ineffective, on the basis of the lack of student
success in proving summarised in Chapter 4. However, many other factors might
come into play, and Fawcett (1938) reports a successful teaching experiment in
which verification was the role ascribed to proof. De Villiers (1991a, Mudaly &
de Villiers, 2000) has looked at explanation as a motivation and reports that proofs
are accepted by students seeking an explanation. Other roles remain to be explored,
and given the complexity of teaching proof the roles of verification, explanation
and communication must be examined further.
An assumption in the literature and in this chapter is that the practice of
mathematicians and the role of proof in professional mathematics should have
some bearing on the teaching of proof in schools. Exactly what bearing professional
practice should have, however, is not clear. In fact, given that many students will
have no contact with mathematics after they leave school, it might be argued that
the practices of those who use mathematics (e.g., scientists, tradespeople, engineers,
medics, etc.) should be more significant. From that perspective a role such as
developing logical thinking might be more important than verification or explana-
tion. On the other hand, school mathematics could be seen as similar to subjects in
which students are prepared not to practice in the field, but rather to appreciate the
products of it. In that case a focus on proof reading rather than writing, and a
greater emphasis on roles such as aesthetics might be called for.
82

via free access
CHAPTER 6
TYPES OF REASONING
In this chapter we will describe different types of reasoning that are relevant to
teaching and learning proof. Types of reasoning have been described in the lite-
rature on philosophy and logic, and we will draw on that literature to make some
initial comparisons. However, human thinking is more complex than the abstractions
of logic can account for, so we will elaborate our descriptions with reference to the
thinking of students. We will also make specific links to mathematics education
research that has made special reference to one or more types of reasoning.
Recall that in the introduction to Part 2 we referred to Reid’s model of the
relationship between needs, reasoning and proof. In this chapter we will refer back
to that model when describing each type of reasoning, in order to note connections
to needs.
– What kinds of reasoning are most important to be aware of when teaching proof ?
– How does the terminology used to describe reasoning in the proof research
literature affect how one interprets that literature?
Four types of reasoning will be our focus here: deductive reasoning, inductive
reasoning, abductive reasoning and reasoning by analogy. One way of distinguishing
between these is by looking at how they use cases, rules, and results. A case is a
specific observation that a condition holds. A condition describes an attribute of
something, or a relation between things. The statement “Chino is a dog” is a case,
in which being a dog is the condition. A rule is a general proposition that states that
if one condition occurs then another one will also occur. “Dogs are animals” is a
rule. The conditions “being a dog” and “being an animal” are linked. A result is a
specific observation, similar to a case, but referring to a condition that depends on
another one linked to it by a rule. “Chino is an animal” is a result in this example.
In deductive reasoning a case and a rule imply a result. “Chino is a dog” and
“Dogs are animals” imply “Chino is an animal”. In inductive reasoning a case and
a result (or many similar cases associated with many similar results) lead to a rule.
“Chino is a dog” and “Chino is an animal” lead to “Dogs are animals”. In abductive
reasoning a result and a rule lead to a case. “Chino is an animal” and “Dogs are
animals” lead to “Chino is a dog”.
Symbolically these three types of reasoning can be shown like this:
A ∧ (A → B) ⇒ B
A ∧ B ⇒ (A → B)
B ∧ (A → B) ⇒ A
83

via free access
CHAPTER 6
There is a nice symmetry to these expressions that suggests that they encompass
all possibilities. And many authors, following Peirce (see e.g., 1867, 1878), have
focussed on deductive, inductive and abductive reasoning. However, thinking does
not always fit nicely into the abstract patterns suggested by logical symbols.
Reasoning by analogy involves using a well known situation to assert something
about a less well understood situation. It can either go from a case to another case,
or from a rule to another rule. For example, if you know that “Chino is a dog” and
you hear someone talking about taking Chino and Tressie for a walk, you might
conclude “Tressie is a dog” by an analogy from one case to another. Going from
“Dogs are animals” to “Cats are animals” based on a sense that cats are similar to
dogs is an analogy from one rule to another. Symbolically, reasoning by analogy
looks much different from the other types of reasoning:
(A ≈ C) ∧ A ⇒ C
(A ≈ C) ∧ (B ≈ D) ∧ (A → B) ⇒ (C → D)
In the next four sections we will describe each of these four types of reasoning:
deductive, inductive, abductive and by analogy, in more detail. In each section we
will include examples from students’ reasoning and point out relationships between
types of reasoning. We will then consider some other types of reasoning that are
relevant to teaching and learning proof.
DEDUCTIVE REASONING
We will begin our descriptions with deductive reasoning, as it is often considered to

be the basis of proof, and hence the development of students’ deductive reasoning is
a goal of teaching proof. In fact, for those whose research perspectives (see Chapter 3)
include a narrow category of “proof ”, and for some of those with a broader category,
deductive reasoning is a requirement of proof. There are many places where one
can read about deductive reasoning. From the syllogisms of the Greeks to modern
symbolic logic, deductions have been the main focus of the formal study of logic.
This is because deduction is the only kind of reasoning which is thought to
establish certainty.
As we noted above, the simplest deductions involve a rule and a case, from which
we conclude a result. They can be characterised by syllogisms. Lewis Carroll’s
Symbolic Logic (1897/1958) includes this one:
All cats understand French
Some chickens are cats
Some chickens understand French
p. 57
In this example the rule is, “All cats understand French,” the case is, “Some
chickens are cats,” and the result is, “Some chickens understand French.” Note that
in this example the case is a general rule as well. When the case is a specific case
84

via free access
TYPES OF REASONING
and not a general rule then the deduction is a “specialisation” of the rule. The
following famous syllogism is an example of a specialisation:
All men are mortal
Socrates is a man
Socrates is mortal
These two syllogisms are of a type called “modus ponens” or “affirming the
antecedent”. Another form of deductive reasoning is “modus tollens” or “denying
the consequent”.
All men are mortal
Socrates is immortal
Socrates is not a man
The research literature suggests that reasoning using modus tollens is more difficult
than using modus ponens, but that students in elementary school can learn, and
usually do learn, to reason in these ways (For a review of this literature, see
Stylianides & Stylianides, 2008. For interesting examples of elementary school
students’ reasoning see Stylianou, Blanton & Knuth, 2009.).
Simple deductions like these are common in the speech of students and because
they are so common, they are not all that interesting for researchers and teachers
interested in students learning proof. More complex chains of deductions are
involved in proving, and are deductive reasoning worth considering.
Our first example of such a chain is Maya’s explanation to her peers of her
method of solving the problem of determining the number of squares in an n by n
grid (the Count the Squares problem, see Figure 7). Maya and the other students
(below) who worked on the Count the Squares problem are grade 5 students in
Vicki Zack’s classroom. See Zack (1997, 1998, 1999ab) for other episodes and
interpretations involving some of the same children.
In these transcripts an em dash (—) indicates a silence of about one second.
Ellipses (...) indicate omitted speech. Three question marks (???) indicate inaudible
speech or a guess at partly audible speech.
Maya is explaining why the number of squares of a certain size in a 10 by 10
grid is always a square number, and more specifically, that the square numbers
involved are 100, 81, 64, …, 4, 1. At the beginning she is gesturing to show how a
2 by 2 square fits across the top of the grid 9 times.
Maya: Can everyone see? So you count 1, 2, 3, 4, 5, 6, 7, 8, 9 right? …
Since a square—this—any square—the square is 10 by 10 no
matter how you turn it, it’s always going to be…the same. So you
don’t have to measure it again. You can go 9 times 9. Do you
understand why? Yeah? OK, So you go 9 times 9 like Gino said, 81.
Then you can do 3 by 3, 1, 2, 3, 4, 5, 6, 7, 8—and then again you
don’t have to measure again you know. It’s going to be the same.
So 8 times 8—64. And you can keep on going.… You can do the
7 times 7—49. And 6 times 6—36. 5 times 5—25. 4 times 4—16.
3 times 3—9.
85

via free access
CHAPTER 6
Find all the squares in the figure on the left. Can you
prove that you have found them all?
What if it were a 5 by 5 square? Can you prove

you have them all?
Extension: What if it were a 10 by 10 square? What
if it were a 60 by 60 square?
Figure 7. The count the squares problem.
Her reasoning can be presented as follows:
There are nine 2 by 2 squares across the top of the grid

“the square is 10 by 10 no matter how you turn it”
“You can go 9 times 9”
or more abstractly as:
There are nine 2 by 2 squares across the top of the grid

A grid has the same number of rows and columns of any size
square
There are nine rows of nine 2 by 2 squares in the grid
In both cases the argument fits the structure of deductive reasoning described
above: a case and a rule imply a result.
Figure 8. Drawing used by Bill and John when proving that

the sum of two odd numbers is even.
86

via free access
TYPES OF REASONING
Deductive reasoning in classrooms is sometimes quite complex and difficult to

analyse. In the following transcript two 15 year old students deduce that the sum of
two odd numbers is an even number (for context see Reid, 1995b). They have just
worked through proving that the sum of two even numbers is even and that the sum
of an odd number and an even number is an odd number, using a pictorial model
which they then formulated algebraically.
1 Bill: An odd plus an odd. ... — — — ... OK this would be — M N — ... And
this would be 2M plus 1. And this would be 2 N plus 1. Therefore, for this
you’d have to have 2. Oh sorry. M. ... Then would it be M? plus N. And
this would be plus 2. Since ... an even, which is ... 2 M ... plus N, ... plus 2.
Since it’s ... 2 more than an even it would be an even again. Right?
2 DR: Why would 2 more than an even number be even?
3 Bill: ’Cause, uh, um, —
4 John: They just go odd even odd even so you go 2 ... on to the odd then add
one more you go back to the even.
5 DR: OK. That sounds good. — If I have an even number and I add 4 to it will
it be even? Will my answer be even?
6 Bill: Umhmm. Because that one’s still an even.
7 DR: OK.
8 Bill: An even added to an even gives you an even. That’s what we explained
here [referring to their previous work].
Their argument can be recast in this form:
An odd number is one more than a multiple of 2 (Rule, based on the
definition of “odd,” expressed as “2n+1”)
The sum of two odd numbers is the sum of two even numbers plus one plus
one (from the previous statement, expressed as “2n+1 + 2m+1 = 2n+2m+1+1”)
2n + 2m = 2(n+m), which is an even number (implicitly using the distributive
law and the definition of “even, ” Bill arrives at this point at the end of line 1)
2(n+m) + 2 is even (from either their previous work, or the alternating pattern
of odd and even described by John in line 4)
Because deductive reasoning is thought to establish certainty, it is often associated
with verification. However, deductive reasoning can also play other roles. In the
examples above deductive reasoning is used to explain to an audience. In both
cases there is no serious doubt of the truth of the statement, but further explanation
is being requested. Reid (1995b, 1996b) also provides examples of students using
deductive reasoning to explore.
It is sometimes claimed that deductive reasoning cannot be used to explore
because, in theory, all the information is already present in the premises. Deductive
arguments are such that “none of them much advance your knowledge of the truth”
(Peirce, 1902, CP 2.189; 1955, p. 127). Note however that theory is different from
practice, and that deductive reasoning can lead to the experience of discovering a
new truth. Consider the following metaphor: If you are given a coded message and
the key to the code, you are in possession of the message (that is, you require no
87

via free access
CHAPTER 6
more information to decode it). You could give the coded message and key to some-
one else and correctly say that you had passed the message on to them. However,
possessing the message is not the same as knowing what it is, and the act of deducing,
like the act of decoding, can make what was implicitly known into something
which is explicitly known, an experience not unlike learning something new.
Maya’s fellow students (above) knew that there are nine 2 by 2 squares across the
top of a 10 by 10 grid, and they knew that the number of rows and columns in a
square grid are the same, but her conclusion that one can find the total number of
2 by 2 squares by multiplying 9 times 9 may not have been something they knew
before she presented her argument.
While deductive reasoning has been described in basically similar ways over a
long history, the other three types of reasoning we will describe have not been, and
so the next three sections will involve making further distinctions and clarifications
that were not needed to describe deductive reasoning. In the next section we will
begin with inductive reasoning.
INDUCTIVE REASONING
“Inductive reasoning” is another of those terms in mathematics education which is

widely but inconsistently used. This reflects a long history in philosophy of
distinguishing between deductive reasoning and reasoning that is not deductive,
especially the reasoning used in the sciences. There are many different ways for
reasoning not to be deductive and scientific reasoning is complex, so it is not
surprising that “inductive reasoning” is defined and used in a number of ways. In
this section we will identify some of the ways of reasoning called “inductive” and
offer more precise terms for them.
Recall that deductive reasoning is thought to have these three characteristics: It
applies a general rule to conclude a specific result; it leads to no new knowledge
(though it might lead to the experience of knowing something new); and it establishes
certainty. Opposing each of these in turn leads to a possible characterisation of
inductive reasoning:
– Inductive reasoning proceeds from specific cases to conclude general rules.
– Inductive reasoning uses what is known to conclude something previously
unknown.
– Inductive reasoning is only probable not certain.
Here we will briefly mention some philosophers and educators who have empha-
sised one of these characteristics in defining inductive reasoning, before introducing
a more refined terminology.
From Specific to General

The dichotomy between deductive reasoning and reasoning which is not deductive
can be traced back to Aristotle. He states (in Topics, Book I, Chapter 12):
Having made these distinctions, we must distinguish how many species there
are of dialectical arguments. There are induction and deduction. Now what
88

via free access
TYPES OF REASONING
deduction is has been said before; induction is a passage from particulars to

universals, e.g. the argument that supposing the skilled pilot is the most
effective, and likewise the skilled charioteer, then in general the skilled man
is the best at his particular task. (Barnes, 1984, p. 174)
Aristotle is clear that there are only these two types of reasoning and that inductive
reasoning is characterised by proceeding from specific cases to general rules.
Already in his example, however, we can see that the situation is not so simple. He
moves from two general rules (one about pilots and the other about charioteers) to
a general rule about a larger class (men) that both pilots and charioteers belong to.
And there are also types of deductive reasoning that proceed from specific cases
to general rules (e.g. proofs by exhaustion) which we will discuss in Chapter 7.
Nonetheless, Aristotle’s dichotomy and his emphasis on specific cases as the
basis for inductive reasoning continue to be influential. For example the NCTM
Standards (1989) state:
a logical verification or a counterexample (deductive reasoning). (p. 143)
From Known to Unknown

Aristotle added a further condition to his description of inductive reasoning later in
his Topics (Book VII, Chapter 1), that inductive reasoning starts with what is
known and concludes with something previously unknown. In other words it
involves discovery of new knowledge.
Induction should proceed from individual cases to the universal and from the
familiar to the unknown. (Barnes, 1984, p. 262)
Topics is thought to be one of Aristotle’s earliest works on logic (von Wright, 1965,
p. 8). In his later Posterior Analytics (Book I, Chapter 1) Aristotle again emphasises
that induction proceeds from the known to the unknown. It comes “from already
existing knowledge” and it proves “the universal through the particular’s being
clear” (Barnes, 1984, p. 114).
John Stuart Mill also makes mention of proceeding from the specific to the
general in his definition of inductive reasoning, but he emphasises that inductive
reasoning is a type of inference, that is, reasoning that allows for the discovery of
something new.
Induction, then, is that operation of the mind, by which we infer that what we
know to be true in a particular case or cases, will be true in all cases which
resemble the former in certain assignable respects. In other words, Induction
is the process by which we conclude that what is true of certain individuals of
a class is true of the whole class, or that what is true at certain times will be
true in similar circumstances at all times. ... Induction, as above defined, is a
process of inference; it proceeds from the known to the unknown. (Mill, 1884,
p. 210, Book III, Chapter 2, section 1)
89

via free access
CHAPTER 6
Von Wright (1965) argues that inductive reasoning need not proceed from the
specific to the general, but that it must involve moving from the known to the
unknown.
We may also extend the conclusion to a limited number of unknown members
of the class, e.g., to the next member which turns up, thus proceeding from
particulars to a new particular. Both cases of inductive inference, that from
particulars to universals and that from particulars to particulars, are covered
by the definition of induction as reasoning from the known to the unknown.
(p. 1)
As we noted above, however, deductive reasoning can certainly lead to the experience
of discovering new knowledge, even if that knowledge was implicit in the premises.
This suggests that defining inductive reasoning as any reasoning proceeding from
the known to the unknown might be confusing in practice.
Probable not Certain

Deductive reasoning is the only reasoning that is thought to lead to conclusions
with certainty. This suggests another definition of inductive reasoning as reasoning
that leads to conclusions without certainty. Such a definition has been used (e.g., by
Smith & Henderson, 1959) but it has the flaw that it includes under the label
“inductive” many types of reasoning with quite different characters. This has led
some authors to invent new terms for the dichotomy between reasoning with
certainty and reasoning without. The terminology of authors who have been cited
in mathematics education are summarised in Table 15.
The three historically established ways of characterising inductive reasoning
mentioned above: reasoning from the specific to the general, from the known to the
unknown and with probability but not certainty, are not in practice useful for
describing students’ reasoning in mathematics classes. To be clear about what kind
of reasoning is being described more precise terms than “inductive” need to be
introduced. In the following we will do so.
Table 15. Reasoning dichotomies based on certainty
With certainty Without certainty

Smith & Henderson, 1959 deductive inductive
Peirce, 1902 A-reasonings B-reasonings
Polya, 1954 demonstrative plausible
Balacheff, 1987 conceptual pragmatic
Duval, 1991 deduction argumentation
(see Chapter 8)
90

via free access
TYPES OF REASONING
More Precise Terms for Reasoning usually called “Inductive”

Drawing conclusions given a sequence of numbers is often described as inductive
reasoning. However, there are several different kinds of reasoning that might occur,
leading to different conclusions. For example, consider the different conclusions one
might draw given the following sequence of numbers:
1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144
One might conclude:
– Many of these numbers are odd.
– Taken in groups of three these numbers display the pattern ‘odd, odd, even’.
– The next triple of these numbers, if the sequence is extended, will also display
the pattern ‘odd, odd, even’.
– If the sequence is extended, all triples of these numbers display the pattern ‘odd,
odd, even’.
In the first two conclusions, an observation has been made that several specific
cases share a common feature. There is not a standard term for such a conclusion.
Mill (1884) calls it “colligation”.
The descriptive operation which enables a number of details to be summed
up in a single proposition, Dr. Whewell, by an aptly chosen expression, has
termed the Colligation of Facts. (p. 214, Book III, Chapter 2, section 4)
However aptly chosen this expression is, it has not caught on. Polya (1954, p. 5)
instead calls such reasoning noticing a similarly and Reid (2002a) calls it pattern
observing.
The third conclusion “The next triple of these numbers, if the sequence is
extended, will also display the pattern ‘odd, odd, even’” is an example of what
Mill (1884) and von Wright (1965) call induction “from particulars to particulars”.
In it a prediction is made about another specific case, but no general claim is made.
Johnson (1924) coined the term “eduction” for such reasoning. Again, this term has
not caught on and we would prefer to call such reasoning predicting, because it
makes a claim about the next case based on past cases.
In the fourth conclusion “If the sequence is extended, all triples of these numbers
display the pattern ‘odd, odd, even’” a general claim is made, and this fits within all
the historical characterisations of inductive reasoning mentioned above. However,
calling it inductive conceals an important ambiguity. What epistemic value (see
Chapter 5) does the general conclusion “All triples of these numbers display the
pattern ‘odd, odd, even’” have? Is it a conjecture requiring further verification
because significant doubt remains, or is it a generalisation that requires no further
verification as sufficient certainty now exists? Logically, it must be a conjecture,
but as such reasoning is generally said to be the basis for science, this leads to the
“problem of induction”, that is, establishing some basis for accepting the conclusions
of inductive reasoning as valid, or at least more valid than the conclusions of
non-sciences like astrology. In practice, given sufficient specific cases suppor-
ting the conclusion, most people would accept it as a generalisation. Thus, from
a psychological perspective (which must be relevant to learning) the conclusion
91

via free access
CHAPTER 6
might be either a conjecture or a generalisation given the evidence. We will use the
word conjecturing to refer to making a general statement from specific cases when
the general statement requires additional verification, and generalising when the
general statement does not require additional verification. Note that the purpose of
inductive reasoning differs in these two cases. Conjecturing addresses a need to
explore. Generalising is used to explore, but also to verify at the same time. Neither
explains.
Let us apply the terms we have introduced so far: pattern observing, predicting,
conjecturing and generalising, to another example. Consider these sums:
6 = 3+3 16 = 3+13 = 5+11
8 = 3+5 18 = 5+13 = 7+11
10 = 3+7 = 5+5 20 = 3+17 = 7+13
12 = 5+7 ...
14 = 3+11 = 7+7 30 = 7+ 23 = 11+19 = 13+17
from Smith & Henderson (1959, p. 123); see also Proof 11 in Chapter 7.
Polya (1954) uses a similar example to discuss inductive reasoning.
You might observe a pattern: The numbers to the far left of the equals signs are all
even. The sums to the right are sums of two odd numbers. So far, this is not astoni-
shing as we know that the sum of two odd numbers is an even number. But there is
something special about these odd numbers. They are all prime.
Now that we have observed some patterns, we might make a prediction:
22, 24, 26 and 28 can be written as the sum of two odd primes.
Or perhaps if we are feeling braver:
All the even numbers up to 100 can be written as the sum of two odd primes.
In both cases we are reasoning from specific cases to other specific cases. No general
statement has been made.
If we are feeling braver still we might make a conjecture:
All the even numbers can be written as the sum of two odd primes.
This is a general statement but until you feel certain it is true, it remains a conjecture
for you, not a generalisation.
While the four terms we have introduced here: pattern observing, predicting,
conjecturing and generalising, describe some of the reasoning often called
“inductive” there is another important kind of reasoning also called “inductive”
that we will consider next.
Testing
Another process that is sometimes included under the label “inductive reasoning”
(e.g., Smith & Henderson, 1959, p. 132) is what we call “testing”. We do not mean
to refer here to the more general everyday meaning of proof which we called
“testing” in Chapter 2. Instead we wish to use “testing” in a technical sense to refer
to a specific type of reasoning which is used to test predictions and conjectures.
92

via free access
TYPES OF REASONING
Testing a prediction is straightforward. For example, we might want to test the

following prediction: “22, 24, 26 and 28 can be written as the sum of two odd
primes.” All we have to do is try to write 22, 24, 26 and 28 as the sum of two odd
primes. It does not take long to do this. In fact we can do it in more than one way:
22 = 11+11 = 5+17 = 3+19 26 = 13+13 = 7+19 =3+23
24 = 11+13 = 7+13 = 5+19 28 = 11+17 = 5+23
Now that we have tested our prediction we can be certain that it is true. In this case
testing amounts to proof by exhaustion (see Chapter 7). Testing our second prediction
(about all even numbers up to 100) is a bit more exhausting, but just as straight-
forward and certain.
Testing a conjecture (as opposed to a prediction) is a slightly different process.
First we need to specialise from our conjecture to generate a specific case. Recall
that specialisation is a simple kind of deductive reasoning described above (page 85).
The four specific cases we just tested are some possibilities. After generating them
through specialisation, we then test them as we did above, and in the end we
find ourselves more convinced (or at least not more doubtful) of the truth of our
conjecture. The more specific cases we test the more confident we become,
although this makes no sense in terms of probability.
If we want to estimate the probability that our prediction for even numbers up to
100 is true, we could use sampling instead of proof by exhaustion. The 13 specific
cases we have examined so far can be seen as a sample of the entire population of
even numbers up to 100 and the probability that our prediction is true is related to
the size and representativeness of our sample. In this case a larger sample, and a
randomly selected one (not just the even numbers up to 30) would be better, but in
any case the sample which we have gives us some confidence, and the larger the
sample, the greater our confidence becomes, until finally our sample includes the
whole population and our confidence is complete.
In the case of a generalisation encompassing an infinite number of cases the
situation is much different. How ever many cases we examine, our sample will be
both biased and woefully inadequate. The bias comes about because there will be
an infinite number of cases larger than the largest case we examine. With a finite
population it is clear that only looking at cases at the lower end is biased but in the
infinite case we cannot avoid it. And with a finite population we can look at a 10%
sample, or a 50% sample and it is clear that the 50% sample is better. With an infinite
population, however, any sample includes 0% of the population. Nevertheless,
experience tells us that the more specific cases we examine the more confident we
become.
Of course, if any of our tests turned up an even number that could not be written
as the sum of two odd primes, our conjecture would be false. In fact, it is. Consider
these two even numbers: 4 and 2. Four can be written as 2+2 or 1+3. In other
words as the sum of two even primes, or as the sum of an odd prime and 1. If 1 is
considered a prime then this does not contradict our conjecture, but 1 is usually not
considered a prime. If it is not, then our conjecture is false. However, we can fix it
easily enough by removing the condition that the two primes be odd. This
procedure, of modifying a conjecture slightly in the face of a counterexample, is
93

via free access
CHAPTER 6
analysed in detail by Lakatos (1976). In this case we substitute a slightly weaker

conjecture that remains true for the specific case of 4.
All the even numbers can be written as the sum of two odd primes.
Two can only be written as 1+1. No modification of our conjecture can save it in
this case. Instead what must be done is to bar the exception from the domain under
consideration:
All the even numbers greater than 2 can be written as the sum of two odd
primes.
Once we’ve started barring exceptions, we may as we return to the original,
stronger, conjecture:
All the even numbers greater than 4 can be written as the sum of two odd primes.
This is the form in which the conjecture (called the Goldbach conjecture) is often
given.
The Goldbach conjecture has been tested in many cases (over 1017 as of 5 January
2010, Oliveira e Silva, 2010). For anyone other than a mathematician that number
of cases is sufficient to make the conjecture a generalisation. And in fact, from a
psychological perspective, the Goldbach conjecture is a generalisation. However,
from a mathematical perspective, it is not yet a theorem.
We have now described five types of inductive reasoning: pattern observing,
predicting, conjecturing, generalising, testing, and some of the relationships between
them. We will now consider how these types of inductive reasoning can be used to
describe students’ mathematical activity.
Inductive Reasoning in Students’ Mathematical Activity

Several types of inductive reasoning can be seen in the episode of students’ mathe-
matical activity below. They include pattern observing, conjecturing, testing, and
generalising. The episode comes from discussions in a small group of three
students, comparing their solutions to the original 4 by 4 Count the Squares problem
(see Figure 7).
1 Will That’s what I did, but just to prove it was right I noticed a pattern. ...
2 I haven’t tested it on different kinds of squares, but on the 1 times 1, like
1 and 1 there’s 16 of them, for the 4 times 4 there are 9, for the 9 times 9
there are 4 and for the 16 times 16 there’s 1 ...
3 You can connect the diagonals and they’ll equal ... If you connect it
diagonally you get 1... If you connect the 4 times 4 diagonally you get 4. ...
[See Figure 9]
Figure 9. Will’s Count the Squares pattern.
94

via free access
TYPES OF REASONING
4 Lew Where do you get 16 times 16?

5 Will Sorry about that, Just ignore these. I made a mistake. This is just the
total amount of squares. So there’s 1 in there, 4 in there, 9 in there. ... I’ll
just test this on another square now.
6 Ross That’s cool... good pattern.
7 Will And I used that just to help prove there’s-
8 Ross That’s very clever. How did you get that?
9 Will I just noticed it. ... I was pretty sure there would be a pattern so I was
keeping my eyes open and I found it.
[Will tests his pattern on another case, a 5 by 5 grid.]
10 ... 25 of 1 ...the next size is 4 ... OK 16 of the 4s ... So the next biggest
size will be 9 ... There are nine, which makes sense because there’s -
would be an odd number of these so the nine should be nine.
11 So the next number we get here should be 16.... so 16, ... so look it’s
equal. So how many 16s? ... 4 of the 16s. Now the next biggest ... would
be 25 and there’s only 1. Equals, equals equals equals and equals
12 Lew Very clever
13 Will The chances are if it works for those then it works for the rest of them...
Will explicitly states that he observed a pattern (“to prove it was right I noticed
a pattern,” line 1) and he describes it to Ross and Lew. Will’s pattern matches the
number of unit squares in each size of square with the number of squares of
another size. 16, the number of unit squares in a 4 by 4 square, is linked to 16, the
number of 1 by 1 squares. When he says, “You can connect the diagonals and
they’ll equal” (line 3) he explicitly states the pattern that he has observed. His phrase
“to prove it” is probably meant in the everyday sense of “to verify it,” indicating
that is the need his reasoning satisfies. That observing a pattern is a way of verify-
ing for Will makes sense given the history of problem solving he has had (see Reid,
2002a). He expects there to be patterns in mathematics, so when he finds one it acts
as a confirmation of the accuracy of his work.
When Ross asks how Will arrived at his observation, Will necessarily responds
in a vague way: He “just noticed it. ... I was pretty sure there would be a pattern so
I was keeping my eyes open and I found it” (line 9). As Polya’s (1968) discussion
makes clear, pattern observing is not a systematic activity, although Will’s
expectation of patterns no doubt contributed to his observation of one.
Will not only observed a pattern in the 4 by 4 case, he also made a conjecture
about all cases of the Count the Squares problem: “you can connect the diagonals
[from the list of quantities to the list of sizes] and they’ll equal” (line 3). That he is
conjecturing is indicated in his comment “I haven’t tested it on different kinds of
squares” (line 2). A pattern that has been observed does not need to be tested. It
applies only to the cases being observed. A conjecture, however applies to all cases
and must be tested to increase confidence in it.
Recall that testing a conjecture consists of a specialisation and then testing the
specific case(s) produced to see if they are true. In this case Will’s conjecture is
specialised to the 5 by 5 case. Had he voiced it, it would have been something like:
“In the 5 by 5 case the diagonals connecting the list of quantities to the list of sizes
95

via free access
CHAPTER 6
will be equal” Before that specific case can be tested, a number of other speciali-
sations need to be made, for each number in the list of sizes. A comparison is then
made with the quantities of squares of each size in the 5 by 5 case, arrived at by
counting them, independently of the conjecture. Will’s statement “So the next
number we get here should be 16” (line 11) indicates one result of these two
specialisations. As his specialisations began with a conjecture, not a generalisation,
he is not certain of the result until he calculates the actual value and finds “it’s
equal”.
Once he has tested his conjecture, Will is prepared to generalise it as a rule.
“The chances are if it works for those then it works for the rest of them...” (line 13)
The same statement, “You can connect the diagonals and they’ll equal” was at first
an observation about the 4 by 4 case, and then a conjecture about all cases which
was tested by specialising it for the 5 by 5 case and then independently determining
the values for that case and comparing, and finally generalised to all cases on the
basis of that test. Will remains aware that his generalisation is not absolutely
certain, but “the chances are… it works for the rest of them” and that is sufficient
for him to base his subsequent reasoning on it.
Inductive Reasoning and Mathematics Education Research

In Chapter 4 we discussed two findings from research related to inductive reasoning:
Proof 6: Textbook proof from Chazan, 1993, p. 365
Statement Reason
1. AB = CD 1. Given.
2. AC = AC 2. Any number equals itself.
3.∠1 = ∠2 3. Alternate interior angles formed
by a line crossing two parallel lines
are equal.
4. Therefore ΔBAC ≅ΔDCA 4. SAS.
5. Therefore BC = DA 5. Corresponding sides of congruent
triangles have equal length.
96

via free access
TYPES OF REASONING
Chazan (1993) set out to explore why students behave in this way. He inter-
viewed seventeen students from five different classes, chosen on the basis of the
wide range of views they held concerning the use of examples in mathematics. They
were asked to comment on two arguments, a deductive proof from their textbook
(see Proof 6) and an argument based on four specific examples (see Proof 7).
Proof 7: Empirical argument from Chazan, 1993, p. 366
In any triangle, a line connecting the midpoints of two

of its sides is parallel to the third side.
∠ADE = ∠ABC ∠ADE = ∠ABC
∠ADE = ∠ABC ∠ADE = ∠ABC

Each of these four different sized triangles was measured and angle ADE was congruent to
angle ABC. Thus, in each case DE is parallel to BC. Therefore the statement is always true.
Chazan found that the students who accepted the empirical argument were not
totally naïve in their use of examples. For example, the four specific examples given
represent four different kinds of triangles: acute, obtuse, isosceles and right. Several
students commented that four examples of triangle all of the same type would not be
a convincing argument. Examples are convincing when they are representative exam-
ples. The must not have “special” features that make them poor representatives. A few
students noted that the four examples in the argument do not exhaust all types of
triangles (no equilateral triangle is included) so the argument could be improved. How-
ever, the basic idea of an argument by representative examples remained acceptable
for these students. And one student suggested that she went beyond the examples by
trying to picture all triangles of each type in her mind, using the examples as guides.
Thus students’ inductive reasoning is more sophisticated than one might think.
As we noted in Chapter 4 there are some good reasons students prefer empirical
97

via free access
CHAPTER 6
arguments, and they are not unaware of their limitations. And we will see in Chapter 7
that it is possible to identify more and less sophisticated arguments based on
inductive reasoning.
Inductive and Deductive Reasoning

Inductive and deductive reasoning are related in several ways. As we saw above,
deductive specialisation is part of testing. More importantly, conjecturing and
generalising are related to deductive reasoning.
We quoted above a common image of the relationship between inductive and
deductive reasoning:
a logical verification or a counterexample (deductive reasoning). (NCTM,
1989, p. 143)
Note that the NCTM use “generalise” and “test” differently from our usages above.
However, the process they describe can occur: A conjecture is made that an obser-
ved pattern is general, and then the conjecture becomes a generalisation after a
deductive proof is found. Whether this happens “often” either in the practice of
mathematicians or of students, is debatable, but this could be settled by empirical
research.
An example of such a process is the use of deductive reasoning to both verify
and explain why the conclusion “All triples of the numbers 1, 1, 2, 3, 5, 8, 13, 21,
34, 55, 89, 144 display the pattern ‘odd, odd, even’” comes about. Of course, one
must be given the recursive rule defining the sequence as a starting point. Bill and
John’s deductive reasoning about sums of odd numbers (above) came about in the
context of trying to use deductive reasoning to explain this conclusion (see Reid,
1995b).
Another way in which deductive reasoning is related to inductive reasoning
concerns the starting point of deductive reasoning. Recall that deductive reasoning
proceeds from general rules. Where do these general rules come from? Some can
be deduced from other general rules but there must be a starting point, the axioms,
which are general rules that are not deduced from other general rules. Where do the
axioms come from? The formalist position is that they are completely arbitrary, but
this is certainly not true in the history of mathematics. One possibility is that the
axioms are generalisations made by inductive reasoning. However, we will encounter
other possibilities in the next two sections.
Deductive reasoning is also related to inductive reasoning in a way that is more
of a confusion than a relationship. In mathematics there is a proof technique
called “mathematical induction” (MI), “complete induction” or sometimes simply
“induction”. The French phrase “raisonnement par récurrence” is less confusing
but it is rare to find English authors using “reasoning by recurrence”, however
clarifying that might be. Proof 8 is an example of a proof using reasoning by
98

via free access
TYPES OF REASONING
Proof 8: The sum of the first n integers (by MI)
The sum of the first n positive integers is n(n + 1)/2.

For n = 1 it is true since 1 = 1(1 +1)/2.
Assume it is true for some arbitrary k, that is,
S(k) = k(k + 1)/2. Then consider
S(k + 1) = S(k) + (k + 1)
= k(k + 1)/2 + (k + 1)
= (k + 1)(k + 2)/2
Therefore, the statement is true for k + 1 if it is
true for k. By mathematical induction,
the statement is true for all n.
Peano’s fifth axiom: If S is a set of numbers containing 0 and if the successor

of any number in S is also in S, then S contains all the numbers.
recurrence. “Mathematical induction” is based on deductive reasoning, not inductive

reasoning. The confusing terminology comes about because reasoning by recurrence
makes use of specific cases, and as we saw above inductive reasoning is sometimes
defined as reasoning beginning from specific cases. Reasoning by recurrence, how-
ever, also makes use of a general rule, first stated as an axiom by Peano (1858–1932).
A proof using reasoning by recurrence refers to a specific case (it shows that 0
is in a set), and to the relationship between specific cases (if a number j is in the
set, then so is j+1). But it also refers (usually implicitly) to Peano’s fifth axiom
which is a general rule allowing one to deduce a general rule from the specific
cases.
In this section we have described inductive reasoning as a broad category,
perhaps too broad to be useful for describing students’ reasoning. We introduced
five terms to refer to different types of inductive reasoning. We also noted that
inductive reasoning is important in mathematics education research because students
make use of it in context where deductive reasoning might be expected. Finally we
noted that deductive reasoning requires rules and inductive reasoning might be one
way that rules are arrived at.
It may be worthwhile at this point to look back to the guiding questions for this
chapter and to consider how more precise terms might affect your reading of the
research literature. In the next two sections we will be describing two other types
of reasoning, abductive reasoning and reasoning by analogy, that are sometimes
included within inductive reasoning. As you read these sections the question of the
usefulness of more precise terminology, especially for research and teaching, may
continue to be interesting to consider.
ABDUCTIVE REASONING
As we noted at the beginning of this chapter, abductive reasoning is related to

deductive and inductive reasoning. Like them it involves a case, a rule and a result
99

via free access
CHAPTER 6
linked in a particular order. Abductive reasoning can be seen as the reverse of

deductive reasoning or “reasoning backward” as Sherlock Holmes describes it in
Study in Scarlet:
Most people, if you describe a train of events to them, will tell you what the
result would be. They can put those events together in their minds, and argue
from them that something will come to pass. There are few people, however,
who, if you told them the result, would be able to evolve from their own inner
consciousness what the steps were which led up to that result. This power is
what I mean when I talk of reasoning backward.
The starting point for abductive reasoning is the observation of a surprising case. In
a detective story it might be that someone was murdered in a locked room. For a
physician it might be that someone has suddenly gone blind. For a scientist it might
be that Mercury does not follow the orbit Newtonian physics says it ought to. For a
mathematician it might be that the products 202×203×204, 483×484×485 and
757×758×759 are all multiples of six. Abductive reasoning leads to a new rule
which makes these specific cases less surprising.
The word “abductive” was introduced into the discussion of reasoning by
Charles Sanders Peirce (1839–1914). Most researchers in mathematics education
refer back to Peirce when describing abductive reasoning, so it is worthwhile
considering carefully what he meant by “abductive”. Peirce described abductive
reasoning in different ways at different times, and a comparison of his descriptions
reveals important aspects of abductive reasoning.
In his early work Peirce (c. 1867) emphasises the logical form of abductive
reasoning. At first he focused on syllogisms and on the role of what he calls
characters of specific cases and classes (we referred to “characters” as “conditions”
in our description of cases, rules and results, above). A case S (“this ball”) might
be a member of a class M (“the balls in this bag”) and have a number of characters
P', P'', etc. (“white” “made of wood” etc.). In 1867 Peirce described deductive,
inductive and abductive reasoning by the following syllogisms (CP 2.474,511; CE,
Vol. 2, pp. 27, 46):
Deduction Induction Abduction

M is P S' S'' S''', etc. taken at random as M’s Any M is, for instance, P' P'' P''', etc.
S is M; S' S'' S''', etc. are P; S is P' P'' P'''
S is P Any M is probably P S is probably M
S in these syllogisms is the subject, a specific case of interest, and S', S'', S'''
are a number of specific cases. P is the predicate, which describes a character of
the subject. P', P'', P''' are a number of characters. Peirce’s symbolic presentation
may be difficult to interpret. We have constructed the following example to
clarify the difference between inductive and abductive reasoning in Peirce’s 1867
description.
100

via free access
TYPES OF REASONING
Induction Abduction
Balls 1, 2, and 3 are chosen at Any ball in this bag is white, wooden, 1 cm across,
random from this bag. and smooth.
Balls 1, 2, and 3 are white. This ball S is white, wooden, 1 cm across, and smooth.
Any ball in this bag is probably This ball probably came from the bag.
white.
The difference between inductive and abductive reasoning in this formulation

has to do with whether a number of cases are found to share a character, leading
to the conclusion that all similar cases share that character (by inductive reason-
ning); or if a single case is found to share a number of characters with a class of
cases, leading to the conclusion that the case belongs to that class (by abductive
reasoning).
Later (c. 1878) Peirce rephrased his descriptions in terms of rules, cases and
results, and this description is often quoted (e.g., Arzarello, Micheletti, Olivero,
Robutti, 1998ab; Mason, 1996). His description is summarised in the following
syllogisms (1878, based on CP 2.623; CE, Vol. 3, pp. 325–326)
Deduction Rule – All the beans from this bag are white M is P
Case – These beans are from this bag S is M
Result – These beans are white S is P
Induction Case – These beans are from this bag S is M
Result – These beans are white S is P
Rule – All the beans from this bag are white M is P
Abduction Rule – All the beans from this bag are white M is P
(Hypothesis) Result – These beans are white S is P
Case – These beans are from this bag S is M
The differences between this formulation and Peirce’s formulation of 1867 are
slight, but significant. Instead of “These beans” sharing a number of characters,
only one character “being white” is involved in this canonical example. This
suggests that Peirce saw abductive reasoning as possible on very limited evidence,
perhaps because in examining instances of abductive reasoning in scientific
discovery, he encountered such situations. As well, the specific cases S', S'', S'''
enumerated in his 1867 formulation are now subsumed under a single subject
“these beans”. The specific nature of the cases is thus downplayed, allowing for
abductive reasoning in which the “case” is in fact a generality (in fact “These beans
are white” could be phrased as “All the beans in this sample are white,” which has
the same generality as “All the beans from this bag are white”).
Eco (1983) makes some useful distinctions based on Peirce’s 1878 formulation
of abductive reasoning. Eco describes abductive reasoning as the search for a
general rule from which a specific case would follow. He identifies three kinds of
abductive reasoning (the same three are described by Bonfantini & Proni, 1983,
using different terms).
101

via free access
CHAPTER 6
Given a specific case, the reasoner may be aware of only one general rule from
which that case would follow. This Eco calls “Hypothesis or overcoded abduction”
(p. 206). It is the same as Peirce’s 1878 formulation, except that Eco requires that
the case be specific. If there is more or less than one rule known to the reasoner,
then the situation becomes more complex. As Mason (1996) points out, “the tricky
part about abduction is locating at the same time the appropriate rule and the
conjectured case” (p. 37).
If there are multiple general rules to be selected from, Eco calls it “undercoded
abduction” (p. 206). This is difficult to structure as a syllogism, because while the
rule is known, and could therefore be treated as a premise, which rule to choose is
not known, and could be seen as a consequence.
It can also happen that there is no general rule known to the reasoner that would
imply the specific case in question. Then the reasoner must invent a new general
rule. This act of invention can also occur when there are general rules known that
would lead to the specific case, but they might be unsatisfactory for some reason.
Abductive reasoning that involves the invention of a new general rule Eco calls
“creative abduction”. Here the only premise of an abductive reasoning is the result,
and both the rule and the case are consequences of it.
Consider the physician’s patient who has suddenly gone blind. In addition to
that surprising specific case, the physician also has available a collection of general
rules of the form “If such and such is the case, then the patient will suddenly go
blind.” If the physician knows of only one such general rule, then the abduction is
Eco’s “Hypothesis or overcoded abduction.” For example, if the only suitable
general rule the physician knows of is “If a patient suffers a stroke in their visual
cortex, then the patient will suddenly go blind,” then this abduction can be made:
This patient suddenly went blind. (Result)
If a patient suffers a stroke in their visual cortex, then the patient
will suddenly go blind. (Rule)
∴ The patient suffered a stroke in their visual cortex. (Case)
If the physician has more than one general rule available that could account for the
patient’s sudden blindness, then the diagnosis, “The patient suffered a stroke in
their visual cortex,” is again an abduction, Eco’s “undercoded abduction”, but a
less reliable one, since the physician knows that there are other general rules which
would account for the result.
The scientist trying to account for the anomalies in the orbit of Mercury
provides an example for Eco’s third type of abductive reasoning. It is possible to
make use of the general rules already available concerning the motion of planets.
One hypothesis of this kind that was proposed was the existence of an unknown
planet close to the sun that was perturbing Mercury’s orbit. In this case the
reasoning is undercoded abduction. Many such hypotheses were proposed but in
the end it was a creative abduction, the creation of a new general rule (Einstein’s
theory of relativity), that accounted for the anomalies.
While Peirce, in his early work, emphasised the differences in logical form
between deductive, inductive, and abductive reasoning, in his later work (e.g.,
102

via free access
TYPES OF REASONING
“Lectures on Pragmatism”, 1903, CP 5.14–212) he shifts the emphasis to the

function satisfied by each kind of reasoning. The logical form of abduction has
been reduced to the form given at the beginning of this chapter:
B ∧ (A → B) ⇒ A
Abductive reasoning for Peirce has become identified with “explanatory hypo-
thesis” and his criteria for abductive reasoning to be good abductive reasoning
has come to include, at least, that it “must explain the facts” (1903, CP 5.197). It
has become part of a process of inquiry in which abductive, deductive, and
inductive reasoning play particular roles in a special order. Abductive reasoning
explains by introducing a new rule, deductive reasoning draws necessary conclu-
sions from the consequent of the abductive reasoning, inductive reasoning evaluates
the consequent by comparing the conclusions draw from it to experience (1908,
CP 6.469–476). In terms of the roles discussed in Chapter 5: abductive reasoning
explains and explores; inductive and deductive reasoning verify. Note also that
the nature of inductive reasoning has shifted in Peirce’s writing from what we
called above conjecturing to testing. This shift in emphasis from logical form to
the role addressed by reasoning can lead to confusion, as we have seen in this
chapter and in Chapter 5 that it is far from clear that each type of reasoning is
always associated with the same role, or that each role is addressed by only one
type of reasoning.
Examples of Abductive Reasoning

Below we will give four examples of abductive reasoning in students’ mathematical
activity. Each example illustrates different formulations of what abductive reasoning
is, classified by Peirce’s and Eco’s descriptions, and different roles abductive
reasoning plays in mathematical activity. As is evident from Table 16, all of Eco’s
types, and abductive reasoning fitting both of Peirce’s early descriptions, can be
observed. Peirce’s later description focusses on the role of the reasoning and
specifies the logical structure less precisely than in his earlier descriptions, so the
Handshake problem example in which the role of the abductive reasoning is to
explain and explore also fits his later description.
Table 16. Types of abductive reasoning used in examples
Peirce Eco Role

Pythagorean theorem 1878 undercoded exploring
Exploring
Handshake problem 1867 creative
explaining
Football field problem 1878 creative exploring
Count the Squares problem 1878 overcoded explaining
103

via free access
CHAPTER 6
Abductive reasoning in proving the Pythagorean Theorem. In this example a

teacher in France is guiding a level 4 (age 13–14) class through a proof of the
Pythagorean Theorem. They have concluded that ABCD is a rhombus, because it
has four equal sides (see Figure 10). The transcript is our translation from Knipping
(2003b, p. 158).
Figure 10. Diagram used in French classroom for Pythagorean Theorem proof.
1 T: It’s a rhombus. That’s a sufficient condition to have a rhombus. There is no

need for anything else. … But that’s not what I asked you to prove. I asked
you to prove that it’s a …
2 S: Square.
3 T: Square. So, under what condition is a square, uh, is a rhombus a square?
4 Ss: If it has a right angle.
5 T: If it has …?
6 Ss: A right angle.
7 T: A right angle, that’s enough.
Here the abduction is:

Case – ABCD is a rhombus
Result – ABCD is a square
Rule – If a rhombus has a right angle then it is a square
Case – ABCD has a right angle
There are several things of interest about the logical form of this abductive reasoning.
One is the presence of the first premise. Because of Peirce’s habit of structuring
abductions as syllogisms with two premises, this form would not have been used
by him. It is also interesting to reflect on the epistemic value of the various
premises. The case and the rule have been established, and the result is suggested
by the diagram, as well as by the teacher’s injunction to prove it, as a normal part
of the didactic contract is that statements the teacher asks students to prove should
be true. Otherwise this abductive reasoning is more or less similar in form to what
Peirce described in 1878. In Eco’s terms this is an undercoded abduction as there
104

via free access
TYPES OF REASONING
are many rules the students could have invoked. For example, If the diagonals of a
rhombus are equal, then it is a square, or If two adjacent angles of a rhombus are
congruent then it is a square. Either of these rules would have led to conclusions that
might or might not have been helpful to the students in proving that ABCD is a
square.
Turning to the role of this abductive reasoning, it is not, as Peirce’s later
comments on abductive reasoning would have it, intended to explain why ABCD is
a square. Rather its use is in exploring what features of ABCD need to be established
in order to create a deductive argument for the truth of the theorem that will satisfy
the teacher’s needs. Recall that for Peirce the use of abductive reasoning is
explaining surprising cases. The fact that ABCD is a square is hardly surprising.
That this example fits Peirce’s early descriptions of abductive reasoning but not his
later description might lead us to conclude that this is not really an example of
abductive reasoning at all. However, we would suggest instead that what it tells us
is that Peirce’s different descriptions are useful in identifying issues central to
describing reasoning, but they are not always compatible with each other, nor
complete by themselves.
Abductive reasoning in the Handshakes problem. Our second example of abductive

reasoning occurred when Jason, Nicola and Sofia, three Canadian grade 8 students
(age 13–14), were solving the problem of determining the number of handshakes
that occur when n people shake hands (from Reid, 2003). They were first asked to
explain why 15 handshakes occur when 6 people shake hands, which they did
using a diagram (see Figure 11).
They were then asked to determine the number of handshakes for 26 and 300
people. They solved the case of 26 people by adding 25+24+…+2+1. As Jason
added the numbers using a calculator, Sofia claimed to “know an easier way”. She
tried to explain her way using a diagram (see Figure 12) but Jason protested “How
would you write that down so you wouldn’t have to draw all those lines? Is there a
way?”
Sofia tried to show how to use her method to solve the case of 300 people,
proposing 150 times 300 as the answer. Jason suggested seeing if it worked for 26.
Sofia said it should be 26 times 13, which Jason calculated, getting 338, not 325 as
Figure 11. Handshake diagram for six people.
105

via free access
CHAPTER 6
Figure 12. Sofia’s diagram.
he had obtained by adding. He checked his addition, once more obtaining 325, “So
that can’t be right. But you were close.” Sofia began guessing at other numbers
near 13 and 26. This transcript begins after they have tried 14 and 13.5.
1 Sofia: Well, it was just a guess, OK?
2 Jason: Well you see that would be a problem.
3 Nicola: Yeah but you can’t use that because13 and 13 is 26, and 13.5 and
13.5 is 27.
4 Jason: OK, let’s see…
5 Sofia: Umm, Oooh, Oooh!
6 Jason: Umm, I know, maybe it’s… just a second, just a second..
7 Sofia: Wait, wait, nono, it’s 27, it’s 27!
8 Jason: Maybe it’s the number times half the number, umm, subtract half the
number.
9 Sofia: You lost me.
10 Jason: Because that would work, 325 subtract 13, which is half of 26 is
right.
11 Sofia: Try it again.
Jason has used abductive reasoning to arrive at the general rule:
[The number of handshakes is] the number [of people] times half the number,
subtract half the number (line 8).
from the specific case:
Because that would work, [the number of handshakes for 26 people is] 325
[which is 338] subtract 13, which is half of 26, is right (line 10).
This can be formulated as Peirce’s formulated abduction in 1867, if it is expre-
ssed as:
Any formula for H(n) gives the answer for the case n, and is
based on the value of n
26×(26/2) – (26/2) gives the answer for the case 26, and is based
on the value of 26
n n
n× − is probably a formula for H(n)
2 2
As this formulation would suggest, the premises list a number (two here) of characters
of a formula, and then assert that the specific subject in question has those characters,
106

via free access
TYPES OF REASONING
leading to the conclusion that the subject is probably a formula. In Eco’s terms it is
a creative abduction, as Jason has invented a general rule to account for the
unexpected near success of Sofia’s method.
This abductive reasoning is used (as the later Peirce would suggest) to explore
(in finding a formula) and to explain (why Sofia’s method gave an answer that was
“close”). The next step in Peirce’s sequence of uses is a deductive specialisation of
the conclusion of the abductive reasoning to another specific case, which is in fact
what happened.
After proposing testing his conjecture for 300 people, then 100 people, Jason
settled on 10 as a good number.
21 Jason: Ok let’s try it with 13, or, no, no, 10, let’s try it with 10. 10 times 5
equals 50,
22 Sofia: Try it with 20, or whatever number…
23 Jason: OK, then you would subtract five because that’s half of ten right? All
right.
24 Nicola: Yeah, that’s 45.
25 Jason: So that’s what we got 45.
The next step is an inductive test of this result. Jason made errors in his adding
however, first skipping 10, then adding it twice.
26 Jason: 35?
27 Sofia: [Laughing hysterically]
28 Jason: Dammit! I forgot to push 10! Shut up! 10+9+8+…+1 equals 55? OK,
so if the number was 10…
29 Sofia: Can I try something different?
30 Jason: I’m positive this has to be right though. The number was ten.
This abductive reasoning then can be described as a creative abduction having
the form described by Peirce in 1867, and being used according to the pattern he
set forth in the early 1900s. This does not make it a better example of what Peirce
meant by abductive reasoning than the previous example, however. It only points
out that Peirce’s early and late descriptions can be useful in combination, though
they are not always. In addition we would question whether declaring one example
to fit Peirce’s descriptions and the other not to fit it would be a useful conclusion
for mathematics education research. Our goal, after all, is to describe the reasoning
of students in mathematics classrooms, not to make it fit predetermined categories
from philosophy. Peirce’s work is useful not because it tells us what categories to
fit students’ reasoning into, but because examining Peirce’s descriptions allows us
to identify issues of importance in describing reasoning.
Abductive reasoning in the Football field problem. This example comes from an
activity in a graduate course for practising mathematics teachers taught by the
authors. As a starting activity groups were asked to solve this problem:
Abby and Billy have been hired to mow a football field. Abby can mow a
football field in 2 hours. Billy can mow the same football field in 3 hours. If
they work together how fast would they be able to mow the football field?
107

via free access
CHAPTER 6
The groups quickly produced a number of different methods of solving this problem,
based on algebra, arithmetic and successive approximations. One group presented a
solution that involved a confusing rearrangement of various equations, resulting in
a complicated formula. While no one, including the members of the presenting
group, understood the derivation of this formula it was defended on the basis that it
produced the same answers as other methods. The objection was raised that there
could be any number of formulae that would produce the desired answer but that
was no basis for concluding they were valid solutions to the problem. For example,
the simplest way to get the answer ( 65 hours) from the data is to divide the product
of the two givens by their sum. That this simple method of finding the answer
worked in this case was mildly surprising to the class, but that it worked when the
data was changed to 3 hours and 4 hours was more surprising. These surprising
cases could be accounted for by the abduction that a × b is a general formula for
a +b
solving problems of this type. This is a creative abduction as a new general rule
was created, and fits Peirce’s 1878 description. Here abductive reasoning was used
to explore, but while the general rule explained why the specific case worked, it
gave rise to an even stronger need to explain. Deductive reasoning (in the form of
an algebraic derivation of the formula) was then used to verify and explain the
validity of this formula.
Now that we have pointed out some differences what “abductive” has meant in
Peirce’s work and have seen some examples of different types of abductive
reasoning in students’ mathematical activity, we turn to the proof research literature
that has referred to abductive reasoning. We will consider especially the ways in
which deductive and abductive reasoning have been considered together.
Abductive Reasoning and Deductive Reasoning in Mathematics Education

Research
There is a growing literature on the importance of abductive reasoning in mathematics
education (e.g., Arzarello et al., 1998ab; Cifarelli & Sáenz-Ludlow, 1996; Mason,
1996). Although they do not name it abductive some of the reasoning considered
by Polya (1968) and Smith and Henderson (1959) under the headings “guessing,”
“conjecturing” and “hunches” also seems to be abductive in nature. Much of this
literature looks (in different ways) at the relationship between deductive reasoning
and abductive reasoning. Abductive reasoning is an important component of two
research projects, one focussed on “cognitive unity” and the other on professional
mathematicians’ proofs. We consider these in turn.
Cognitive unity. A group of researchers in Italy, under the leadership of Boero,

has proposed the construct of “cognitive unity”. This construct opposes a vision of
proof as a process independent of and distinct from the process of exploration.
Instead, Boero and his colleagues propose that a continuity exists between the
production of a conjecture and the possible construction of its proof (Garuti, Boero &
Lemut, 1998). They refer to the production of a conjecture as “argumentation” and
108

via free access
TYPES OF REASONING
we will describe their work in more detail in Chapter 8. Here we will note that one
type of reasoning used in the production of a conjecture is abductive reasoning, and
so research on abductive reasoning is an important thread in the work of this group
of Italian researchers.
The interest in the case of abduction lies in the fact that, according experimental
evidence, exploration supporting a conjecture is very often accompanied by
arguments showing this structure, so that passing from conjecturing to proving
would require transformation from an abductive into a deductive structure.
(Mariotti, 2006, p. 186)
Boero, Garuti and Lemut (1999) discuss argumentation leading to the conjecturing
of an implication as “the process of generation of conditionality (PGC)”. They
present a typology of PGCs including one which is abductive:
Generally speaking, a PGC4 consists in a reasoning which can be described
as follows: the regularity found in a particular generated case can put into
action “expansive” research of a “general rule” whose particular starting case
was an example; during research, new cases can be generated (cf. Peirce’s
“abduction” ; see Arzarello et al. 1998[ab]) (pp. 141–142)
These Italian researchers use abduction in Peirce’s third sense, referring chiefly its
role in exploring and explaining and to the simple logical structure:
B ∧ (A → B) ⇒ A
Explicit references to this structure are made by Cabassut (2005), Krummheuer

(2007), and Pedemonte (2003). Other researchers (e.g., Arzarello et al., 1998ab)
refer to Peirce’s 1878 “bean” syllogisms.
Abductive reasoning in professional mathematics. Weber (2002) discusses a use

of proofs in professional mathematics that also involves abductive and deductive
reasoning working together. He considers proofs that justify the use of definitions
and axiomatic structures. Conclusions are already known, but there is a question
about whether a certain axiomatic structure is sufficient to support proofs for all the
desired conclusions. The axioms are produced by abductive reasoning from the
conclusions, and then deductive reasoning is used to produce proofs of the conclu-
sions from the axioms. What is being verified in this case are not the conclusions,
but that the definitions and axioms lead to the desired conclusions.
In this section we have seen that abductive reasoning is not a simple category
but includes several types of reasoning connected by several characteristics. The
work of Peirce and Eco to clarify the nature of abductive reasoning have been
reviewed and applied to students’ reasoning. Finally two ways in which the connec-
tion between abductive reasoning and proof are being explored in mathematics
education research have been mentioned.
As we conclude our discussion of abductive reasoning it might be useful to look
back once more at this chapter’s guiding questions. Is precise terminology necessary,
109

via free access
CHAPTER 6
or does it lead to confusion? Are there better ways to refer to the reasoning used by
the students in the examples we have included here? These questions will be
important also in reading the next section, on reasoning by analogy.
REASONING BY ANALOGY
Reasoning by analogy the last of the four main types of reasoning we will consider
in this chapter. Its importance in mathematical activity has been pointed out by
Polya (1968) and it is receiving increased attention in mathematics education.
English (1997), for example, collects together articles on analogy, metaphor and
imagery and their role in mathematics education. However, the connection between
reasoning by analogy and proof is only beginning to be explored.
Reasoning by analogy involves making a conjecture based on similarities between
two cases, one well known (the source) and another, usually less well understood
(the target).
Analogy is a sort of similarity. It is, we could say, similarity on a more
definite and more conceptual level. ... The essential difference between
analogy and other kinds of similarity lies, it seems to me, in the intentions of
the thinker. Similar objects agree with each other in some aspect. If you intend
to reduce the aspect in which they agree to definite concepts, you regard
those similar objects as analogous. (Polya, 1968, p. 13)
Reasoning by analogy is also mentioned by Smith and Henderson (1959) and Benis-
Sinaceur (2000) asserts that “Virtually every breakthrough relies on analogies,
either within a given field of mathematics or between different fields” (p. 281).
While in the study of language metaphor is often considered a type of analogy,
analogy is usually presented as different from metaphor in the context of mathe-
matics education. Sfard (1997) provides several useful distinguishing characteristics:
The main point to remember is that metaphor has a constitutive power
and thus functions a priori: It brings the target concept into being rather
than just sheds new light on an already existing notion. This is not nece-
ssarily the case with analogy, which is normally understood as a result of a
comparison between two already constructed concepts (of course, the speaker
would usually be better acquainted with one of these concepts and would
use its similarity to the other in order to get a better sense of the less familiar).
In view of the earlier discussion it seems useful to make the following
distinction between the terms analogy and metaphor: Analogy enters the
scene when we become aware of a similarity between two concepts that
have already been created; the act of creation itself is a matter of metaphor.
(pp. 344–345)
I believe that the process of drawing analogies is dialectic in nature, so it does
not leave our understanding of either the target or the source unchanged,
neither is its appearance clearly restricted to [a] certain well-defined stage in
the development of a new concept. All this says that analogy does have some
110

via free access
TYPES OF REASONING
constitutive power as well. Nevertheless, the use of analogy does presuppose

that we have already a rather good grasp of the source and at least some initial
understanding of the target. (p. 345)
It would be useful to restrict the term analogy to the context of reasoning,
namely to thought processes that are applied to investigating existing concepts.
(p. 345)
Bill’s Explanation by Analogy
A striking example of reasoning by analogy is described by Reid (1995b). It

occurred while he was interviewing Bill and John. Reid asked them to explain
“Why is the sum of two odd numbers even?” (Their later deductive response to this
question was discussed above.) Bill’s first answer was: “Because a negative times a
negative is a positive.” His analogy relates even and odd numbers to positive and
negative integers. The rule ‘An odd number plus an odd number is an even number’
corresponds to the rule ‘A negative number times a negative number is a positive
number.’ The sums of even numbers and odd numbers are related in a similar way
to products of integers.
In this analogy we can identify several important similarities (See Table 17).
First, in a context where the truth of a statement in one domain of mathematics has
been questioned, support is obtained by making reference to a statement in another
domain that is held to be more securely known. From the point of view of a more
experienced mathematician, Bill’s choice might seem strange, as the characteristics
of odd numbers seem to be more elementary than the characteristics of negative
numbers. In Bill’s context however, the knowledge that the sum of two odd
numbers is even had only recently been made explicit, and had been called into
question by the interviewer, while the rule that the product of two negative
numbers is positive had been taught to him as a rule in school, giving it a secure
foundation.
Second, there are several similarities between the rule ‘An odd number plus
an odd number is an even number’ and the rule ‘A negative number times a
negative number is a positive number.’ There is a binary operation that combines
two like numbers into a number of one kind, and that combines two unlike
numbers into a number of the other kind. The two kinds divide the integers in
half, into two disjoint sets (excepting that zero is left out of the positive/negative
division). Less mathematically, but perhaps just as significant psychologically,
in both cases there are connotations attached to the words used to describe the
two parts, which makes one half “better” than the other. For example, both ‘even’
(as in ‘even-handed’) and ‘positive’ have good connotations. All of these features
mark links between the two domains of the analogy. Whether Bill would have
been able to explicate these links himself or not, the number of links, and the
degree of match between the features they link, is a measure of the strength of
the analogy.
111

via free access
CHAPTER 6
Table 17. Bill’s analogy
Source Abstraction Target

A negative number times a a*a=(not a) An odd number plus an odd
negative number is a positive number is an even number
number
Multiplication Binary operation Addition
Positive/Negative Half of the integers Even /Odd
Positive Good Even
Negative Bad Odd
Reasoning by analogy has a number of important characteristics that we will describe

further below.
– Reasoning by analogy can be difficult to distinguish from generalising and then
specialising.
– Reasoning by analogy can be used both to explore and to explain in mathematics.
– Reasoning by analogy and reasoning deductively can be formulated in ways that
are very similar, and the conclusion of deductive reasoning can be interpreted as
a result of reasoning by analogy.
Analogy, Generalising and Specialising

To illustrate the relationship that can exist between reasoning by analogy, generalising
and specialising, consider this transcript. In it Ryan and Walt are working on the
Count the Squares problem (See Figure 7). They are recalculating the number of 4
by 4 squares that would fit in a 10 by 10 square. Walt had observed that the
numbers of squares that would fit going across the grid is the same as the number
going down the grid, so he could count the number going across and then square
that number to find the total number of squares. Ryan’s method is to count every
square, with the aid of squares of the appropriate size cut out of paper.
1 Ryan: 1, 2, 3, 4, 5, 6, 7, 7 in each row, then
2 Walt: and I got 7 in each row too.
3 Ryan: 8, 9, 10, 11, 12, OK, I think I’ll go with your way because there’s 7,
and then you could do it from 1, 2, 3, 4, 5, 6, 7,-
4 Walt: do you want to go on to-
5 Ryan: -and then you -
6 Walt: With the 60 by 60, what I did is, was I said “hey” this, this, and I tried
that, the same thing, my way, that way, with the 20 by 20. It worked
perfectly. So I did it with the 60 by 60. Now I’m not sure if my
calculations were correct. I got 7300, 73245.
In line 6 Walt tells what he did next to solve the 60 by 60 problem. It is clear
that he did the “same thing” to solve the 20 by 20, and found that “it worked
perfectly” so he felt confident to do the same thing to solve the 60 by 60. What
reasoning he used to go from the 10 by 10 case to the 20 by 20 case and then to the
112

via free access
TYPES OF REASONING
Figure 13. The relationship between analogy, generalisation

and specialisation (Polya, 1954, p. 16).
60 by 60 case is not clear. There are two possibilities. He might have made an
analogy directly between the 10 by 10 case and the 20 by 20 case, and then on to
the 60 by 60 case. Or he might have generalised that his method works in all cases
and then specialised from his generalisation that it works in the specific cases of
the 20 by 20 and the 60 by 60.
Polya (1954) diagrams the relationship between analogy, generalisation and
specialisation as shown in Figure 13. Although he uses “generalisation” and “speciali-
sation” in different senses than we have above, the relationship he illustrates also
applies to reasoning like Walt’s.
If we reinterpret Polya’s diagram in our terms is says that it is possible to reason
by analogy from the Pythagorean Theorem shown graphically in I to the relation
between the triangle areas shown in II. The three squares on the outside of the right
triangle in I are mapped onto three triangles in II, which are a bit hard to see as
they lie inside the right triangle. One is the right triangle itself, and the other two
are formed by the division of the right triangle by its height. Alternately, the
Pythagorean Theorem can be the inspiration for a generalisation that areas of
similar polygons built of the three sides of a right triangle will be related in the
same way as the squares in the Pythagorean Theorem. This can then be specialised
to the specific case of triangles.
Some hints as to the kind of reasoning Walt was using occur in the subsequent
transcript which occurred after Walt and Ryan joined Mona and Sue to compare
solutions.
10 Walt So I — and then I went down because this row can start going down
like that. This can too so I – counting down (counts 1–8). I can’t start
here, it’s 8 by 8 that’s 64, it’s 8 squared, 8 to the power of 2. Right?
And so for all the 10 by 10s that worked for me. So for the 60 by 60,
all you have to do is 60 squared plus 59 squared plus 58 squared plus
57 squared…dot, dot, dot, dot.
11 Sue That’s extremely confusing.
12 Walt But do you get it?
13 Sue No! I get what you mean by the 3 by 3. I can’t see how you could
possibly have done it for the 60 by 60.
113

via free access
CHAPTER 6
14 Walt Pretend this is the 60 by 60. This is 60 up and 60 across. 60 times 60 is

60 squared. if this is 59 by 59 this is 59 times 59, 59 squared.
15 Sue I rest my case. — We were having a debate. I said, my theory works.
She said, how can you prove it, so I said OK, let’s pretend that, —
where’s my 6? —
Walt begins by repeating his explanation for how to find the number of 3 by 3
squares in a 10 by 10 grid (line 10). He then asserts without further explanation
that “for all the [different sizes of squares in the 10 by 10] that worked for me” and
he makes the transition immediately to the 60 by 60 “all you have to do is 60 squared
plus 59 squared plus 58 squared plus 57 squared…dot, dot, dot, dot.”
For Sue, this is “extremely confusing” especially “how you could possibly get it
for the 60 by 60” (lines 11, 13). Walt attempts to clarify, by using an analogy, in
which the 10 by 10 square he has in front of him stands for the 60 by 60 square that
only exist in his imagination. He is interrupted by Sue, who is excited that he is
using an analogy, as she attempted to use one earlier, between the 6 by 6 and the
60 by 60, in her discussions with Mona. It seems that for Sue it is more interesting
to establish the validity of her method of reasoning by analogy, “let’s pretend that,”
than to understand Walt’s way of solving Count the Squares problems.
16 Sue OK 6. Here’s my 60. I know it’s 6, but pretend it’s 60. Here’s my 5 by 5
but we’re pretending it’s 59. It fits on (counts 1…4) 4 times. That’s
how you know. She said- Excuse me, we have to find an answer.
17 Mona I am going to say what I said. If you are representing the 6 for a 60,
the difference is 10 right?
18 Walt No, the difference is 54.
19 Mona No, I mean not the difference but the multiplication is 10, you are
removing 10s. —So you’ve got 60 and you’ve got 6. If your doing 59
and you’re removing those 10s, you would end up with 5 remainder 9,
not just 5.
20 Walt No, 5 times 10 would be 50 so therefore the 5 by 5 should be 50 —
because the 6 by 6 is 60, 6 times 10 is 60. You have to do the same
thing for a 5 by 5. And 5 times 10 is 50.
In line 16 Sue uses her 5 by 5 square on the 6 by 6 grid to show that there are
four 59 by 59 squares in the 60 by 60 grid. Mona’s objection (lines 17, 19) is to the
mapping between the two cases. If 60 is mapped on to 6, then Mona feels that 59
should map onto “5 remainder 9”, not 5. Walt seems to agree (line 20), though he
argues in the other direction, that if 6 is mapped on to 60, then 5 should map onto
50, not 59.
The interesting thing about this exchange is what it reveals about Walt’s thinking
earlier. If his reasoning to go from the 10 by 10 case to the 20 by 20 case and then
to the 60 by 60 case was based on generalising to an n by n case then one would
expect him to accept Sue’s analogy as its conclusions make sense as specialisations
from the general case. But he does not accept her analogy. This suggests that he
was reasoning by analogy himself (in line 6 above) and that his analogy was based
on the last digits of the numbers he chose. For him, if 10 maps onto 20, then 9 maps
114

via free access
TYPES OF REASONING
onto 19, not because 9 is 10–1 and 19 is 20–1, but because 9 and 19 have the same
last digit. If his focus is on digits that are the same, then it makes sense that he
would interpret Sue’s mapping of 6 onto 60 as a mapping in which the first digit of
the number is important, and so 5 ought to map onto a number with the first digit 5,
and as much like 60 as possible in other ways. 50 is a better candidate than 59 because
50 is more like 60 than 59 is, making a stronger analogy.
Reasoning by Analogy to Explore and Explain

Polya (1968) states, “Analogy seems to have a share in all discoveries, but in some
it has the lion’s share” (p. 17). He gives the example of Euler’s search for an exact
sum for the infinite series: 1 + 14 + 19 + 161 + L
Euler found the value ( π6 ) by applying well known rules for working with
equations of finite degree to equations of infinite degree. As Polya puts it:
Euler’s decisive step was daring. In strict logic, it was an outright fallacy:
he applied a rule for a case for which the rule was not made, a rule about
algebraic equations to an equation which is not algebraic. In strict logic,
Euler’s step was not justified. Yet it was justified by analogy ... Euler
passed from equations of finite degree (algebraic equations) to equations of
infinite degree, applying the same rules made for the finite to the infinite.
(p. 21)
Reid (1995b) gives an example of a much less successful use of analogy in
exploring. An undergraduate student called Wayne was exploring to try to find the
solution to the Arithmagon problem (see Figure 14).
Wayne’s exploring resulted in the diagrams shown in Figure 15. He has observed
a similarity between the problem situation (the target) and one with which he is
more familiar (triangle geometry). So he has begun to explore the unknown situation
of the problem making use of relations he knows are important in geometry: The
angle measures of equilateral triangles, Pythagorean triples, and the area formula
are all applied in the problem situation, but without leading to any useful discoveries.
Walt and Sue’s use of reasoning by analogy (above) is also directed towards
exploring and discovery. They use analogies with grids they have solved like the
10 by 10 grid to find specific answers in the 20 by 20 and 60 by 60 cases and to
find a general method for solving grids of any size.
A secret number has been assigned to each corner of this

triangle. On each side is written the sum of the secret
numbers at its ends. Find the secret numbers.
Generalize the problem and its solution.
Figure 14. The Arithmagon problem (as posed by Reid, 1995b).
115

via free access
CHAPTER 6
Figure 15. Triangles showing geometric properties drawn by Wayne while exploring.
Reasoning by analogy is not used solely to explore, however. Bill offered his
analogy between even and odd numbers and positive and negative integers not as a
means to discover something new in mathematics, but as a way of explaining
something in response to a question “Why?” This shows that reasoning by analogy,
at least in the mathematical activity of students, can have a function other than that
of discovery.
Bill’s analogy, though explanatory for him, would not satisfy everyone. Analogies,
when used to explain, can be problematic. This is apparent in the following episode
of mathematical activity involving Wayne and his group members Rachel, Ben and
Eleanor (from Reid, 1995b).
Rachel had discovered a formula for determining the value at a vertex x. It is:
(b + c − a)
x=
2
Here a, b, and c are the values on the sides of the triangle, with a opposite the
vertex x.
Wayne gave his interpretation of this formula in words, and then raised a question:
1 Wayne: You pick any vertex and it’s going to be the two sides that make the
angle, subtract the side opposite the angle, and divide by two. I under-
stand everything except why you divide by 2.
[A short discussion of the differences between the way Wayne
describes the process and Rachel’s equation omitted ]
2 Ben: You know why you divided by 2, is because-
3 Rachel: Because there’s two sides.
4 Ben: No. No, it’s because-
5 Wayne: There’s two other points, to be solved for, no?
6 Ben: No. No. No. We found out that Y, X + Y + Z is half of the outside
points.
7 Wayne: That’s right!
116

via free access
TYPES OF REASONING
In line 1 Wayne wants an explanation: “Why you divide by 2?” Shortly thereafter,
Ben attempts an explanation (see lines 2-6). He is interrupted by explanations from
both Rachel and Wayne, before he manages to offer an explanation of his own.
In all three cases the explanation attempts to make a link by analogy between the
2 that is divided by in the formula and some other 2, or division by 2, they have
encountered before.
Rachel’s explanation “Because there’s two sides” (line 3) makes an analogy
between the number of sides and the number to be divided by in the formula. It was
probably not clear to the others what sides she was referring to. In the problem
there are in fact three sides, making her analogy very weak indeed. If one tries to
make a sensible analogy out of what she said, perhaps she meant there are two
sides that are connected at the vertex under consideration. In Wayne’s retelling they
are “the two sides that make the angle” (line 1).
Wayne’s explanation (line 5) makes a different analogy, between the two
vertices other than x, and the number 2 to be divided by. It is true that there are two
more vertices to be found once the first is known, but there is no obvious
connection between the division by 2 and the number of vertices remaining to be
solved, other than the number 2. This makes this a weak analogy.
Ben’s analogy (line 6) makes a connection with a relationship they found earlier:
a+b+c = 2(x+y+z). Here the analogy is stronger than in Wayne’s and Rachel’s. In
addition to the occurrence of the number 2 in both situations, there is also a
similarity in that both involve two equations with variables, and in both cases the
2 is a factor for a multiplication or division (see Table 18). The relative strength of
this analogy is likely to have led to Wayne’s acceptance of Ben’s explanation over
his own.
Table 18. Ben’s analogy
X+Y+Z is half of the outside points. (b + c − a)

why you divide by 2 in x =
[(x+y+z)= (½)(a+b+c)] 2
half ½ divide by 2
Algebraic (b + c − a)
(x+y+z)= (½)(a+b+c) x=
formula 2
Reasoning by Analogy and Deductive Reasoning

We will now turn from the use of reasoning by analogy to explore and explain to
three ways in which reasoning by analogy can be linked to deductive reasoning:
– Analogy leading to discovery verified deductively
– Deductive explanation being understood as an analogy
– Analogy going from one deductive proof-text to another
117

via free access
CHAPTER 6
Analogy leading to discovery verified deductively. In Ben’s case and with Euler’s
discovery as described by Polya an analogy can lead to the discovery of a relation-
ship that can later be verified deductively. Ben never verified the relationship he
found which he claimed explained why they divided by two, but it is possible to do
so. Euler spent considerable time and effort after finding the sum of the series by
analogy searching for a deductive verification of it, finally being successful (Polya,
1954, p. 21).
Deductive explanation being understood as an analogy. Rachel’s analogy is related

to deduction in a different way. In her case what her colleagues interpreted as a weak
analogy may have been intended as the conclusion of a deduction. Rachel was the
one who discovered the formula in the first place, through an algebraic derivation:
x+y =b
x+z =c
2x + y + z = b + c
2x + a = b + c
2x = b + c − a
b+c − a
x=
2
When the “2” appears in the third line, it is because x is involved in the totals of
two sides, b and c. When the derivation is completed the “2”, which came from the
combination of two sides at the beginning, becomes the “2” that is divided by at
the end. So when she said “Because there’s two sides” (line 3) Rachel may not
have been making an analogy at all, but instead making reference to her deductive
derivation of the formula. Given this derivation it makes perfect sense to say that
the “2” appears in the final formula because there are two sides, b and c, whose
values are added at the beginning of the derivation.
This example suggests that in some contexts, an analogy is a better explanation
than a deduction. Deduction is a process that must be formulated to be communic-
ated and must be followed with some care to be understood. In this situation the
social dynamic did not afford Rachel the opportunity to make her case clearly.
Ben’s analogy, on the other hand, could be understood immediately by Wayne and
Eleanor, who were familiar with the context to which he was making links.
Analogy going from one deductive proof-text to another. Another connection

between reasoning by analogy and deductive reasoning is the way in which analogy
can lead from one proof-text to another analogous proof-text.
For example, one might make an analogy from triangles to irregular tetrahedra.
Both are the figures in their dimension bounded by the fewest number of sides. They
also have other properties in common (for example, being rigid). This analogy can be
the basis for discovering new theorems about tetrahedra by analogy from theorems
about triangles. It can also be the basis for developing proof-texts that prove theorems.
118

via free access
TYPES OF REASONING
Proof 9: The perpendicular bisectors of a triangle meet in a point
The perpendicular bisector of a side is the locus of points equidistant from the two endpoints
of the side. It is a line.
Two lines perpendicular to two non-parallel lines cannot be parallel, so they must meet in a
point. Therefore the intersection of two perpendicular bisectors of a triangle exists.
The intersection of two perpendicular bisectors of a triangle must be equidistant from all
three vertices. (Call the vertices A, B, C and consider the perpendicular bisectors of
AB and BC. The intersection point being on one perpendicular bisector ensures it is
equidistant from A and B, and being on the other ensures its distance from C is the same as
its distance from B.)
Because the intersection of two perpendicular bisectors of a triangle is equidistant from all
three vertices, it must also lie on the third perpendicular bisector, so all three perpendicular
bisectors meet in one point.
Proof 9 is a proof of a well known theorem about triangles which states:

The perpendicular bisectors of the sides a triangle meet in a single point (called
the circumcentre).
The analogous theorem for irregular tetrahedra is:
The lines perpendicular to the faces of an irregular tetrahedron, passing
through the circumcentres of the faces, meet in a single point.
In Table 19 we use the notation “L(ABC)” to refer to the line perpendicular to the
face ABC passing through the circumcentre of ABC.
The analogy between triangles and tetrahedra can be used to discover the
analogous theorem about tetrahedra, or to explain why it is the case:
The lines perpendicular to the faces of an irregular tetrahedron, passing
through the circumcentres of the faces, meet in a single point, because the
perpendicular bisectors of a triangle meet in a single point.
One can also make this claim:
The lines perpendicular to the faces of an irregular tetrahedron, passing through
the circumcentres of the faces, meet in a point, for the same reasons that the
perpendicular bisectors of a triangle meet in a point.
119

via free access
CHAPTER 6
Table 19. Some links between triangles and tetrahedra

Two dimensions Three dimensions
Triangle Closed shape bounded by the minimal Tetrahedron
number of elements of dimension n-1
Sides (line segments) Bounding elements of dimension n-1 Faces (triangles)
Perpendicular Bisector = Locus equidistant from L(ABC) = locus
locus equidistant from points defining bounding element equidistant from the
endpoints of a side vertices of face ABC
Perpendicular Bisector = Line L(ABC) = line
line perpendicular to a perpendicular to bounding element, perpendicular to a
side, through the through its “centre” face, through the
midpoint circumcentre
This suggests that the proof for the theorem about triangle perpendicular bisectors
somehow applies to the theorem about tetrahedra. Of course it does not directly,
but it can be used as the basis for an analogous proof of the theorem. The reader
may wish to develop this proof in order to explore this use of reasoning by analogy
personally. The resulting proof-text is certainly based on deductive reasoning, but
that reasoning was guided by analogies, making a simple description of the overall
process impossible.
Note that this use of analogy to guide deduction is not simple. The choice of the
analogue is not always as clear as in this case. Here, choosing the perpendicular
line through the circumcentre was the analogue chosen for the perpendicular line
through the midpoint of the side of a triangle. However, consider what happens if
we attempt to make an analogy with another well known theorem about triangles:
The three angle bisectors of a triangle meet in a single point.
Here the situation is more complicated. There are at least three reasonable analogues
for the angle bisector:
– the line equidistant from the three edges meeting at a vertex (the trihedral angle
“bisector”),
– the intersection of the three dihedral angle bisectors bisecting the dihedral angles
between the faces meeting at a point,
– the intersection line of the three planes perpendicular to three faces and passing
through the angle bisectors of each face.
Whichever choice is made, an analogous argument can be produced, but in some
cases gaps occur. For example, if one chooses the trihedral angle bisector, an
argument can be made which is very close to the one made above for the perpen-
dicular bisectors, with “equidistant from a point” replaced by “equidistant from a
line”. The gap occurs when one asserts that the trihedral angle bisectors intersect in
a single point. If they intersect, then they intersect in a single point, but in general
they do not intersect.
120

via free access
TYPES OF REASONING
In the history of mathematics there is a notable case where reasoning by analogy

was used to go from one deductive proof-text to another. Dhombres (1993) describes
the work of Gregory of Saint-Vincent, a seventeenth century mathematician. He
analyses in some detail two of Gregory’s proofs of a single proposition, which have
very similar structures in spite of starting from very different figures. Why have
two parallel proofs for one result?
As we already said, the two parallel proofs are not independent. The first
proof “helps” the second one. But this does not come from a logical argument.
The “deduction” stems from a similar reduction; it uses an analogous path.
Thus for Gregory it was important from a stylistic viewpoint, to exhibit a
strict parallelism between the two proofs. (pp. 412–413)
How does one proof “help” another one? As we noted above, reasoning by analogy
guides the development of the second proof. In fact, in Gregory’s case, this guidance
can be so strong that errors can be made.
In the right column Gregory claims that the same is true, i.e., the area of the
sum of the two rectangles EI and LH is larger than half the area of the
concave hyperbolic segment EGHD. But he does not even attempt to prove
this. And he certainly does not wish to, because he relies on an argument by
symmetry with what has been done according to the proof in the left column.
This seems sufficient. This is an instance of reasoning by analogy. The
trouble is that the result is in general false! (p. 412)
This error is similar to what might have occurred if we had tried to develop an
analogous proof for the “angle bisector” of a tetrahedron, using the trihedral angle
bisector. Relying on the strong analogy between our new argument and the previous
one for the perpendicular bisectors of irregular tetrahedra we might have missed the
gap that occurs because the trihedral angle bisectors do not, in general, intersect.
However we were using reasoning by analogy to prove two different statements.
Gregory uses it to develop parallel arguments that prove the same theorem.
Our final example of using reasoning by analogy to go from one deductive
proof-text to another come from Hanna and Jahnke (2002b). They begin with a
proof from physics (Proof 10). They then produce the analogous argument for the
irregular tetrahedron to show that the four lines connecting the vertices to centres
of gravity of the opposite triangles meet at a single point:
To prove this complex geometrical theorem, we consider the tetrahedron as
loaded at its four vertices with equal masses of weight 1 [Figure 16].
Again, the vertices are connected by rigid and weightless rods. Then, we
consider the centre of gravity of one part-triangle. It will be the point of
intersection of the three medians of this triangle and will be loaded with
weight 3. Therefore, the centre of gravity of the whole tetrahedron must lie
on the line connecting this point and the remaining vertex and, by the law of
the lever, it must divide this line in the ratio 3:1. Since this argument may be
applied to every part-triangle and its opposite vertex, it is clear that all these
lines meet in one and the same point, the tetrahedron’s centre of gravity.
121

via free access
CHAPTER 6
Proof 10: The medians meet in a single point, physics proof
By a rather simple physical argument, we can demonstrate the theorem of geometry that
the medians of a triangle meet at a single point. From the three principles above, we can
immediately find the centre of gravity of the triangle by first considering the triangle as
loaded at its vertices with equal masses of weight 1. (The vertices are considered to be
connected by rigid and weightless rods.)
Then, the mid-point of a side is its centre of gravity, loaded with weight 2. If we connect
this mid-point to the third vertex to form a median, the centre of gravity of the whole
triangle must lie on this median, and, by the law of the lever, must divide it in the ratio 2:1.
Since this construction can be repeated using the other two sides, the three medians must
meet in one and the same point, the centre of gravity.
Hanna & Jahnke, 2002b, pp. 40–41
Figure 16. A tetrahedron with equal masses at its vertices from

Hanna and Jahnke, 2002b.
In this section we have describe reasoning by analogy, and have provided

examples of students’ use of analogies in mathematics. We have pointed out how
analogy is related to generalisation (a type of inductive reasoning) and specialisation
(a type of deductive reasoning) and to needs to explore and explain. Finally, we
have considered how reasoning by analogy is related to proof.
122

via free access
TYPES OF REASONING
A pause to look back at the guiding questions for this chapter may be helpful
this point. Specifically, reasoning by analogy may be important in teaching proof,
as analogy provides a way to bridge the gap between what it known and what is
not. However, it has weaknesses that must be considered carefully.
OTHER KINDS OF REASONING
The four kinds of reasoning described above: deductive reasoning, inductive

reasoning, abductive reasoning and reasoning by analogy, are important because
they have been discussed in the literature related to proof and proving. There are
many other kinds of reasoning that we will not consider here (for example narrative
reasoning, Bruner, 1986) as they are not relevant to our topic. There are some kinds
of reasoning, or replacements for reasoning, that should at least be mentioned:
– Avoiding reasoning by making reference to an authority. This is important
because it occurs in schools.
– Avoiding reasoning by using a tool or technique. This is central to much mathe-
matical activity.
– Transformational reasoning. This has been proposed as another kind of reasoning
and is worth discussing in some detail, as it brings into question whether the
four kinds of reasoning described above are sufficient.
Avoiding Reasoning by Making Reference to an Authority

Smith and Henderson (1959), Reid (1995b), and Harel and Sowder (1998) have
discussed students’ use of outside authorities as a substitute for reasoning.
Reid (1995b) found that making reference to an authority (usually a teacher or
someone acting in the role of a teacher) was the second most popular method of
verification of the students he studied. He notes:
Verifying by reference to authority is not necessarily a poor method of verifi-
cation. Matters of historical fact, for example, cannot be verified without
consulting and analysing various texts, which act as authorities. In mathematics
verifying by making reference to published proofs saves considerable time
and effort, although at some risk, as Wiles found when some of the proofs on
which he had based his first proof of Fermat’s Last Theorem turned out to have
flaws. The elevation of verification by reference to authority over reasoning
as a method of verifying has problems, however. Authorities make errors (or
lie ...) and reasoning in various ways can discover these errors (as occurred in
the case of Wiles’ proof of Fermat’s Last Theorem...). (p. 52)
Harel and Sowder call this use of authority to verify mathematical statements the
“authoritarian proof scheme”. It includes “accepting the statements of others
considered to be more expert” (Sowder & Harel, 2003, p. 253) The students’ “main
source of conviction is a statement appearing in a textbook or uttered by a teacher”
(Harel & Sowder, 1998, p. 247).
Hanna (1983) has pointed out that even within the professional mathematics
community, authority counts for something. One of the five factors she lists as
123

via free access
CHAPTER 6
contributing to the acceptance of theorems (all of which she claims rank higher
than rigourous proof ) is authority: “The author has an unimpeachable reputation as
an expert in the subject matter of the theorem” (p. 70).
Mechanical Reasoning and Formulaic Proofs

Another way to avoid reasoning is to make use of formulae, symbols, or mechanisms
to reach conclusions in reliable and sometimes acceptable ways. Tall (1995) refers
to the use of symbols in ways which ignores their meaning as “Manipulative proof ”:
The previous algebraic identity [(a+b)(a–b) = a2–b2] can also be proved by
manipulation. To show that (a+b)(a–b) = a2–b2 all that is necessary is to
multiply out the brackets on the left hand side and cancel the terms ba and -ab.
(p. 33, see Proof 26 in Chapter 7)
This is the most commonly occurring method of “proof ” in the English
National Curriculum, and occurs widely in numerical investigations. However,
it involves meaningful manipulative facility in algebra rather than logical
deduction. (p. 8)
Reid (1995b) includes using symbols to avoid reasoning within a wider category of
“mechanical deduction”
Mechanical deduction involves the use of a technique or technology that is
based on deductive principles, but that conceals the operation of these
principles in a set of mechanistic rules. Algebraic manipulations are included
in mechanical deduction, as is computer programming. (p. 39)
Harel and Sowder (1996, 1998, 2007) describe two proof schemes in which it is the
form of the presentation that is valued: the ritual proof scheme and the symbolic
proof scheme. In the ritual proof scheme the form should resemble that of a
familiar mathematical proof (e.g., it should have two columns). In the symbolic
proof scheme the use of symbols is central, as in Tall’s manipulative proofs, but
Harel and Sowder require that the symbols and manipulations have “no potential
coherent system of referents (e.g., quantitative, spatial, etc.) in the eyes of the
students” (2007, p. 809). This means that they would not include correct algebraic
manipulations, which Tall and Reid are referring to. Instead they mean a mani-
pulation like
(a + b) (a + b/ ) a
= =
(c + b) (c + b/ ) c
Reid (1995b) calls such a manipulation a “formulaic proof ”.
Transformational Reasoning
Simon (1996) introduced the term “transformational reasoning” into the vocabulary
of mathematics education. According to Simon,
124

via free access
TYPES OF REASONING
transformational reasoning is the mental or physical enactment of an operation

or set of operations on an object or set of objects that allows one to envision
the transformations that these objects undergo and the result of these operations.
Central to transformational reasoning is the ability to consider, not a static
state, but a dynamic process by which a new state or continuum of states are
generated. (p. 201)
Transformational reasoning, according to Simon, also has these three charac-
teristics:
– it is “neither inductive nor deductive” (p.198)
– students have a sense that transformational reasoning can answer questions they
have
– students have “a spontaneous desire to pursue this way of knowing” (p.198)
While Simon’s article is referred to, his category “transformational reasoning”
has not become standard. To see why, we will look at some of the examples Simon
gives of transformational reasoning, and re-examine them in order to see whether
the other kinds of reasoning described above are sufficient to include reasoning
that Simon would call transformational.
One of Simon’s examples is a grade 10 student who, after having constructed an
isosceles triangle by specifying two equal base angles, explained what she did to
Simon and her teacher by saying:
Mary: Well, I know that if two people walked from the ends of this side at
equal angles towards each other, when they meet, they would have
walked the same distance.
Author: What would happen if the person on the left walked at a smaller
angle to the side?
Mary: (without hesitation) Then that person would walk further (than the
person on the right) before they meet (Simon, 1989, p. 373)
What kind of reasoning is Mary using? Note that she is mapping an unfamiliar
situation (Euclidean geometry) onto a more familiar situation (walking to meet
someone). This suggests she is reasoning by analogy. Because both the source and
the target of her analogy are general situations, she can use her reasoning by analogy
to make a generalisation. Her source generalisation (If two people walk from two
points in directions that make equal angles with the line between them, then they
will meet after walking the same distance) is probably not something she learned
explicitly, but rather something she learned through many similar experiences.
Until this moment she may have never voiced this generalisation, but at the same
time in a situation like the one described she would know, without thinking, what
to do. Her knowledge of walking to meet someone was embodied but not conscious.
Other examples involve making connections to real world experiences, but not
through analogy. For example,
A child envisioning a set of five blocks and a set of three blocks and becoming
two rows of four blocks, which she recognises as a total of eight. In a second
instance the child might envision two odd numbers, each as a set of paired
125

via free access
CHAPTER 6
blocks plus a single, coming together resulting in the pairing of the single
blocks. Such an image allows the student to ‘see’ the result of adding (any)
two odd numbers. (1996, p. 202)
Here the child is using the imagined blocks as representatives of a general concept.
We will discuss this use of representations more thoroughly in Chapter 7, but here
it is sufficient to note that the reasoning is deductive. It is the same reasoning that
Bill uses, above, when reasoning deductively about sums of odd numbers.
In summary, transformational reasoning seems to include some cases of
reasoning by analogy to familiar contexts, and some cases of deductive reasoning
using everyday experiences and objects as representations. We are not convinced
that it adds to the descriptive potential of the four kinds of reasoning described above.
SUMMARY
In this chapter we have focussed on four kinds of reasoning: deductive, inductive,

abductive and by analogy. Within these classifications we have made some distin-
ctions, which are summarised in Table 20. Note that deductive reasoning is the
basis of most of the arguments that we will classify in Chapter 7, and inductive
reasoning is the basis for some of them, so further distinctions will be made there.
Table 20. Kinds of reasoning
Kind of reasoning Distinctions

Simple: Modus ponens, with general conclusion
Simple: Modus ponens, specialisation
Deductive Simple: Modus tollens
Chains of deductions
Reasoning by recurrence, mathematical induction
Pattern observing, colligation
Predicting, eduction
Inductive Conjecturing
Generalising
Testing
Peirce 1867: Case having a number of special features
Peirce 1878: Case having a single special feature
Peirce 1903: Reasoning used to explore and explain
Abductive
Eco: Overcoded (only one rule known)
Eco: Undercoded (many rules known)
Eco: Creative (no rule known)
By analogy
126

via free access
TYPES OF REASONING
Table 21. Kinds of reasoning and roles
Kind of reasoning Roles

Deductive verification, explanation
Inductive exploration, verification
Abductive exploration, explanation
By analogy exploration, explanation
In Table 21 we have noted the role that each kind of reasoning can play in mathe-
matical activity, as we have shown in the examples above. Some authors have
claimed that some kinds of reasoning can play roles we have not indicated here.
For example, in Chapter 5 we mentioned exploration and discovery as a possible
role for deductive proofs. This has been suggested by de Villiers (1990, 1999) and
Reid (1995b). As we discussed above, there are logical objections to this, though it
is entirely possible that deductive reasoning could be used in a way that would be
psychologically indistinguishable from exploration.
IDEAS FOR RESEARCH
While recent research in mathematics education has made considerable contributions

to widening our knowledge of the kinds of reasoning involved and related to proving,
more remains to be done. Relationships between kinds of reasoning, students’
understanding and use of them, the goals and obstacles in teaching reasoning and
the connection between reasoning, needs, formulation and formality all require
more research. More broadly, mathematics education could be a fruitful context in
which to explore the question of how people have come to reason in these ways.
The traditional stereotype of inductive reasoning leading to discovery and
deductive reasoning following to verify has been shown to be too limited to
describe mathematical reasoning. But it is not clear what combinations of reasoning
and needs are sufficient. Is it possible for any kind of reasoning to satisfy a need, to
explain for example? This question could be explored theoretically, empirically, or
by applying results from research in psychology. In Chapter 11 we will look at one
empirical approach to investigating how different kinds of reasoning and needs are
connected in students’ mathematical activity, but such research is just beginning.
We have noted that abductive reasoning and reasoning by analogy have a role in
exploration and discovery, but these are both areas in which research is ongoing
(see, for example, the work of Hofstadter, 1995, 2001) and again, mathematics
education may turn out to be a good context for such research.
There exists a body of research in both psychology and mathematics education
on the kinds of reasoning students of different ages can use and choose to use (See
Stylianides & Stylianides, 2008, for a review of some of the key research in
psychology). Considerable progress has been made since Piaget’s theorising in the
early twentieth century (e.g., 1928/2002). However, more needs to be done. One key
127

via free access
CHAPTER 6
point that has received little attention is the process of formulating reasoning; that
is, becoming aware of ones reasoning processes. Formulation is a central concern
in teaching proving as it is necessary for the articulation of reasoning as a proof-
text, and for what Harel and Sowder (1998) call interiorisation, which makes a
method of proving available in other contexts.
A related question is how students might develop the emotional reactions to
kinds of reasoning that mark what Reid (2002a) calls a mathematical emotional
orientation. This includes having a feeling of certainty related to deductive reasoning
and not to other kinds of reasoning, and desiring that certainty. Fischbein (1982)
describes this as developing an intuition towards deductive reasoning.
Finally, Lakoff and Johnson (1980) have argued that metaphor and analogy are
the basis for all human thought. If this is the case how then have other kinds of
reasoning arisen, especially deductive reasoning, which seems very different from
reasoning by analogy? Are other kinds of reasoning based on reasoning by analogy
or somehow independent? Mathematical activity, because it has a special place for
deductive reasoning, may be a context in which this question can be explored.
128

via free access
CHAPTER 7
CLASSIFYING PROOFS AND ARGUMENTS
One of the most important results from research into proof and proving in mathe-
matics education is a better understanding of the diversity of reasoning and argumen-
tation that takes place under the names “proof ” and “proving”. We mentioned some of
the most important efforts to expand the field’s understanding of types of reasoning
that might be related to proof in Chapter 6. We also mentioned the challenge that
faces a scholar reading this literature, that the classifications offered differ,
according to the perspective of the researcher, the focus of the research, and the
particular data being analysed.
For example, consider Balacheff’s categories and those of the preformalists outlined
in Table 22.
Balacheff’s categories and those of the preformalists both overlap and differ.
They seem to agree on the extremes: naïve empiricism and experimental proofs
share many characteristics, as do scientific proofs and démonstrations. Balacheff and
Table 22. Comparison of Balacheff and the preformalists’ classifications of arguments
Balacheff (1988ab, 1991b) Preformalists: Blum and Kirsch (1991);

Wittmann and Müller (1988, 1990);
Branford (1908, 1913)
naïve empiricism in which the truth of a “experimental ‘proofs’ consist of the
result is asserted after verifying several cases; verification of a finite number of examples”
(Wittmann and Müller 1990, pp. 30, our
italics)
crucial experiment in which a proposition is
verified on the basis of testing a special case
that is chosen to be “typical” in the sense The basis for an inhaltlich-anschaulich
that it has no obvious special properties; proof are intuitive premises, which include
“concretely given real objects, geometric-
generic example in which a specific case is intuitive facts, reality-oriented basic ideas,
use to stand for all possible cases in a or intuitively evident,’commonly
general argument; intelligible’, ‘psychologically obvious’
statements” (Blum & Kirsch, 1991, p.187).
thought experiment in which the reasoning An inhaltlich-anschaulich proof should also
is detached from action even if it might be formalisable in principle.
refer to it, and only mental operations are
involved.
Mathematical proof (démonstration) Scientific proofs: those typical of
professional mathematics.
129

via free access
CHAPTER 7
the preformalists differ, however, in what features they use to distinguish proofs
that fall between the extremes. For the preformalists the nature of the premises
(concrete objects, etc.) are of central importance, while for Balacheff it is the role
of action and language that is of most importance. Both of these ways of describing
proofs offer useful insights, and it is an indication of the importance of considering
such proofs to the mathematics education research community that others have also
attempted to describe proofs falling between formal mathematical proofs and
empirical approaches (e.g., Simon, 1996; Tall, 1995). Attempting to reconcile the
categories offered by different researchers can be difficult, as cross references by
the authors are rare, and descriptions are often given according to incompatible
criteria.
Here we will attempt to provide an overall structure into which past classification
systems can be placed. We have found that focussing on the use of representations
in proofs and other arguments provides a sufficiently rich and fine grained basis for
describing them.
PROOFS AND ARGUMENTS DESCRIBED ACCORDING TO THE

REPRESENTATIONS INVOLVED
Four broad categories of arguments can be distinguished by their use, or non-use,

of representations:
– those in which specific examples are used but do not represent a general class,
– those in which specific examples are used as representations,
– those in which words and symbols are used as representations, and
– those in which symbols and words are used without representing anything.
We will call these categories Empirical, Generic, Symbolic and Formal. We will
use both the words “argument” and “proof ” to refer to them, for convenience. Most
classifications of arguments would draw a line somewhere between “proofs” and
“arguments” but that line can be drawn in many places and so we will not attempt
to make that distinction at this point.
Within the four categories Empirical, Generic, Symbolic and Formal several
subcategories can be identified. There are also cases that lie at the borderlines
between these categories. These are summarised in Table 23 and will be described
in the following sections.
Empirical Arguments: Non-Representational Examples

In Empirical arguments, the reasoning operates directly on the topic of the argument.
The examples that are given are non-representational. They stand only for themselves,
not for anything else. Proof 11 is an example of an Empirical argument in which
the numbers used do not represent any numbers other than themselves. A number
of subcategories of this category have been proposed: simple enumeration, extending
a pattern, crucial experiment, kinds or types and the perceptual proof scheme. We
will consider each of these further, below, and refer to authors who include them in
their categorisations.
130

via free access
Table 23. Classification of arguments
Category Subcategories
Simple enumeration
Extending a pattern
Empirical arguments:
Crucial experiment
Non-representational examples
Kinds or types
Perceptual proof scheme
Between non-representational examples and Proof by exhaustion
representational examples Counterexample
Numeric generic example
Generic arguments: Concrete generic example
Examples as representations Pictorial generic example
Situational generic example
Between the Generic and the Symbolic Geometric arguments
Symbolic arguments: Narrative
Words and symbols as representations Symbolic
Between representational and non-
Manipulative
representational symbols
Formal arguments:
Non-representational symbols
Simple enumeration. Proof 11 is not only an example of the broad category

Empirical, but also of the subcategory simple enumeration. In such arguments
a characteristic is observed in a non-exhaustive set of examples and then the
characteristic is generalised to all objects of which the examples are examples.
This category is called “simple enumeration” by Smith & Henderson (1959),
“extrapolation” by Bell (1976), and “naïve empiricism” by Balacheff (1988ab,
1991b).
Proof 11: The Goldbach conjecture
As a mathematical recreation, one junior high school teacher asked his class if they could
find a relationship between the even numbers greater than 4 and the sums of two odd prime
numbers. Here are some of the relations that these pupils discovered:
6=3+3 16 = 3 + 13 = 5 + 11
8=3+5 18 = 5 + 13 = 7 + 11
10 = 3 + 7 = 5 + 5 20 = 3 + 17 = 7 + 13
12 = 5 + 7 .........
14 = 3 + 11 = 7 + 7 30 = 7 + 23 = 11 + 19 = 13 + 17
After working a while on this problem, one youngster exclaimed, “this could go on forever.
There is no end to it.”
Smith & Henderson, 1959, p. 123
131

via free access
CHAPTER 7
Proof 12: Product of negatives

3 × (-4) = -12
2 × (-4) = -8
1 × (-4) = -4
0 × (-4) = 0
therefore
(-1) × (-4) = 4
Extending a pattern. In extending a pattern, the examples are consecutive and a

pattern is observed which then leads to the predictions that the pattern will continue
to apply to additional examples. For example, in Proof 12 the decreasing pattern in
the left and right columns is extended to make a prediction about the next product.
This category is called “extending a pattern” by Smith & Henderson (1959).
Crucial experiment. Balacheff (1988ab, 1991b) identifies a type of argument he

calls “crucial experiment” in which a characteristic of a single, more or less typical,
example is attributed to all.
This type of validation is distinguishable from naive empiricism in that the
pupil poses explicitly the problem of generality and resolves it by staking all
on the outcome of a particular case that she recognises to be not too special.
(Balacheff, 1988a, p. 219).
Proof 13 is a student’s argument that Balacheff offers as an example of this
category.
Kinds or types. Chazan (1993) describes “kinds or types” which is a subcategory

of Empirical argument in which the set in question is divided into subsets
according to some attributes, and then one example from each of the subsets is
examined (p. 369). If examples from all the subsets are found to share a
characteristic, then that characteristic is claimed to be common to the whole set.
For example, Proof 7 in Chapter 6 shows specific triangles of four types: acute,
right angle, obtuse and isosceles, for which a property has been verified.
Proof 13: Try it with 15
[Nadine and Elisabeth are trying to find a formula that will give the number of diagonals of
a polygon Pn.They have conjectured the recursive formula: f (n+1) = f (n) + a(n+1), where
a(n+1) = a(n) +1 and with initial values f (4)=2 and a(5)=3, but Nadine is not convinced.]
She relies on the crucial experiment to decide: ‘try it with 15 and then if it works for that,
well then that means that it works for the others’. In fact, this experiment was carried out on
P10, because P15 seems too complex from the outset.... The experiment they have in mind
confirms this result.
Balacheff, 1988a, p. 224
132

via free access
Proof 14: The diagonals of a rhombus are congruent
Melissa was attempting to prove that the midpoints of any isosceles trapezoid form a rhombus.
She began by trying to prove that FH is congruent to EG, which Harel and Sowder say she
believed because of her “perceptual observation of the figure she drew” (1998, p. 256).
The perceptual proof scheme. All of the prototypes we have offered so far have
involved numerical examples, but pictorial or concrete examples might also serve
as the basis for argument without being used as representations. Such arguments fit
into Harel and Sowder’s (1998) “perceptual proof scheme” in which a single pictorial
example is the justification for a proposition. They give Proof 14 as an example.
Proofs between the Empirical and the Generic

Some arguments lie between our categories of Empirical arguments which use non-
representational examples and Generic arguments which use representational
examples. These “in-between” arguments use non-representational examples but
unlike Empirical arguments they are accepted as verification in the community of
mathematicians. For mathematicians, non-representational examples can be used in
proofs in two situations: when all the examples are checked and found to satisfy
the proposition, and when one example is checked and found to refute the propo-
sition. These two situations are known as “proof by exhaustion” and “refutation by
counterexample”.
Proof by exhaustion. In a proof by exhaustion every element of a finite set is

checked to see if the proposition in question holds. Such proofs are also called
proofs by cases (e.g., Maher & Martino, 1996a, p. 195). Proof 15 is an example
from the work of Bell (1976, p. 40). The italics indicate the student’s writing.
Counterexample. When a statement is made that some property applies to all

elements in a set, then any example of an element that does not have that property
is considered sufficient to refute the general statement. Such proofs are included in
the subcategory counterexample. Proof 16 is an example. Smith and Henderson
(1959) describe some proofs under the label “counterexample” however their
description: “to disprove a proposition, either prove its contradictory or prove that
its acceptance leads to a contradiction of a postulate or theorem” (p. 167) seems
instead to describe reductio ad absurdum (see below). The examples they give fit
our description of counterexamples better than they fit Smith and Henderson’s. In
them a general statement is made about a set of elements, and then one element is
identified for which the statement is false, disproving the generality of the statement.
133

via free access
CHAPTER 7
Proof 15: There are 14 possibilities and all fit

Write down any number up to fifteen. Write down the next number. And add it to the first.
Write down your answer. You have now written down three numbers.
Gail says that one, and only one, of these numbers is in this list:
3, 6, 9, 12, 15, 18, 21, 24, 27, 30
[some of the solution omitted]
Below is the list of possibilities, the ringed ones are the ones mentioned in the list. There are
14 possibilities, and all fit Gail’s rule.
Proof 16: Not all prime numbers are odd

Although there are an infinite number of examples that confirm the proposition
that all prime numbers are odd, the existence of one counterexample, 2, is
enough to refute it.
Generic Proofs: Examples as Representations

The Empirical arguments given above involved examples, but the examples did not
represent anything, instead they were operated on as themselves. In this section we
describe arguments based on examples that represent a larger class (typified by
Proof 17). In it a specific number is used to represent a class of numbers. “10” in
this case, does not represent itself, but any number. The addition of the first 10
numbers is intended to act as a model for the addition of any sequence of the first n
numbers. This category of proofs seems to have been first described by Mason and
Pimm (1984) and became well known after being used by Balacheff (1988ab, 1991b).
Such arguments falls into subcategories according to what kind of examples are
involved: numeric, concrete, pictorial or situational.
Proof 17: Numeric Gauss proof

Prove: The sum of the first n positive integers is n(n+1)/2
Consider the sum 1+2+3+4+5+6+7+8+9+10. Write this sum, and the reverse,
and add them:
1+ 2+ 3+ 4+ 5+ 6+ 7+ 8+ 9+10
10+ 9+ 8+ 7+ 6+ 5+ 4+ 3+ 2+ 1
11+11+11+11+11+11+11+11+11+11 = 10 × 11
Because the sum was added to itself, dividing 10 × 11 by 2 gives the sum.
134

via free access
Proof 18: Divisibility by Nine
To prove “If a whole number is divisible by 9, then the sum of its digits is divisible by 9,”
take a number, say 867. This number can be represented as 8 × 100 + 6 × 10 + 7, which is
(8 × 99 + 6 × 9) + (8 + 6 + 7). Since the first addend, 8 × 99 + 6 × 9, definitely is divisible
by 9, the second addend, 8 + 6 + 7, which is the sum of the number’s digits, must be
divisible by 9.
Adapted from Harel & Sowder, 1998, p. 271
Numeric generic examples. Proof 17 typifies this subcategory as well as the

larger category. Harel and Sowder (1998) offer another example, Proof 18. It is an
example of their category “generic proof scheme”, in which “conjectures are inter-
preted in general terms but their proof is expressed in a particular context” (p. 217).
Note that prior to the development of algebraic notation, mathematicians were
unable to express non-geometric justifications in general terms, and made use of
specific numbers in context where now we would use variables. For example,
Euclid’s proof of the infinitude of primes (Book IX, Proposition 20, see Chapter 1,
Proof 4) requires showing that “if there are exactly n prime numbers then there are
n + 1 prime numbers” in order to arrive at a contradiction, but Euclid could only
show this for a specific case: “if there are three prime numbers then there are four
prime numbers” and assert that this applies in general.
Concrete generic examples: Action proofs. Proofs using concrete objects as

representational examples have been called “action proofs” and “enactive”, reflecting
two intellectual traditions, one arising in central Europe and the other in English
speaking countries. The name “action proof ” was popularised by Semadeni (1984).
His description is as follows:
An action proof of a statement S should consist of the following steps:
1° Choose a special case of S. The case should be generic (that is without
special features), not too complicated, and not too simple (trivial examples
may later be particularly hard to generalise). Choose an enactive and/or
iconic representation of this case of a paradigmatic example (in the sense
of Freudenthal [1980]).
Perform certain concrete, physical actions (manipulating objects, drawing
pictures, moving the body etc.) so as to verify the statement in the given
case.
2° Choose other examples, keeping the general schema permanent but
varying the constraints involved. In each case verify the statement, trying
to use the same method as in 1º
3° When you no longer need physical actions, continue performing them
mentally until you are convinced that you know how to do the same for many
other examples.
4° Try to determine the class of cases for which this method works.
(p. 32)
135

via free access
CHAPTER 7
Proof 19: Action proof of the commutativity of multiplication
The statement to be proved is that the product does not depend on the order
of the factors... .
We choose a pair of numbers, e.g., 3 and 5. We are to show that 3 × 5 = 5 × 3.
As the concretisation of n × m we choose n rows with m counters in each.
Thus we begin the action by arranging the counters as in Figure [a].
We separate them horizontally (Figure [b]) and infer that the number of
counters is 3 × 5. Then we separate them vertically (Figure [c]) and get 5 × 3.
The number of counters in Figure [a] must be independent of the way of
counting. Hence 3 × 5 = 5 × 3.
Figure [a] Figure [b] Figure [c]
Semadeni, 1984, p. 33
His first example (see Proof 19) presents a slightly different picture. In the
example, step 1º is followed, and then he notes “it may be desirable to perform
analogous actions for some other pairs of numbers. ... and to perform the action
in the mind” (p. 33). In other words, performing steps 2º and 3º seems to be
optional.
Pictorial generic examples: Visual proofs. Although Proof 19 appears here as

words and pictures, for it to be an action proof the actions described in words must
be carried out on the objects depicted in the pictures. If the actions are only imagined
and the pictures are used directly it becomes an example of a visual proof.
Note that we are drawing a different border between action proofs and visual
proofs than Semadeni, who claims:
Some visual proofs (like the celebrated “Behold!” in a proof of the
Pythagorean theorem) are, in fact, action proofs. To understand such a proof
requires concrete actions (e.g., cutting the given square and moving the
pieces) performed in the mind. (p. 34)
It seems to us that all visual proofs require “actions performed in the mind” and so
we would prefer to apply strictly Semadeni’s criterion 1º (“concrete, physical
actions”) rather than taking the more flexible approach he advocates later in his
article.
136

via free access
Proof 20: Behold!
Tall, 1995, p. 31
Visual proofs can show discrete objects (as in Proof 19). They can also portray
continuously varying quantities. Proof 20, suggested by Tall (1995) is an example.
Tall comments, “any actual drawing will have specific values for a and b, but such
a diagram can be seen as a prototype, typical of any right-angled triangle. This
gives a kind of proof which is often termed ‘generic’; it involves ‘seeing the
general in the specific’.” (p. 32)
Situational generic examples: Reality oriented proofs. Blum & Kirsch (1991)
describe what they call “reality oriented proofs” in which the proof is based on a
“real” situation. “In a reality-oriented proof, basic ideas meaningful in reality
and easily accessible for the learner are used, such as the derivative as a local
rate of change” (p. 188). In Blum (1998) he gives Proof 21 as an example. It
proves that:
a c a a +b c
< ⇒ < < (a,b,c, d ∈ Q + )
b d b c+d d
Note that the proof uses specific numbers and a pictorial representation of them,
so it resembles arguments based on numeric and pictorial generic examples.
However, central to the argument is an “everyday experience” of judging the
“wininess” of a Schorle. This raises the question of whether the argument is as
effective for a reader who has not had this experience. Is the proof still a “reality
oriented proof” if it is based on a situation one has never experienced but can
imagine?
As we noted when discussing Proofs 19 and 21 it is not always easy to
distinguish between proofs using concrete generic examples (action proofs), using
pictorial generic examples (visual proofs) and those using situational examples
(reality oriented proofs). For example, Schifter (2009) includes them all within
what she calls “representation-based proofs” (p.76) which she describes in the
context of elementary school mathematics.
137

via free access
CHAPTER 7
Proof 21: Schorle proof
Proof: Let’s take 43 and 7

9
as an example.
3
We ... interpret as a mixture of 3 parts wine and 4 parts mineral water, and the same
4
with 79 . Such a mixture of wine and water is in South Germany where Johannes and I come
from, called a Schorle. We assume all parts to be of the same size.
What shall we use in the following? We know from everyday experience with (idealised)
Schorle: The mixture ratio defines the ‘wininess’ of a Schorle, which can be tasted (or seen,
especially if it is red wine). Two Schorles have the same wininess (that is the same taste or
colour) if and only if the fractions are equivalent; that means proportional enlargement of
the two components of a Schorle doesn’t change the wininess. Schorle 1 is less ‘winy’ [than]
Schorle 2 if and only if fraction 1 is less than fraction 2.
3
3 6 3 7
If we want to compare the two given fractional numbers we find = 4 and hence <
4 9 4 9
since 6 43 < 7. So the 43 Schorle is less winy than the 79 Schorle.
Now we pour the two Schorles together ([see figure]; of course, real
Schorles look a bit different!).
We get a Schorle with 3 + 7 = 10 parts wine and 4 + 9 = 13 parts
water. ...
Now it should be clear from everyday experience with mixture ratios
(with respect to its wininess) that the mixed Schorle truly lies
between the two initial Schorles; that is it tastes a bit more winy than
the one and a bit less winy than the other. ...
How can we explain this experience? We could argue in a more detailed way as follows.
It is clear that if we pour together two Schorles with the same wininess then the mixed
Schorle has also the same wininess, for instance 43 and 6.759
.
If we take the 79 -Schorle instead of the 6.75
9
-Schorle and mix it with the 43 -Schorle then we
obviously add a bit more wine, so the new mixed Schorle tastes a bit more winy than the
first one. That’s it!
adapted from Blum, 1998, pp. 70–71
Proofs between the Generic and the Symbolic

The examples we have given above of Generic proofs fall into what Branford
(1908) calls “intuitional” proof and we would agree with him that:
The intuitional ... is merely a preliminary and less rigid species of ideal
scientific proof: there is really no sharp dividing line between these kinds of
proofs, but only a difference in degree of logical rigour. (p. 102)
Hence it is not surprising that identifying a sharp dividing line between arguments
that use examples as representations and arguments that use words and symbols as
representations is not easy. In fact, many arguments that could be considered
prototypical of all proofs fall into this gap, as most geometry proofs make use of
words and symbols, but also rely to some extent on a generic figure.
138

via free access
Proof 22: Two-column proof for the triangle angle sum
Healy & Hoyles, 1999, p. 163
For example, consider Proof 22. The “statement” “p=s” is based not only on the
rule given as a reason, but also on the position of p and s in the figure. The figure
plays a very special role in the proof.
As Herbst (2004) notes:
Some of those features of a diagram, which are available to perception along
with the premises of an argument (and whatever is known), can actually feed
the argument being produced. Perception of a diagram thus supports and at
times even replaces the logical machinery of discourse as a source of state-
ments about the object (p. 132)
And Tall (1995) describes the way in which the words in such a proof point out the
generality of the generic figure:
The ideas of Euclidean geometry are inspired by visual representations but
they are formulated verbally to give the proofs greater generality. A theorem
in Euclidean geometry specifies a certain geometric configuration. A figure
drawn to accompany the theorem is a generic picture which represents any
configuration satisfying the statement. The verbal proof then applies not just
to the specific picture drawn, but generically to the whole class of figures
represented by the theorem. (p. 8)
Symbolic Proofs: Words and Symbols as Representations

Proof 23 is an example of a symbolic argument, in which the words, and letters
represent things other than themselves. For example, “S(n)” represents the sum of
the first n positive integers, “n” represents any positive integer, and the word “row”
represents one of the two equations given above it.
139

via free access
CHAPTER 7
The proofs most teachers and students associate with “correct” mathematical
proofs employ a mix of words and symbols as representations (as in Proof 23).
Some proofs (e.g., Proof 24) use only or mainly words, and have been called
“narrative” (Healy & Hoyles, 2000) for this reason. Others use very few words and
many symbols specific to mathematics (e.g., Proof 25). These have been called
“symbolic” (Healy & Hoyles, 2000). These are not two distinct subcategories,
however, as proofs might be sited anywhere on a continuum from highly narrative
to almost completely symbolic.
While the meaning of the words and symbols in a mathematical proof might not
be immediately obvious, for these proofs, each word and symbol represents
something. For example, in Proof 25 the symbol Bij represents the element in row i
and column j of an arbitrary diagonal matrix B.
Proof 23: Symbolic Gauss proof

Prove: The sum of the first n positive integers is n(n+1)/2
Let S(n) = 1 + 2+ 3 + ... + n.
Then S(n) = n + (n–1) + (n–2) + ... + 1.
Taking the sum of these two rows,
2S(n) = (1+n) + [2+(n–1)] + [3+(n–2)] + ... + (n+1)
= (1+n) + (n+1) + (n+1) + ... + (n+1)
= n(n+1)
Therefore, S(n) = n(n+1)/2
Knuth, 2002a, p. 384
Proof 24: Infinitude of primes

For any finite set {p1,...,pr} of primes, consider the number n = p1p2...pr + 1.
This n has a prime divisor p. But p is not one of the pi: otherwise p would be
a divisor of n and of the product p1p2...pr, and thus also of the difference
n–p1p2...pr =1, which is impossible. So a finite set {p1,...,pr} cannot be the
collection of all prime numbers.
Aigner & Ziegler, 2004, p. 3
Proof 25: The product of two diagonal matrices is diagonal
When i ≠ j , ( AB) ij = ∑A
k
ik Bkj = Ai1B1 j + Ai 2 B2 j + L + Ain Bnj
= Aii Bij , as A is diagonal, so Aik = 0 when i ≠ k

= 0 as B is diagonal, so Bij = 0 as i ≠ j
Segal, 2000, p. 199
140

via free access
Proofs between the Symbolic and the Formal
In a mathematical proof some of the symbols can be representations and others can
be used without representing anything in the course of a calculation. This often
occurs in proofs in which an algebraic expression is manipulated to show that it is
equivalent to another expression. The symbols in the initial and final expressions
are representations, but the intermediate expression may include terms that are not
representational.
Tall’s (1995) category of “Manipulative proof ” includes such proofs. He gives
Proof 26 as an example, and notes that: “This is the most commonly occurring
method of ‘proof’ in the English National Curriculum, and occurs widely in
numerical investigations. However, it involves meaningful manipulative facility in
algebra rather than logical deduction” (p. 34). Longer proofs, involving both
logical deduction and “meaningful manipulative facility in algebra” would also be
characterised by a mix of meaningful symbolic representations and meaningless
symbolic manipulation.
Proof 26: Proof of an algebraic identity
The algebraic identity (a + b)(a – b) = a2 – b2 can be proved by algebraic

manipulation:
(a + b)(a – b) = a2 – ab + ba – b2
= a2 – b2
adapted from Tall, 1995, p. 33
As noted in Chapter 6, such proofs are in principle mechanisable, and may

involve no reasoning at all on the part of a student who produces one. In such
a case mechanical reasoning, rather than deductive reasoning, underlies the
proof.
Formal Arguments: Non-Representational Symbols

Proof 27 involves symbols that are not representational. It is a formal argument
in which the symbols X and O do not represent anything, and the Rule is
concerned purely with the syntax of the strings, not their meaning. Some
meaningful words are included, but they are not considered to be a part of the
proof, merely guides for the reader. Such proofs are the subject of formal proof
theory and unlike the proofs in mainstream mathematics every assumption and
logical rule used is specified. This one comes from a mathematician’s exercise
designed to introduce his students to some of the main concepts used in doing
formal mathematical proofs. We mention such proofs here for completeness, but
their relevance for understanding students’ proving and the teaching of proof in
schools is limited.
141

via free access
CHAPTER 7
Proof 27: A formal proof
Axiom 1: OX
Axiom 2: XO
Rule: For any strings w and v, wv and vw ⇒ w
Theorem : X
Proof:
1. OX (by Axiom 1)
2. XO (by Axiom 2)
3. X (by Rule applied to statements #2 and #1)
QED
adapted from Monks, 2002
Overview
Table 24 gives an overview of the categories and subcategories described above as
well as references to the examples given in this chapter and elsewhere.
OTHER CLASSIFICATIONS OF PROOFS AND ARGUMENTS
Many authors have offered classifications of proofs and arguments. However, as

we noted above, their efforts have often been for their own specific purposes and it
is rare that connections to other classifications have been made. In this section we
will attempt to summarise the classification systems proposed by the most well
known authors, and to place our examples within their systems to illustrate connec-
tions and differences.
We have already mentioned the classification schemes of Balacheff and the
preformalists. In Table 25 we show how their categories relate to ours. We have
used descriptions of categories in Blum and Kirsch (1991), Wittmann and Müller
(1988, 1990), Branford (1908) and Balacheff (1987, 1988ab, 1991b) to establish
related categories. We also include the categories of Bell (1976) and van Dormolen
(1977) who make similar distinctions.
A number of things are evident from Table 25. First, the distinction we make
between Geometric, Symbolic and Formal proofs is not a focus for any of these
researchers. Their interest is primarily in classifying arguments that fall outside
the main stream of mathematical proofs. But it is clear that they have different
concerns within that broader interest. Bell and Balacheff make finer distinc-
tions among empirical arguments than do van Dormolen and the preformalists.
However, the preformalists make distinctions among generic arguments that the
other do not.
142

via free access
Table 24. Overview of categories
Category Subcategories Examples
Proof 11: The Goldbach

Simple enumeration
conjecture
Extending a pattern Proof 12: Product of negatives

Empirical arguments: Non-
representational examples Crucial experiment Proof 13: Try it with 15
Kinds or types Proof 7 in Chapter 6

Perceptual proof Proof 14: The diagonals of a
scheme rhombus are congruent
Between non- Proof 15: There are 14

Proof by exhaustion
representational examples possibilities and all fit
and representational
Proof 16: Not all prime numbers
examples Counterexample
are odd
Proof 17: Numeric Gauss proof
Numeric
Generic arguments: Proof 18: Divisibility by Nine
Examples as representations
Proof 19: Action proof of the
Concrete
commutativity of multiplication
Pictorial Proof 20: Behold!

Situational Proof 21: Schorle proof
Between the Generic and Proof 22: Two-column proof for
Geometric arguments
the Symbolic the triangle angle sum
Narrative Proof 24: Infinitude of primes
Symbolic arguments: Words
and symbols as Intermediate Proof 23: Symbolic Gauss proof
representations Proof 25: The product of two
Symbolic
diagonal matrices is diagonal
Between representational
Proof 26: Proof of an algebraic
and non-representational Manipulative
identity
symbols
Formal arguments: Non-
Proof 27: A formal proof
representational symbols
Proofs by exhaustion and counterexamples are ignored by everyone but Bell,

which is understandable given the focus of these researchers on arguments other
than mathematical proofs. Why then does Bell include them?
Bell developed his system to classify the responses of students to problems
which required an explanation and justification of a conclusion. His categories are
143

via free access
CHAPTER 7
Table 25. Comparison of proof classification systems
Our Our sub- van

Bell Preformalist Balacheff
Category categories Dormolen
Simple
Extrapolation
enumeration Naïve
Extending a Systemic empiricism
pattern empirical
Empirical Crucial Experimental First level

experiment Partially
Crucial
systemic
experiment
Kinds or empirical
types
Naïve
Perceptual Extrapolation
empiricism
Proof by Finite
Between exhaustion Empirical Unclassified
Formal Mathematical
Empirical (Third
scientific proof
and Generic Counter- Counter- level?)
example example
Preformal
Numeric (unspecified Generic
subtype) Second level
example
Generic Concrete Action proof
Unclassified Geometric-
Pictorial intuitive /
iconic
Reality Thought
Situational
oriented experiment
Between
Generic and Geometric
Symbolic
Narrative
Symbolic Intermediate Complete
Formal Mathematical
deductive Third level
Symbolic scientific proof
explanation
Between
Symbolic and Manipulative
Formal
Formal
144

via free access
divided first into “Empirical” and “Deductive” categories, which are then subdivided
into subcategories indicating the degree of validity of the arguments. In setting up
the subcategories of the “Empirical” category, Bell included what he called “Finite
Empirical”, checking all cases (in other words producing a proof by exhaustion), as a
possibility because one of the problems he included was amenable to this approach (see
Proof 15: There are 14 possibilities and all fit, for a student’s response to this item).
Bell did not include proofs by counterexample in his categories prior to examining
the students’ responses because he did not expect them to occur in response to the
problems he set. However, he added them afterwards to classify responses where
counterexamples were used. As with proof by exhaustion Bell includes
“counterexample” in his “Empirical” category, not in his “Deductive” category.
We did not include Bell’s two broad categories “Empirical” and “Deductive” in
Table 25 to save space. Similarly, we have left out Balacheff’s broader categories.
Naive empiricism and crucial experiment belong to the larger category of pragmatic
arguments; the statement is justified by direct actions on examples. Generic
example, thought experiment and mathematical proof are intellectual proofs where
the reasoning is based on language, and objects serve only as place-holders (in
generic examples) or illustrations.
The distinction between pragmatic arguments and intellectual proofs is based on
the reasoning underlying the argument. This distinction parallels two others
Balacheff (1987, 1988b) makes, based on the nature of the concepts employed
(ranging from practical “know-how” to theoretical “scientific” knowledge) and on
the expression of the argument (ranging from “ostension”, where objects stand for
themselves, to naïve formalism, Balacheff’s term for the style of proofs written by
professional mathematicians). Combining Balacheff’s three types of distinctions
between arguments one could arrive at a more detailed classification than is usually
attributed to him, but as far as we know only the distinction between pragmatic and
intellectual proofs has been widely considered by others, and it is not clear how
independent Balacheff’s other types are.
The Concept of “Anschauung”

Most of the preformalists are German speaking and they generally prefer to speak
of “anschaulich” or “inhaltlich-anschaulich” proofs instead of preformal proofs.
Blum and Kirsch (1991) claim that “inhaltlich-anschaulich” “cannot be translated
into English in an adequate manner” (p. 201). This raises the question of what is
meant by it.
In the epistemology of Immanuel Kant the concept of “Anschauung” is central.
Kant understands “Anschauung” as knowing on the basis of an insight related to an
object. For Kant the object is of fundamental importance. In his view the object
evokes sensory impressions that embed the insight in the concrete. Thus insight is
obtained via the senses, not by pure thought. This relation to the senses is, for Kant,
a key aspect of Anschauung, in contrast to pure thinking. This allows us to build
meaningful concepts founded on their creation in relationship to real objects
(Knipping, 2005).
145

via free access
CHAPTER 7
In the context of contemporary mathematics education this idea of “Anschauung”

can be expressed as the attachment of understanding to a mathematical representa-
tion. Historically, Anschauung, in this sense, played a central role in geometry for a
long time. Even today, this type of knowing is central to geometry teaching. In
geometry teaching drawings and material objects are part of the method of learning
geometry. Concepts and relationships are bound to visual and thus material
representations.
The concept of “inhaltlich-anschaulich” proofs may be an interesting example of
a word preceding a concept. Recall that the preformalists cite Branford (1908) as a
source of ideas. Branford describes a type of evidence or proof which he calls
“intuitional”. This word is not always translated in the same way in the German
translation (Branford, 1913). In one instance “intuitional evidence or proof ” (1908,
p. 233) is translated as “Evidenz oder Beweis durch Intuition” (p. 239). Later on
the same pages “Intuitional evidence” is translated as “Evidenz durch Anschauung”.
Whether this translation led the later preformalists to incorporate ideas related to
Kant’s notion of Anschauung into Branford’s concept of intuitive evidence cannot
be known.
Harel and Sowder’s Proof Schemes

Harel and Sowder’s (1996, 1998, 2007) classification of arguments used by students
to verify mathematical statements is well known and often referenced. They refer
to “proof schemes” by which they mean what a person considers adequate to
remove or create doubts about the truth of an observation. Because they define
proof schemes as convincing arguments and we have chosen in this chapter to
focus on proof-texts and the reasoning associated with them, we are working with
different meanings of ‘‘proof ’’ and so it is not surprising that our categories are
significantly different from those of Harel and Sowder.
Harel and Sowder’s broadest distinction is between three classes of proof schemes:
External proof schemes, Empirical proof schemes and Analytical proof schemes.
Harel (2007) later renamed these classes the external conviction proof scheme class,
the empirical proof scheme class and the deductive proof scheme class, but we will
use the original terminology as it is more well known. See Table 26 for a summary
of all their categories and subcategories.
In external proof schemes a statement is justified by making reference to an
outside authority or the form of the argument. We have considered these proof
schemes in Chapter 6 in the section on ways of avoiding reasoning. We have not
included them in our categories in this chapter for several reasons: representations
play no consistent role in them, some of them are not associated with any “text”
and in general they are not based on reasoning. Harel and Sowder divide external
proof schemes into the ritual, authoritarian and symbolic. Authoritarian proof
schemes involve avoiding reasoning by making reference to an authority, for
example a textbook or a teacher. Ritual proof schemes depend entirely on “the
appearance of the argument (for example, proofs in geometry must have a two-
column format)” (Harel & Sowder, 2007, p. 809). Symbolic proof schemes depend
146

via free access
“on symbol manipulations, with the symbols or the manipulations having no potential
coherent system of referents (e.g., quantitative, spatial, etc.) in the eyes of the
student” (p. 809).
In empirical proof schemes conjectures are verified “by appeals to physical facts
or sensory experiences” (1998, p. 252). There are two subcategories “inductive proof
schemes” and “perceptual proof schemes.” Inductive proof schemes are distinguished
by the use of quantitative evaluation (“e.g., direct measurement of quantities, numeri-
cal computations, substitution of specific numbers in algebraic expressions, etc.”
p. 252, footnote), while “perceptual proof schemes” make use of mental images
based on direct perceptions without any analysis or appreciation of possible trans-
formations (p. 255).
Analytical proof schemes are characterised by the use of deductive reasoning.
Within this category Harel and Sowder describe a number of subcategories. First
they distinguish between “transformational proof schemes” and “axiomatic proof
schemes”.
Transformational proof schemes are “characterized by (a) consideration of the
generality aspects of the conjecture, (b) application of mental operations that are goal
oriented and anticipatory, and (c) transformations of images as part of the deduction
process” (p. 261). Proof 28 is an example provided by Harel and Sowder to illustrate
the transformational proof scheme.
In discussing perceptual proof schemes Harel and Sowder refer to “transforma-
tional reasoning” and make a specific reference to Simon (1996). However, they do
not use the phrase “transformational reasoning” in describing transformational
proof schemes, leading us to suppose that they are using “transformational” in a
different sense than Simon. In fact, as we discussed in Chapter 6, Simon includes
examples of reasoning by analogy in his description of transformational reasoning,
which do not fit under the heading “transformational proof scheme” because they
are not deductive.
Harel and Sowder further distinguish two levels of the transformational proof
scheme, which they label “internalised” and “interiorised”. If the proof scheme is
internalised it has become a standard heuristic that is used in proving similar
conjectures (Harel & Sowder, 1998, p. 262). However, there is no awareness of it
as a standard heuristic. When the prover reflects on his or her proving and becomes
aware of it, it is said to be interiorised (p. 264). Harel (2001) explores mathematical
induction (reasoning by recurrence) in connection with interiorisation. These two
levels are not types of arguments or proof-texts, so we will not consider them
further here. They correspond to Reid’s notion of more or less formulated reasoning
described in the Introduction to Part 2.
In addition to these two levels of the transformational proof scheme Harel and
Sowder also distinguish further subcategories. Some of these are grouped under the
heading “restrictive proof schemes” suggesting that the other examples they give
could be called “unrestrictive”, though they do not do so. Restrictive proof schemes
are limited by being tied to a context, or by a lack of language suitable for expressing
generalities, or by a rejection of some logical principles. Such restrictive proof
schemes are called “contextual”, “generic” or “constructive” accordingly.
147

via free access
CHAPTER 7
Proof 28: A transformational proof
Amy ... demonstrated to the whole class how she imagines the theorem, “The
sum of the measures of the interior angles in a triangle is 180°.” Amy said
something to the effect that she imagines the two sides AB and AC of a
triangle ∆ABC being rotated in opposite directions through the vertices B and
C, respectively, until their angles with the segment BC are 90° (Figure [b]).
This action transforms the triangle ABC into the figure A'BCA'', where A'B
and A'C are perpendicular to the segment BC. To recreate the original
triangle, the segments A'B and A'C are tilted toward each other until the
points A' and A' merge back into the point (Figure [c]). Amy indicated that in
doing so she “lost two pieces” from the 90° angles B and C (i.e., angles A'BA
and A'CA) but at the same time “gained these pieces back” in creating the angle
A. This can be better seen if we draw AO perpendicular to BC: angles A'BA and
A'CA are congruent to angles BAO and AOC, respectively (Figure [d]).
Harel & Sowder, 1998, p. 259
Contextual proof schemes are tied to a specific context. For example, a student
might prove a statement about vector spaces in general in the context of Rn, or a
statement about a finite geometry in the context of Euclidean plane geometry. This
subcategory corresponds roughly to our subcategory of arguments based on situational
generic examples. Generic proof schemes correspond to our category of Generic
arguments. They make use of specific examples to stand for the class of which they
are examples. The argument is made with specifics, but has the same form as a
completely general argument, if only the specifics were replaced with appropriate
general terms. Constructive proof schemes require that the existence of an object
be proven by the construction of the object. Students with this proof scheme reject
proof by contradiction and indirect proof. This does not correspond to any of our
categories, though our example illustrating a proof by exhaustion would fall into it.
Transformational proof schemes operate on the basis of what Freudenthal (1971)
calls “local” organisation (see Chapter 12). Propositions are related deductively to
each other, but not within any larger framework. In contrast, axiomatic proof
schemes involve an understanding that the ultimate basis for a proof is a set of
148

via free access
axioms and definitions. Among axiomatic proof schemes Harel and Sowder make
three distinctions. If the axioms used must conform to the prover’s intuitions they
describe the proof scheme as “intuitive-axiomatic”. For example, someone might
be prepared to prove on the basis of the axioms of Euclidean geometry because
they conform to that person’s intuitions about space, but not be prepared to work
with non-Euclidean axiom systems. In a “structural proof scheme” it is accepted
that there are a number of different interpretations or models of the axiom system
and a recognition that the axiom system concerns the underlying structure of all the
models, not any particular one. Finally in an “axiomatizing proof scheme” “a person
is able to investigate the implications of a varying set of axioms, or to axiomatize a
certain field” (Harel & Sowder, 1998, p. 274).
The categories used by Harel and Sowder cannot be mapped onto our categories
in any simple way. In the cases of Empirical and Transformational proof schemes
there are some clear equivalents, which is not surprising as Harel and Sowder’s
work is one influence on our categorisations. Empirical Inductive proof schemes map
onto our Empirical category (excluding our perceptual subcategory) and Empirical
Perceptual proof schemes map onto our Empirical: Perceptual subcategory. Trans-
formational proof schemes that are not restrictive map fairly well onto our Symbolic
category and among the Restrictive Transformational proof schemes, Contextual
proof schemes map onto our Generic: Situational subcategory while Generic proof
schemes map onto the remainder of our Generic category (see Table 26).
Table 26. Harel and Sowder’s proof schemes
Category Subcategory 1 Subcategory 2 Subcategory 3 Our category or example

Ritual
External Authoritarian See Chapter 6
Symbolic
Inductive Empirical
Empirical
Perceptual Empirical: Perceptual
[Unrestrictive] Symbolic
Generic Generic
Transform-
ational Restrictive Contextual Generic: Situational
Proof 15 There are 14
Constructive
Analytical possibilities and all fit
Intuitive- Proof 22: Two-column proof
axiomatic for the triangle angle sum
Axiomatic Proof 16: Not all prime
Structural
numbers are odd
Axiomatizing Proof 27: A formal proof
149

via free access
CHAPTER 7
Harel and Sowder’s other proof schemes do not map neatly onto our categories,
but there is value in having two well elaborated systems to work with. In Table
26 we indicate our corresponding categories and in cases where no correspondence
exists we indicate which of our example proofs fit into each proof scheme,
however, this is meant only to illustrate that the same proof-text can be classified
in both systems, not to imply that the categories containing the examples are
related.
Proof Techniques
Smith and Henderson (1959) offer an extensive, but not often cited, classification
of arguments and proofs. They make distinctions primarily based on the type of
reasoning used in the argument, dividing “probable or inductive inference” from
“necessary inference (deduction)”. Within these broad categories, however, they
make distinctions according to a variety of criteria. We have referred in Chapter 6
to categories Smith and Henderson describe which refer to the reasoning employed
(“the Method of Analogy”, “Hunches”, “Testing hunches”) or to avoiding reasoning
(Recognition of Authority). Above we have incorporated some of their categories
into our classification of empirical arguments (“the Method of simple enumeration”
“Extending a pattern of thought”). Here we will refer to Smith and Henderson’s
descriptions under the heading of “necessary inference (deduction)” of various
proof techniques. These represent another way to categorise proofs and allow us to
introduce some terminology that is used in the literature on teaching and learning
proof.
Some of the proof techniques Smith and Henderson describe we have already
considered either above or in Chapter 6. These include counterexample (above),
“detaching the antecedent” or modus ponens, “developing a chain of propositions”
or chain of deductions, and mathematical induction or reasoning by recurrence.
Smith and Henderson also describe “proving a conditional,” “reductio ad absurdum,”
“indirect proof,” and “proving a statement of equivalence” which we will describe
briefly here.
Proving a conditional. As we discussed in Chapter 6, detaching the antecedent or

modus ponens, depends on having a general rule “if p then q” to use. If, however,
one wishes to prove such a rule, one begins with the assumption “p” and attempts
to derive the consequent “q” from it. This process is called “proving a conditional”
(p.167). An alternative strategy is to assume the negation of the consequent, and to
derive from that assumption the negation of the antecedent. In logic there is a law
(the law of contraposition) that states that a general rule “if p then q” is true
precisely when its contrapositive “if not q then not p” is true (p.168). For example,
proving that no things that are not white are swans is equivalent to proving that all
swans are white.
Reductio ad absurdum. The proof technique reductio ad absurdum (also known

as proof by contradiction) is used to prove a proposition “p”. It begins by assuming
150

via free access
the negation of “p” and then deriving from that negation a false statement. In other
words one proves the conditional “If not p then not True”. The law of contra-
position then is used to replace this condition with it contrapositive “If True then
p”. As “True” is known to be a true statement, one can then detach the antecedent
“True” leaving the consequent “p”. Proof 24: Infinitude of Primes (above) is the
classic example of a proof by contradiction. In order to prove that there is an
infinite number of prime numbers, one assumes that there is only a finite number
of them. If this is so, then one can (in principle) list them all. But from this list
one can generate a prime number which is not on the list, resulting in a
contradiction.
Indirect proof. Indirect proof involves showing that among several possibilities at
least one must be true and then showing that all but one are false. The conclusion is
that the remaining case must be true. Reductio ad absurdum is a special case of
indirect proof where there are only two cases: p and not-p. A proof that proves that
A=B by showing that A<B and A>B lead to contradictions is an example of an
indirect proof with three cases (A<B, A=B, A>B).
Proving a statement of equivalence. Smith and Henderson also describe “Proving

a statement of equivalence” (p. 169) as a proof technique. This applies when one
wishes to prove that two statements are true under exactly the same conditions. The
strategy is to first show that when the first is true, the second must be, and then to
prove that when the second is true, the first must be.
IDEAS FOR RESEARCH
Systems for classifying proofs and arguments have used a number of different
bases: degree of formality (e.g., the preformalists), reasoning employed (e.g., Bell),
the role of representations (e.g., ours), and the degree of abstraction of the objects
referred to (e.g., Balacheff). It is not clear how these different bases are interrelated
and to what degree they are independent of each other. This question could be
explored theoretically, or empirically as Harel and Sowder (1998) did.
Such systems have been developed to aid in describing research results, but also
to help clarify the goals of teaching. For example, Harel and Sowder propose that
the aim of teaching is to help students move from external and empirical proof
schemes to analytical (deductive) proof schemes and Stylianides and Stylianides
(2009) suggest that teaching approaches that provoke cognitive conflict with respect
to empirical arguments support a transition to deductive arguments. How facility
with some types of arguments can support or become an obstacle to learning to
produce and value other kinds of arguments requires more study. It may be that
students already possess attitudes towards different types of arguments that could
provide a basis for learning. Research by Galotti, Komatsu and Voelz (1997)
suggests that even young children recognise the difference between arguments
based on inductive reasoning and those based on deductive reasoning and show more
151

via free access
CHAPTER 7
confidence in deductive arguments. On the other hand, it may be that facility with
some types of argument could be a cognitive obstacle to learning others. Fischbein
(1982) suspects that an entirely new “intuition” or “basis of belief ” (p. 17) is required
for students to come to treat deductive arguments as special in the way that mathe-
maticians do.
152

via free access
CHAPTER 8
ARGUMENTATION
Argumentation is often linked to proof in research in mathematics education. For

example, the European Society for Research in Mathematics Education (ERME)
has held working groups under the title “Argumentation and Proof ” at three of
its recent conferences. Duval (1999) includes a summary of some of the reasons
for an increasing attention to argumentation in mathematics education. These
include:
– the recognition in the disciplines of philosophy and linguistics that natural
languages rather than formal languages are the basis for human thought and
communication.
– the recognition in mathematics education of the importance of social processes
in learning. Duval points to the early work of Balacheff (e.g., 1982) as an example
of this.
Duval also points out that “Ce qu’on appelle ‘argumentation’ n’est pas facile à
définir.” [What we call ‘argumentation’ is not easy to define.] (1990, p. 195). Just
as there is diversity in usages of the words “proof ” and “proving”, it is not always
clear what is meant by “argumentation” in mathematics education.
One consequence of this diversity of usages is that several incompatible conclu-
sions have been reached on the relationship between argumentation and proof. These
can be summarised as:
– Argumentation is completely distinct from proof and the cause of students’
misunderstanding of proof
– Argumentation is related to proof in complex ways but is an obstacle to learning
proof
– Argumentation is distinct from proof but compatible with learning proof
– Argumentation is essential to teaching and learning proof
In the following we will examine some of the differences in approaches to
argumentation that give rise to these contradictory conclusions.
Lo que dice
Balacheff
Balacheff (1999) notes the diversity of approaches to argumentation in mathematics
education and traces them to different theoretical starting points.
The diversity which we may observe among the problématiques of argumenta-
tion and of its relations with mathematics, notably with mathematical proof,
is in my view basically due to profound differences in the theoretical research
in this domain. (Different theoretical conceptions of argumentation, para. 1)
He asserts that:
Three authors, by the contrast of their problématiques and their distance, can
be used to provide a system of benchmarks by reference to which one can
153

via free access
CHAPTER 8
situate works on argumentation: Chaïm Perelman, Stephen Toulmin and

Oswald Ducrot. (Different theoretical conceptions of argumentation, para. 1)
Briefly, Perelman sees argumentation as being about convincing. Toulmin sees it as
being about the structure of the argument and its reference to premises accepted in
a community. Ducrot places argumentation at the heart of the activity of discourse
and focusses on grammatical structures. This gives a possible classification:
– argumentation is what convinces another person
– argumentation has a logical structure accepted in a community
– argumentation is present in all discourse and founded on grammatical elements
While these theoretical starting points do not tell the whole story of argumen-
tation in mathematics education, it is clear that, as Balacheff puts it:
Reference to one or another of these conceptions of argumentation is likely to
make us adopt a different position with regard to what the argument can
represent in the practice of mathematics, notably with the intent of teaching
and in relationship with mathematical proof. (Different theoretical conceptions
of argumentation, para. 7)
In addition to looking at which theorists researchers base their work on, one can
also examine the definitions and descriptions they give to see what they mean by
“argumentation.” As we will see in this chapter, there are several possibilities,
including:
– A kind of reasoning
– A social behaviour
– A process that produces a logical discourse
– A process that gives rise to conjectures
In terms of the role of argumentation, most researchers follow Perelman to the
extent that they see argument being about convincing. However, others see argumen-
tation as being more about explaining the reasoning process used to come to a
conclusion. More specifically, in mathematics education argumentation has been
seen as playing several special roles, including:
– A pre-condition for learning
– An epistemological obstacle to the learning of mathematical proof
– Integral to proving
– Opposed to proving
Argumentation is also linked to “argument” in various ways. “Argument” can be
seen as:
– Giving rise to argumentation
– Being the result of argumentation
– Being a part of argumentation
– Being identical to argumentation
In this chapter we will discuss argumentation in the work of the researchers
most active in the area. We will try to clarify and relate their usages of “argumenta-
tion” and “argument” and also to discuss the key concepts in their work. We will
also note an important cultural difference in the nature of argumentation in Japan
compared to the approaches to argumentation taken in the West.
154

via free access
ARGUMENTATION
– What does “argumentation” mean to you, and how is it related to proof?
– Which researchers’ notions of argumentation make sense to you?
– What are the important implications of each researchers’ work, independent of
their take on argumentation?
ARGUMENTATION VERSUS PROOF
Raymond Duval proposes a simple claim about the relationship between argumen-
tation and proof:
Deductive thinking does not work like argumentation. However these two
kinds of reasoning use very similar linguistic forms and propositional connec-
tives. This is one of the main reasons why most of the students do not under-
stand the requirements of mathematical proofs. (1991, p. 233)
Already a number of things are clear about what Duval means by “argumentation”.
For Duval, argumentation is a kind of reasoning, and it is opposed to proving.
Reference to the meaning of statements as opposed to their form is, for Duval, an
important distinction between argumentation and deductive reasoning.
Le raisonnement argumentatif ... a des règles implicites qui relèvent en partie de
la structure de la langue, et en partie des représentations des interlocuteurs: le
contenu sémantique de propositions y est donc primordial. [Argumentative
reasoning has implicit rules that are partly from the structure of the language,
and partly from the statements of the speakers: the semantic content of propo-
sitions is essential.] (1991, p. 235)
In contrast, in deductive reasoning, what is important is not content but the “statut
opératoire” [operational status] of each step, which is its role in the functioning of
the inference.
Duval cites Perelman as one of his sources, and so it is not surprising that
argumentation is linked to justification or convincing.
La notion d’argumentation est étroitement associée à celle de justification d’une
affirmation ou d’une thèse [The notion of argumentation is closely associated
with that of justification of a claim or a thesis] (Duval, 1992–1993, p. 38)
Argumentation arises “spontaneously as soon as there is an argument with someone.”
(1999, Part I, para. 1). What, then, is an “argument” according to Duval?
An argument is considered to be anything which is advanced or used to
justify or refute a proposition. This can be the statement of a fact, the result of
an experiment, or even simply an example, a definition, the recall of a rule, a
mutually held belief or else the presentation of a contradiction. They take the
value of a justification when someone uses them to say “why” he accepts or
rejects a proposition. An argument is the answer to the question why “do you
say that?... do you believe that? ...” (Section II.1, para. 2, ellipses in original)
155

via free access
CHAPTER 8
Duval’s notion of an “argument” refers to three features. They are the choice to
argue, the motivation for that choice, and the goal intended:
The notion of argument ... involves taking into account two dimensions. To
speak of argument is first to refer to the choice of a subject to achieve a
determined goal. Next it is to refer to the context for the production of the
argument. A production context is determined according to two points. On the
one hand there is whatever motivated the recourse to arguments: a weight on
the sense of a decision to be taken, the resolution of a conflict of interest, the
resolution of a problem presenting technical or logical constraints. On the
other hand there is the objective: convince someone else or, on the other
hand, diminish the risk of error or uncertainty in the choice of a process.
(Section II.1, para. 4)
The motivation for arguing and the goal intended (which Duval considers together
as the “production context”) can differ, and an important difference in production
contexts allows Duval to distinguish between “heuristic argumentations” and
“rhetorical argumentations”. “Heuristic argumentations” are found in mathematics.
In mathematics the motive and the objective of the argumentation are specific
to the problem to be solved. ... it is the constraints of the problem which
determine the choice of arguments and not first the beliefs of the person to
whom the argument is directed. The force of an argument depends primarily
on how appropriate it is to the situation and not on its resonance in the
universe of the person being addressed (Section II.1, para. 5)
Outside of mathematics the beliefs of the person being addressed are important,
and “rhetorical argumentations” occur:
When it is a question of convincing someone about a decision to be made, or
resolving a conflict of interest or getting consensus on a question, there is an
inversion of priority: one takes into account first the convictions of the person
being addressed. (Section II.1, para. 5)
Here an important shift has occurred. “Argumentation” for Duval is a kind of rea-
soning, but “an argumentation” in the context of distinguishing between rhetorical
and heuristic argumentations, seems to be simply an argument. That is, an heuristic
argumentation is an argument that gives rise to deductive reasoning, while a rhetorical
argumentation is an argument that gives rise to argumentation (a kind of non-
deductive reasoning used to convince).
We can now return to Duval’s claim that there is a rupture or gap between
argumentation and proof and that this is a cause of students’ difficulties with
proof.
Il est important, en effet, de ne pas négliger l’écart existant entre des argumen-
tations rhétoriques et les argumentations heuristiques ... Le développement de
l’argumentation même dans ses formes les plus élaborées n’ouvre pas une
voie vers la démonstration [It is indeed important not to overlook the gap
156

via free access
ARGUMENTATION
between rhetorical argumentations and heuristic argumentations ... Teaching

argumentation even in its most developed form is not a path to mathematical
proof] (Duval, 1992–1993, p. 60)
What he means by this is that the kinds of non-deductive, meaning based, arguments
used in everyday life to convince others are not useful in helping students to
understand the deductive formal structure of mathematical proofs. This is not far
from the position of other researchers. For example, Fischbein (1982) asserts that
“the concept of formal proof is completely outside the main stream of behavior”
(p. 17) and that it requires a new “basis of belief ” which is distinct from the basis
of other kinds of reasoning.
Balacheff, at times, seems to take a position like Duval’s. For example he writes:
Argumentation and mathematical proof are not of the same nature: The aim
of argumentation is to obtain the agreement of the partner in the interaction,
but not in the first place to establish the truth of some statement. As a social
behaviour it is an open process, in other words it allows the use of any kind
of means; whereas, for mathematical proofs, we have to fit the requirement
for the use of knowledge taken from a body of knowledge on which people
(mathematicians) agree. (1991a, pp. 188–189)
Balacheff’s position then was based on the work of Perelman, who wrote:
Whereas mathematical proof in its most perfect form is a series of structures
and of forms whose progression cannot be challenged, argumentation has a
non-constraining character. It leaves to the author hesitation, doubt, freedom
of choice; even when it proposes rational solutions, none is guaranteed to
carry the day. (1970, p. 41, cited and translated in Balacheff, 1999, Argumen-
tation, a problématique resulting from the study of social interactions, para. 3)
Later, however, Balacheff distanced himself from this position. After considering
Duval’s “rupture” between argumentation and proving, and the concept of cognitive
unity (see below) between argumentation and proving, Balacheff (1999) arrives at
a third way:
From the perspective of teaching and learning, I arrive at supporting neither
the thesis of continuity nor that of a rupture between argumentation and mathe-
matical proof (or proof in mathematics), but at proposing the recognition of
the existence of a relationship which is complex and is part of the meaning of
both: argumentation constitutes an epistemological obstacle to the learning
of mathematical proof, and more generally of proof in mathematics. (Argu-
mentation, epistemological obstacle to the teaching of mathematical proof,
para. 1)
Balacheff distinguishes here between three concepts: argumentation, mathematical
proof (démonstration in French) and proof in mathematics (with proof/preuve
having the everyday meaning of verification). Thus, it is clear that verification in
mathematics is still distinct in Balacheff’s mind from convincing (argumentation).
157

via free access
CHAPTER 8
Certainly, he would disagree with a position like that of Hersh that “In mathematical
practice, in the real life of living mathematicians, proof is convincing argument, as
judged by qualified judges.” (1993, p. 389).
I would maintain thus that there is no mathematical argumentation in the
frequently suggested sense of an argumentative practice in mathematics which
is characterized by the fact that it escapes certain of the constraints present for
mathematical proof. (Balacheff, 1999, The risks of recognizing a “mathematical
argumentation”, para. 4)
However Balacheff acknowledges that argumentation (i.e., non-deductive reasoning)
is used in mathematics:
If there is no such thing as mathematical argumentation, there does nonetheless
exist argumentation in mathematics. The resolution of problems, in which
I would like to say that there are no holds barred, is the context in which
to develop the argumentative practices using means which could be used
elsewhere (metaphor, analogy, abduction, induction, etc.) but which disappear
in the construction of a discourse acceptable with regard to the rules specific
to mathematics. (1999, The risks of recognizing a “mathematical argumenta-
tion”, para. 5)
ARGUMENTATION IN ACCORD WITH PROOF
Nadia Douek has written a number of papers on argumentation and proof in the
past decade. She argues against Duval’s position (outlined above) that the differ-
ences between argumentation and mathematical proof lie at the heart of students’
difficulties in learning mathematical proof. Among the theorists Balacheff names
she makes only occasional references to Toulmin (e.g., in 2005, p. 152). She does,
however, give very clear statements of what she means by “argumentation” and
“argument” and is consistent in her definitions from (1998) to (2007). For Douek,
“argumentation” means two related things:
First, it denotes the individual or collective process that produces a logically
connected, but not necessarily deductive, discourse about a given subject
(2002, p. 304)
Second, it points at the text produced through that process (p. 304)
Douek adopts the Webster’s dictionary definition of “argument” which is “a reason
or reasons offered for or against a proposition, opinion or measure” (cited, e.g., in
2005, p. 152). Arguments “may include verbal arguments, numerical data, drawings,
etc.” (1999b, p. 274). Argumentation and argument are linked in that “an ‘argumenta-
tion’ consists of one or more logically connected ‘arguments’” (p. 274), “connected
by deduction, induction or analogy” (1999a, p. 91). In Chapters 7 and 9 we use
“argument” in this way.
Douek’s and Duval’s definitions of “argument” are quite similar (See Table 27)
however there is a sharp distinction in the way they see argumentation. For Douek,
argumentation is “not necessarily deductive” while for Duval, it is necessarily not
158

via free access
ARGUMENTATION
deductive: “Deductive thinking does not work like argumentation”. Douek also
considers what Duval calls “démonstration” to be “formal proof ” as opposed to
mathematical proof which is “what in the past and today is recognized as such by
people working in the mathematical field” (1998, p. 128). Given these differing
definitions, it is not surprising that Douek disagrees with Duval’s claim that there is
a “rupture” between argumentation and mathematical proof. Douek’s position, based
on different definitions, is that “mathematical proof can be considered as a particular
case of argumentation” (p. 129).
Boero (e.g., 1999, Boero, Douek & Garuti, 2003) has adopted Douek’s definitions
of “argumentation” and “argument”. He discusses argumentation in six phases of
mathematical activity:
– production of a conjecture
– formulation of the statement
– exploration of the content (and limits of validity) of the conjecture
– selection and enchaining of coherent, theoretical arguments into a deductive
chain
– organization of the enchained arguments into a proof that is acceptable
according to current mathematical standards.
– approaching a formal proof.
In the first two phases, argumentation concerns inner (and eventually public)
analysis of the problem situation, questioning the validity and meaningfulness
of the discovered regularity, refining hypotheses, discussing possible formu-
lation(s). In the third phase, argumentation plays three important roles:
producing (or resuming from the first phase – “Cognitive Unity of Theorems”,
[Boero, Garuti], Lemut, & Mariotti, 1996; Garuti et al., 1998) arguments for
validation, discussing their acceptability according to requirements about their
nature (for instance, although empirical arguments may be relevant in the first
phase and even in the approach to validation, they must be progressively
excluded from this phase on), and finding possible links leading from one to
another. I could add that the nature of the whole third phase is argumentative,
and the fourth phase is also largely argumentative (especially as concerns the
control of argument enchaining). In the fifth phase, argumentation may play
a role when comparing the text under production with current standards of
“rigour”, textual organisation, etc. resuming from the first phase – (Boero,
1999, para. 20)
Here Boero seems to say that argumentation is a part of all phases of mathematical
activity, including the creation of a formal proof.
Boero refers to the important concept of “cognitive unity” connecting argumenta-
tion in the first phase (conjecturing) to argumentation in the third phase (exploration
of the conjecture). It is worth spending some time elaborating this concept here.
Duval (1999) raises the following important question, about which he says “we
do not yet have available enough really usable observation results”:
With reference to the work of a mathematician, a lot of emphasis is put on the
moment of developing a conjecture. But, at least for the students, do the
159

via free access
CHAPTER 8
arguments which lead to the formulation of a conjecture also make it possible

to find the means to prove it? (Section III.4, para. 3)
A group of researchers including Boero, Douek, Mariotti, Pedemonte, Bartolini
Bussi and their colleagues would say that there is a significant body of empirical
evidence suggesting that the formulation of a conjecture does make it possible to
find the means to prove it. They propose that a continuity exists between the
production of a conjecture and the possible construction of its proof, which they
refer to as “cognitive unity” (Garuti et al., 1998) (see Chapter 6).
They arrived at this idea as a hypothesis based on episodes in the history of
mathematics (such as those described by Lakatos, 1976) where conjecturing and
proving were interwoven. Recall Lakatos’s comment:
This intrinsic unity between the ‘logic of discovery’ and the ‘logic of justi-
fication’ is the most important aspect of the method (p. 37 italics in original)
Boero and his colleagues observed students proving and found experimental evidence
of cognitive unity. They describe cognitive unity in these terms:
During the production of the conjecture, the student progressively works out
his/her statement through an intensive argumentative activity functionally
intermingled with the justification of the plausibility of his/her choices.
During the subsequent statement-proving stage, the student links up with this
process in a coherent way, organising some of the previously produced
arguments according to a logical chain. (Boero, Garuti, Lemut, & Mariotti,
1996, p. 113)
Not every researcher with a focus on cognitive unity sees argumentation in the way
Boero does, however. Notably, Pedemonte restricts argumentation to the process of
conjecture generating.
My research has shown the importance of considering the process of problem-
solving by comparing the process connected with the conjecture, denoted by
argumentation in this article, and the subsequent proof as a product (2007, p. 25)
This usage of “argumentation” leads to a conclusion that is diametrically opposed to
Duval’s claim that teaching argumentation is not a path to proof. In fact, Pedemonte
claims that “argumentation activity might favour the construction of a proof.”
Experimental research (Boero, [Garuti & Mariotti], 1996; [Boero, Garuti,
Lemut, et al., 1996;] Garuti et al., 1998; Mariotti, [2000]) shows that proof is
more ‘accessible’ to students if an argumentation activity is developed for the
construction of a conjecture. The teaching of proof, which is mainly based on
‘reproductive’ learning (proofs are merely presented to students, they do not
have to construct them) appears to be unsuccessful. A didactical consequence
of this study is that suitable open problems (Arsac, [Germain & Mante],
1991) which call for a conjecture could be used to introduce the learning of
proof. Indeed, according to these studies, when cognitive unity can be realised,
argumentation activity might favour the construction of a proof. Moreover,
160

via free access
ARGUMENTATION
the idea of cognitive unity can be used to foresee and analyse some difficulties
that students might have in the construction of proof, especially when
cognitive unity cannot be realised to perform proving. (2007, p. 25)
ARGUMENTATION ACCORDING TO KRUMMHEUER
Götz Krummheuer has written extensively on argumentation using the work of

Toulmin as a starting point, though he also echoes Perelman’s description of
argumentation as being about convincing. For Krummheuer argumentation is a
process, and “argument” is the product of it:
One common and obviously also more frequently successful way of giving
structure is by using language; that is by choosing verbal means in order to
express the sense made in this moment and the reasons for this specific sense
making. These language-bound methods or techniques of expressing the
reflexive claim of acting rationally will be called argumentation. Its aim is to
convince oneself as well as the other participants to this special kind of
“rational enterprise.” The final sequence of statements accepted by all partici-
pants, which are more or less completely reconstructable by the participants
or by an observer as well, will be called an argument. (1995, pp. 246–247)
While Krummheuer is not explicitly interested in mathematical proof, his descrip-
tions of argumentation make it clear that he is interested in a process leading to the
verification of a statement, which presumably could include proving.
Usually, these techniques or methods of establishing the claim of a statement are
called an argumentation. Thus a successful argumentation refurbishes such a
challenged claim into a consensual or acceptable one for all participants (p. 232)
In addition, argumentation has to do with the formulation of the reasoning used:
Empirically the concept of argumentation will be bound to interactions in the
observed classroom that have to do with the intentional explication of the
reasoning of a solution during its development or after it. (p. 231)
However, unlike proving which is often seen as a personal, individual activity, argu-
mentation is a social process:
Argumentation is seen here primarily as a social phenomenon, when coope-
rating individuals tried to adjust their intentions and interpretation by verbally
presenting the rationale of their actions. (p. 229)
Because of the emergent nature of social interaction, argumentations are
usually accomplished by several participants. Such a case is called a collective
argumentation (p. 232)
Krummheuer’s approach to argumentation is distinct from those described
above in that he does not see argumentation as interfering with learning proof,
or supporting learning proof, but rather as essential for learning mathematics:
Argumentation is not only a teaching aim in the sense that one has to
design mathematics instruction in a way that ultimately the students reach
161

via free access
CHAPTER 8
this goal and are able to argue in a sophisticated mathematical level: this is
the notion of “learning to argue”. If we think in terms of an interactional
arena in everyday mathematics classroom situations, argumentation also
appears as a major feature of it, though in a different way: with regard to
learning mathematics one usually assumes that participation in argumen-
tations, which appear to be rather explicit and sophisticated, is a pre-
condition for the possibility to learn, not only the desired outcome. In this
sense, learning mathematics is argumentative learning. (Krummheuer, 2007,
p. 62)
Erna Yackel (2001) argues that one can use Toulmin’s argumentation scheme, as
elaborated by Krummheuer (1995), as a methodological tool to demonstrate how
learning progresses in a classroom. Presumably then she shares a similar notion of
argumentation, however she does not make this explicit.
In her work with her colleagues Paul Cobb and Terry Wood, Yackel elaborates
the concept of “sociomathematical norms”.
Our prior research has included analyzing the process by which teachers
initiate and guide the development of social norms that sustain classroom
microcultures characterized by explanation, justification, and argumentation
(Cobb, Yackel, & Wood, 1989; Yackel, Cobb, & Wood, 1991). Norms of this
type are, however, general classroom social norms that apply to any subject
matter area and are not unique to mathematics. For example, ideally students
should challenge others’ thinking and justify their own interpretations in
science or literature classes as well as in mathematics. In this paper we
extend our previous work on general classroom norms by focusing on nor-
mative aspects of mathematics discussions specific to students’ mathematical
activity. To clarify this distinction, we will speak of sociomathematical
norms rather than social norms. For example, normative understandings of
what counts as mathematically different, mathematically sophisticated, mathe-
matically efficient, and mathematically elegant in a classroom are sociomathe-
matical norms. Similarly, what counts as an acceptable mathematical explanation
and justifycation is a sociomathematical norm. (Yackel & Cobb, 1996,
pp. 460–461)
While they do not refer explicitly to proof or proving, it is clear that “an acceptable
mathematical explanation and justification”, at least in a community of mathema-
ticians, is a mathematical proof.
Wood (1999) later departed from her colleagues’ approach to argumentation,
and provided some explicit definitions to clarify her interests.
I define argument as a discursive exchange among participants for the purpose
of convincing others through the use of certain modes of thought. In the study
reported here, argumentation is viewed as an interactive process of knowing
how and when to participate in the exchange. As will be seen, this study of
argumentation as a discursive and interactive process differs from Yackel’s
(1992) examination of the type of arguments children offer when they disagree
162

via free access
ARGUMENTATION
(e.g., forms of proof by counterexample) and from Krummheuer’s (1995)

analysis of the accounting of statements children give in situations of disagree-
ment (e.g., claims, warrants). (p. 172)
SUMMARY
In Table 27 we summarise the theorists cited, definitions of “argumentation” and
“argument”, and the link between argumentation and argument, for some of the key
researchers discussed above.
Table 27. Notions of argumentation
Author Theorists Argumentation is... Argument is...

cited
Duval Perelman – a kind of non-deductive reasoning anything which is
– arises when there is an argument advanced or used to
justify or refute a
proposition.
Balacheff Perelman – a social behaviour used to
convince
– an epistemological obstacle to
the learning of mathematical
proof
Douek, Toulmin – the process that produces a a reason or reasons
Boero (briefly) logically connected (but not offered for or against a
necessarily deductive) proposition, opinion or
discourse about a given subject measure
– the text produced through that
process
– one or more logically connected
“arguments”
Pedemonte Toulmin the process of conjecturing
Krummheuer, Toulmin – a social phenomenon, the result of
Yackel, – the intentional explication of the argumentation
Cobb reasoning of a solution,
– techniques or methods of
establishing the claim of a
statement,
– a pre-condition for the
possibility to learn
Wood Toulmin an interactive process of knowing a discursive exchange
how and when to participate in the among participants for the
exchange purpose of convincing
others through the use of
certain modes of thought
163

via free access
CHAPTER 8
ARGUMENTATION IN JAPAN
It is important when discussing the diversity of usages of “argumentation” in mathe-
matics education to point out that all the researchers mentioned above are working
in the Western cultural context. Sekiguchi and Miyazaki (2000) provide a description
of the process of “hanashi-ai” which they see as the counterpart to argumentation
in the West. They point out that the Toulmin model is a Western one, based on a
metaphor of conflict and war (Lakoff & Johnson, 1980, p. 4).
In contrast, in Japan, exchanging talks in either public or private is usually
referred to as “hanashi-ai”: The word means mutual conversation or consul-
tation, and does not signify a war. Because people try to avoid direct confron-
tation, they try to put their opinions ambiguously so that they can withdraw
or change them easily when others indicate opposition (Nakayama, 1989). As
a result, people in “hanashi-ai” do not usually bring up such full logical
defense devices like “grounds,” “warrants,” and “backing.” Even in those
situations where the social exchange model is working, people tend to avoid
bringing up logical armaments because they feel that arguing logically is
impersonal (“katakurushii”). (Sekiguchi & Miyazaki, 2000, Communication
and Argumentation in Japanese and Western Cultures, para. 10)
Though “hanashi-ai” may eventually conclude which solution is better, correct,
efficient, elegant, or whatever, competition among children is generally discouraged.
Therefore, in principle, no winner and no loser exist in “hanashi-ai,” unlike the
Western-style argumentation.
IDEAS FOR RESEARCH
Although research in mathematics education on argumentation is expanding, there

is still work to be done. Most fundamentally, the disagreement between Duval and
the Italian researchers is based not only on differing uses of the word “argumentation”
but also on the need for more research in this area. The relationship between
argumentation and proof, or perhaps more broadly between problem solving and
proof, is far from clear. As we have indicated in this chapter there has been a good
theoretical groundwork laid for research, but now more empirical work needs to be
do to apply and develop that groundwork.
The comments of Sekiguchi and Miyazaki, above, also suggest another direction
for future research. Comparative studies on proof and argumentation are rare, and
essays like Sekiguchi and Miyazaki’s offer us insight into another perspective that
sheds light on aspects of argumentation that might go unnoticed because they are
generally assumed by Western researchers. These comparisons could be extended
to other languages and cultures. Another way in which comparative research could
be informative is through comparing classroom proving practices in schools,
curricula, or textbooks. Knipping’s (2003b) research in French and German class-
rooms (see Chapter 10), and Cabassut’s (2005) research on French and German
textbooks offer two models for such research.
164

via free access
CHAPTER 9
TEACHING EXPERIMENTS
Most of the research in the previous chapters has looked at students’ proving
outside classrooms (in interviews or on assessments for example) or in classrooms
where the approach to proof was typical of the time and place. In this chapter
we will describe four approaches to teaching proof that could be considered
experimental. That is, they differ from standard practices in significant ways. The
first is that of Fawcett (1938) whose teaching was at the high school level (ages 13–17)
in the US, over a two year period including a total of 68 weeks of four, 40 minute
classes. The second we call the “debate” approach and is described in a number
of publications from the 1980s. This teaching was directed at younger students
(age 11–12) in France. The third comes from the Canadian elementary school
(grade 5, age 10–11) classroom of Vicki Zack, which has been described by her and
her colleagues since 1997. The fourth has been employed in Italian Grade 5–8 class-
rooms and has not been thoroughly described in English previously.
– What are the common elements in the teaching approaches?
– Are these approaches related to particular views of what proof is and what it is
for?
– What are some barriers to the wider use of these approaches in schools?
FAWCETT
One of the most extensively described successful teaching experiments related to

teaching proof is that of Fawcett (1938). Fawcett’s goal was to teach proof “as a
means for cultivating critical and reflective thought” (p. 1). He was working in the
context of the transition from what Herbst (2002b) calls the “era of originals” to
the “era of exercise”. In the era of originals high school geometry students were
expected to learn the proofs of major results, but also to prove corollaries and less
important theorems themselves. The content of geometry was central but oppor-
tunities for students to prove also were included (p. 290). In the era of exercise,
proof became an explicit object of study and textbooks restructured the presentation
of geometry (which up to that time had pretty much followed Euclid’s Elements) to
include a few basic fundamental propositions and numerous exercises in which
students would prove propositions on the basis of the fundamental propositions. In
contrast to the “originals” students had proved in the past, the “exercises” were
“many, easy, and carefully graded” (Young et al., 1899, p. 136, cited in Herbst, 2002b,
p. 299). This was intended to make the high school geometry course a good context
for learning proof, while learning the content of geometry was largely shifted to the
junior high school and earlier.
165

via free access
CHAPTER 9
Herbst dates the beginning of the era of originals to the recommendations of the
Committee of Ten (Eliot et al., 1893, cited in Herbst, 2002b, p. 286). Over thirty
years later, however, Fawcett found that while there was general acceptance of the
idea that the main goal of high school geometry was teaching proof, there was little
evidence that this was being achieved. Fawcett’s work was intended identify teaching
methods that would achieve this goal.
As noted in Chapter 3, for Fawcett, a student “understands the nature of deductive
proof when he understands:”
1. The place and significance of undefined concepts in proving any conclusion.
2. The necessity of clearly defined terms and their effect on the conclusion.
3. The necessity of assumptions or unproved propositions.
4. That no demonstration proves anything that is not implied by the assump-
tions. (p. 10)
It is further assumed that a pupil who understands these things will also
understand that the conclusions thus established can have universal validity
only if the definitions and assumptions which imply these conclusions have
universal validity. The conclusions are “true” only to the extent that the
fundamental bases from which they were derived are “true.” Truth is relative
and not absolute. (p. 11)
For Fawcett the purpose of proving is the determination of truth. He makes no
reference to proving as explaining or exploring although the students in his course
did a fair bit of both. Fawcett also believes that the main reason to teach mathe-
matical proof in schools is as preparation to apply the infallible methods of
deductive reasoning to verify not only in mathematics, but also in other areas.
The concept of proof is one concerning which the pupil should have a
growing and increasing understanding. It is a concept which not only pervades
his work in mathematics but is also involved in all situations where conclusions
are to be reached and decisions to be made. Mathematics has a unique
contribution to make in the development of this concept. (p. 120)
Fawcett’s methods for teaching proof are, in some respects, quite traditional. He
makes no reference to students working together, except in the context of whole
class discussions led by the teacher. The context for teaching proving is geometry
and although Fawcett does note the importance of students being able to transfer
their ability to prove to non-mathematical domains, he does not discuss proving in
other areas of mathematics. Other aspects of his teaching are fairly radical, at least
compared to current practice. He summarises his methods as follows:
1. No formal text is used. Each pupil writes his own text as the work develops
and is able to express his own individuality in organization, in arrangement,
in clarity of presentation and in the kind and number of implications
established.
2. The statement of what is to be proved is not given the pupil. Certain
properties of a figure are assumed and the pupil is given an opportunity to
discover the implications of these assumed properties.
166

via free access
3. No generalized statement is made before the pupil has had an opportunity

to think about the particular properties assumed. This generalization is made
by the pupil after he has discovered it.
4. Through the assumptions made the attention of all pupils is directed
toward the discovery of a few theorems which seem important to the
teacher.
5. Assumptions leading to theorems that are relatively unimportant are
suggested in mimeographed material which is available to all pupils but
not required of any.
6. The major emphasis is not on the statement proved, but rather on the
method of proof.
7. The extent to which pupils profit from the guidance of the teacher varies
with the pupil and the supervised study periods are particularly helpful in
making it possible to care for these variations. In addition individual
conferences are planned when advisable. (p. 62)
Fawcett used a number of measures to evaluate how successful his teaching
approach had been. On a state-wide test of geometry the students in his class had
scores that were in the same range as the scores for all students who wrote the test
and the median for his class was above the median for all students. That Fawcett’s
students did as well as the other students is significant as his students “had
covered only a small part of the geometric content usually studied in plane geo-
metry” (p. 102). His students had to derive many results that other students had
memorised, and they commented afterwards that they found the time too short for
the test and “if the time had been longer they could have worked out many of the
results which, in view of their limited acquaintance with the subject matter, was
entirely new to them” (p. 102). Fawcett also compared his students to other
classes with a test in which they had to analyse arguments in non-mathematical
contexts. Not surprisingly, given that this was a focus in Fawcett’s teaching and
not in the other classes, his students showed improvement on this test between a
pre-test and a post-test, and the students in the other classes did not (p. 103). The
teacher of a control class also developed a test requiring the students to prove
geometric statements, and Fawcett’s class did significantly better on this test as
well (p. 105).
In spite of these positive results, Fawcett’s approach was not widely adopted
in schools. The Second World War may have shifted educational priorities away
from the development of critical thinking and towards more technical skills. Or
it may simply have been too radical a departure from existing practice to be
acceptable.
One related approach, with some significant differences, was proposed by
Christopher Healy (1993). His “Build-a-Book” geometry is like Fawcett’s approach
up to point 3 (above), though Healy seems unaware of Fawcett’s work. No text is
used and the theorems are all discovered by the students. Healy, however, did not
focus on proof and the “explanations” his students included in their textbook range
from two-column proofs through generic examples to simple restatements of the
167

via free access
CHAPTER 9
“theorem” with no additional argument. Healy did provide a context in which

students were highly motivated and became passionate about mathematics, and it
would be interesting to see what the combination of his approach with Fawcett’s
would produce.
Another related approach is the Moore method based on teaching methods of
the mathematician R. L. Moore (1882–1974). One of Moore’s students, F. B. Jones
(1977) describes Moore’s teaching this way:
After stating the axioms and giving motivating examples to illustrate their
meaning he would then state some definitions and theorems. He simply read
them from his book as the students copied them down. He would then instruct
the class to find proofs of their own and also to construct examples to
show that the hypotheses of the theorems could not be weakened, omitted, or
partially omitted. ... When a student stated that he could prove Theorem x, he
was asked to go to the blackboard and present his proof. Then the other
students, especially those who hadn’t been able to discover a proof, would
make sure that the proof presented was correct and convincing. ... When a
flaw appeared in a “proof ” everyone would patiently wait for the student at
the board to “patch it up.” If he could not, he would sit down. Moore would
then ask the next student to try or if he thought the difficulty encountered
was sufficiently interesting, he would save that theorem until next time and
go on to the next unproved theorem ... Occasionally theorems got left over
indefinitely but nearly all of these would be proved in some subsequent year.
(pp. 274–275)
The Moore method is now used chiefly in advanced university mathematics
courses. We can find no evidence that Fawcett was aware of Moore’s teaching
methods but by the time Fawcett was doing his teaching experiment in the late
1930s Moore was very well known for his mathematical work.
Given Fawcett’s success one might ask why Fawcett’s methods could not be
used in schools today. It may be that they can; however, there are significant
differences in the teaching context now that may make this difficult.
Reid (1997) describes some obstacles he encountered when adopting Fawcett’s
methods. These included time limits imposed by the curriculum, the nature of
school examinations, and the existing didactic contract in the classroom. Fawcett
was able to work with his class over a two year period focussed on Euclidean
geometry, while Reid was limited to two months in which the students studied both
co-ordinate and Euclidean geometry. The approved textbook, provincial curriculum
and examinations included considerable algebraic manipulations even in the
context of geometry which the students were expected to be fluent in, further
limiting the time that could be spent on proof. In addition, the students had been
schooled to expect their teacher to provide information clearly and for their role to
be accurate reproduction of procedures. The more open structure of Fawcett’s
method was initially resisted as a violation of this didactic contract. In addition by
calling into question parts of the contract it left the students confused as to what
exactly their role was.
168

via free access
Another contextual difference between Fawcett’s situation and the present is the
nature of the student body. In Fawcett’s time only about 75% of 14–17 year olds
attended school and only about 50% graduated (James & Tyack, 1983, p. 401, cited
in Stanic, 1986, p. 194). This suggests that high schools were then selective in a way
that they no longer are. Hence Fawcett’s students may have been better prepared or
more motivated than some students in high schools today.
There are also hints in Fawcett’s report that suggest further difficulties that a
teacher adopting his approach might face. They are related to his goal that his
students be able to transfer their ability to prove to “non-mathematical material”
(Fawcett, 1938, p. 21) and so are relevant to anyone who advocates teaching proof
as a way to improve reasoning in non-mathematical contexts. Fawcett achieved his
goal of transfer, as is indicated by these comments he received from parents by
way of another teacher who conducted interviews with them:
The parents fear that the course may tend to inhibit in the boy the power of
imagination for creative writing in English. For example, when he was writing
of a personal experience for an English assignment he resented some sugges-
tions his mother made in order to add interest to the composition on the basis
that the suggestions were not facts. He wished to write only in a scientific
manner.
The mother fears that the girl may carry her criticism to the point of quibbling,
however. In some cases she has gone to the point of criticising authorities on
subjects about which she knew nothing. (p. 109)
Fawcett’s 1938 NCTM yearbook was widely read, and it was reissued in 1995. His
ideas have been influential, but his approach has not been widely adopted by
teachers, perhaps for the reasons discussed above. The next approach we will
describe, the debate approach, differs in content, age level and context but shares
this pattern of being more influential in the world of ideas than implemented in
classrooms.
THE DEBATE APPROACH
A group of researchers working in Grenoble and Lyon in the late 1980s explored
the potential of a teaching approach centring around “scientific debates”. They
were attempting to address a shift in the curriculum of the time. Previously, proof
had been introduced abruptly in the third year of collège (when students are about
13 years old). The new curriculum called for an introduction to deductive reasoning
in the first year of collège (11 year olds). The aim was to lessen students’ difficulties
with proof, but the curriculum change left teachers unclear about what exactly was
meant by an introduction to deductive reasoning. The Grenoble/Lyon group sought
to offer a possible model for teaching in the first two years of collège (11–12 year
olds) (Arsac, Chapiron, Colonna, Germain, Guichard & Mante, 1992, p. 1).
This group included Balacheff and adopted his distinction between “proofs”
(preuves) and “mathematical proofs” (démonstrations) which we described in
Chapter 2. Briefly, a proof is a verification accepted by a social group, and a
169

via free access
CHAPTER 9
mathematical proof is a proof accepted by mathematicians. Mathematical proofs

are based on axioms and deductive logic, and make reference to abstract objects,
not objects in the tangible world (Balacheff, 1987, p. 148, Arsac, Chapiron et al., p. 6).
Two roles are seen for mathematical proofs: convincing/verifying, and understanding/
explaining (Arsac, Chapiron et al., pp. 6–7) however the main focus, as we will
see, is on convincing. In classifying proofs the researchers use Balacheff’s categories,
which we described in Chapter 7.
The teaching is based on students exploring mathematical statements chosen so
as to be within the students’ capabilities to explore and likely to provoke disagreement
and debate. For example, students were asked to respond to the question: “Dans
l’expression n × n – n + 11, si on remplace n par n’importe quel entier naturel, obtient-
on toujours un nombre qui a exactement deux diviseurs ?” [In the expression
n × n – n + 11, if n is replaced by any whole number is the result always a number
with exactly two divisors?] (Arsac, Chapiron et al., p. 25). The students explore the
statement individually, and then in groups in what is called the “research period”.
During a first phase, called the research period, the students’ task is to solve
a given problem and to write their solutions on a poster. The problem is
chosen in such a way that several solutions can be reasonably expected
within the time allocated to a session (one hour or one hour and a half). The
fundamental requirement for the teacher is that she must not interfere at a
mathematical level. For example, she must not give indications to students
about the validity of their tentative productions. Insofar as we want students
to accept that they are in charge of their own work, it is necessary that the
aim of producing a solution to be discussed appears by itself sufficient to
motivate their activity. (Arsac, Balacheff & Mante, 1992, p. 9)
The research period is followed by the debate period.
During the second phase, called the debate period, aiming at a collective
discussion about the proposed solutions, the organization is the following:
Students’ solutions are written on a large sheet of paper and are then displayed
as posters on the wall of the classroom. Each team has to analyze the posters
and their spokes-person tells the class their criticism and suggestions. The
criticism must be accepted by the team whose poster is discussed. Since the
students involved are 13 to 14 years old, it is not possible to leave them free
of any regulation. The management of the activity is then left to the teacher.
Therefore, teachers’ interventions are needed here for what we can consider
as ‘technical reasons’. But, more than that, it will be the responsibility of the
teacher to institutionalize the debate’s outcome at the very end. Thus, students
as well as the teacher must have been witnesses of the phenomena to appear
during this debate period. (p. 9)
The role of the teacher in the final synthesis of the debate is to draw attention to
mathematical rules that have been used in the debate and to note the insufficiency
of pragmatic proofs (in Balacheff’s terms) that have been put forward (Arsac,
Chapiron et al., p. 21).
170

via free access
The influence of Lakatos (1976) on this teaching model is clear. In it the evidence
for a conjecture, a proof, is offered to the community and others comment on it and
point out flaws. These are then corrected and perhaps new arguments are brought
forth until finally the proposition is accepted into the body of mathematical know-
ledge. In addition this approach places a strong emphasis on communication. This
could provide a context that encourages the formulation of reasoning.
Lampert (1990) was also inspired by the work of Lakatos and was aware of
Balacheff’s research. She taught a grade 5 class (10 years old) in the US using the
debate approach, with some modifications.
As students volunteered their solutions to a given problem, I write them on
the board for consideration, and I put a question mark next to all of them....
Once the list of students’ solutions was up on the board, they were open for
discussion and revision.... If they wanted to disagree with an answer that was
up on the board, the language that I have taught them to use is, “I want to
question so-and-so’s hypothesis.” ... I always ask them to give reasons why
they questioned the hypothesis, so that their challenge took the form of a
logical refutation rather than a judgment. (p. 40)
Lampert also placed some emphasis on portraying mathematics as exploratory, in
keeping with Lakatos’s historical analysis. She is especially aware of the “cultural”
side of teaching proving:
I assumed that changing students’ ideas about what it means to know and do
mathematics was in part a matter of creating a social situation that worked
according to rules different from those that ordinarily pertain in classrooms,
and in part respectfully challenging their assumptions about what knowing
mathematics entails. (p. 58)
Teaching based on the debate approach does have some shortcomings. One of these
is pointed out by Arsac, Balacheff and Mante (1992), who report that in classrooms
the arguments offered are often not entirely founded on mathematical bases, but
include appeals to social and personal factors. Students rely on their personal
authority as members of the social structure of the class to verify their statements
by reference to their own authority.
A second weakness in the debate approach may be the focus on verifying as the
main role for mathematical proofs (or proofs more generally). While explaining
was also included as a role, the debate context stresses convincing/verifying more
than explaining. As we noted in Chapter 5 proof serves many purposes and a focus
on verifying might be contrary to students’ inclinations.
There are also difficulties with the debate approach related to the time available
and the teacher’s role and feeling of responsibility as an expert. Arsac, Balacheff
and Mante describe the pressure the teacher felt to prompt the students in order to
save time, and her (sometimes unconscious) shifting of attention towards statements
she knows will be productive and away from statements she knows will not be.
While the contexts of Fawcett’s teaching and that of the Grenoble/Lyon group
are very different (separated by fifty years, different cultures, and focussed on students
171

via free access
CHAPTER 9
of different ages) there are some suggestive common features. Students begin with
a problem, not with a statement to be proven. What they prove are their own
conjectures. And their expression of their reasoning is largely their own choice but
constrained by a social context.
EXPECTING EXPLANATIONS
Fawcett’s students were in high school, those taught by the Grenoble/Lyon group
were in the first years of collège (roughly junior high school in North American
terms), and Lampert taught in an upper elementary school. This is important as
groups such as the NCTM (2000) have advocated that proof be taught at all levels.
Reid and Zack (2009) provide a description of another teaching method used at the
upper elementary level, in Vicki Zack’s grade 5 classroom.
Zack taught in a private school with a fairly diverse student body. The provincial
mathematics curriculum is focussed on the development of competencies such as
reasoning and problem solving and these are emphasised more than content. In
Zack’s school there is a problem-solving culture in which the students were expected
to support their positions and present arguments for their point of view in most
areas of the curriculum. This, along with the initiatives envisioned by the NCTM
Standards (NCTM, 1989, 1991, 2000), provided the context for Zack’s approach to
teaching mathematics.
It should be noted that Zack did not intend to teach proof. Her focus was on
problem solving and communication. However, her classroom turned out to be a
context in which students proved.
Reid and Zack (2000) structure their description of Zack’s teaching around five
aspects: problem solving, teaching time, conjecturing, expectations, and teacher
expertise. While problem solving is very important in the school, not all mathematics
lessons involved non-routine problem solving. The children learned traditional
algorithms and dealt with the language of the mathematics textbook as well, so that
they would be familiar with the “formal school textbook” language of mathematics.
One notable feature of the non-routine problem solving that the students did is that
the criteria for deciding whether they had solved the problem were in the hands of the
students. Specific results were checked by repeating calculations, identifying mistakes,
or comparing with solutions obtained using other methods. The methods themselves
were evaluated by the students, according to criteria they developed (Zack, 2002).
The students were given a great deal of time to experiment with, think through,
discuss and refine their solutions to problems. For example, the Problem of the
Week tasks were assigned and worked on independently in a 90 minute class on a
Monday. They recorded this work in their Math Logs, which Zack looked over that
evening. They were given time on Tuesday to review their Logs and Zack’s
comments on them. On Wednesday discussions took place, again in a 90 minute
class. The students worked first in pairs or threes and then came together in a group
of four or five. In these small groups they compared solutions and discussed further,
and then reported to the class, with more discussion following. On occasion class
time on Thursday and Friday was also used to allow the discussions to come to a
172

via free access
fruitful conclusion. This means that the exploration of the problems occured over a
significant period of time, almost four hours per task. This allowed for conjectures
to be made and explanations sought without being artificially cut short by time
constraints. Extensive time allotted for work related to proving is also an aspect of
other studies in teaching proving at the upper elementary level (for example, the
generous time periods in the studies reported by Maher & Martino, 1996ab).
In this approach, the processes of conjecturing and proving are intertwined in
two ways. Proving makes use of insights gained through the explorations that led
to conjectures (Zack, 2002). Conjectures are also used as the basis for proving
(Reid, 2002a).
As we mentioned above, the school is one in which the children are expected to
publicly express their thinking, and engage in conjecture, argument, and justification
throughout their elementary school life. The students pursue questions of personal
interest in mathematics (Zack & Graves, 2001) as well as in literature and social
studies (Zack, 1991, 1999b).
The groundwork laid during the year included an expectation that the children
would be looking for patterns, and that they could be nudged to think about
the mathematical structure underlying the pattern (Zack, 1997). In addition,
there were expectations that everyone’s answers should be considered and
that answers should not be changed without discussing how they arose and
what might be the source of an error. These support the development of
beliefs which Lampert (1990) identifies as important to mathematical thinking.
(Reid & Zack, 2009, p. 140)
There is a constant expectation for both explanation and for generalisation.
The final aspect of the teaching method that Reid and Zack describe is teacher
expertise. This teaching approach is based not in mathematical expertise but in
language expertise. They note that “in terms of her mathematical background,
Zack could be considered a typical elementary school teacher in that she describes
her background in formal mathematics as weak” (p. 143). However, she has other
expertise that is relevant to teaching proof. She has studied closely how meaning
is constructed as her students express their ideas, and listens closely, recognising
potentially fruitful avenues and seizing opportunities to provoke discussion.
Looking back at the teaching the Grenoble/Lyon group, Zack’s teaching presents
some interesting contrasts. Most notably, her intent was not to teach proof and her
background is not in mathematics, however, in some ways her approach was more
successful than the debate approach. This may be linked to the different roles given
to proving in the two contexts: primarily convincing in the debate approach and
primarily explaining in Zack’s classes.
ITALY
Boero, Mariotti, and their colleagues (Boero, Garuti, Lemut, et al., 1996; Boero,
Garuti & Mariotti, 1996) have reported a number of results related to a teaching
experiment in an Italian grade 8 class in which an extended study of sun shadows
provided the basis for conjecturing and proving. The two classes observed had 20
173

via free access
CHAPTER 9
and 16 students, and the students were beginning their third year with the same
teacher. In previous years the students had experiences with producing conjectures,
supporting those conjectures with arguments, and writing down their reasoning
(see Boero & Garuti, 1994 and Boero, Chiapini, Garuti & Sibilla, 1995).
The context of sun shadows was chosen because it is a context in which problems
can be explored and conjectures made in a number of ways, with the support of
everyday experience, experiments, and drawings. It allows for “conjectures which
are meaningful from a space geometry point of view, not easy to be proved and
without the possibility of substituting proof with the realization of drawings”
(Boero, Garuti, Lemut, et al., 1996, p. 115). Prior to the activity described here the
students had spent about 80 hours of class time on sun shadows, including obser-
vation and record keeping over hours and months, geometrical modeling, problem
solving concerning the height of inaccessible objects, and activities in which they
had to imagine “different positions of the sun and of the observer in order to
produce hypotheses concerning the shape and the length of the shadows” (Boero,
Garuti & Mariotti, 1996, p. 123).
As an example of the sort of activites that support students’ proving undertaken
in these classes, consider the following activity, which took place over about 10
hours of class time. The problem posed was:
In the past years we observed that the shadows of two vertical sticks on the
horizontal ground are always parallel. What can be said of the parallelism of
shadows in the case of a vertical stick and an oblique stick? Can shadows be
parallel? At times? When? Always? Never? Formulate your conjecture as a
general statement. (Boero, Garuti, Lemut, et al., 1996, p. 115)
The students worked on this problem individually or in pairs, according to their
own preference. Some long thin sticks and polystyrene platforms were provided to
allow for concrete experiments, but the lighting in the classroom was unsuitable for
producing shadows, so the students’ experiments had to be, at least partly, thought
experiments. Many students began by using the sticks or pencils and moving the
sticks or changing their own perspective to help them visualize the situation. Others
considered the problem with their eyes closed, relying entirely on mental represen-
tations. The students then wrote down their conjectures.
The teacher led a discussion, in which the students’ conjectures were collected
and clarified, resulting in a list like this one:
If sun rays belong to the vertical plane of the oblique stick, shadows are parallel.
If the oblique stick moves along a vertical plane containing sun rays, then
shadows are parallel.
The shadows of the two sticks will be parallel only if the vertical plane of the
oblique stick contains sun rays. (Boero, Garuti, Lemut, et al., 1996, p. 115)
The first two statements express different (correct) results, differing because in
the first the oblique stick is held fixed and the possible positions of the sun are
considered, while in the second the position of the sun is fixed and the possible
positions of the oblique stick are considered. The third statement asserts that the
174

via free access
two results specify all the situations in which the shadows can be parallel. Through
further discussion these statements were combined to form two new statements:
If sun rays belong to the vertical plane of the oblique stick, shadows are
parallel. Shadows are parallel only if sun rays belong to the vertical plane of
the oblique stick.
If the oblique stick is on a vertical plane containing sun rays, shadows are
parallel. Shadows are parallel only if the oblique stick is on a vertical plane
containing sun rays. (Boero, Garuti, Lemut, et al., 1996, p. 115)
A mathematician might object that the language could be made more concise; for
example, “Shadows are parallel if and only if sun rays belong to the vertical plane
of the oblique stick,” but in Italian, as in English, this phrasing would not be
distinguished in everyday speech from “only if ”. In this case, clarity of expression
is valued over conciseness.
The students then worked individually, comparing their original conjectures to
the conjectures produced in the class discussion and considering whether it would
be possible to test the collective conjectures by making an experiment. The feasibility
of experimentally establishing the truth of the conjectures was then discussed by
the whole class:
During the discussion, gradually students realize that an experimental testing
is “very difficult”, because one should check what happens “in all the infinite
positions of the sun and in all the infinite positions of the sticks”. (Boero,
Garuti, Lemut, et al., 1996, pp. 115–116)
This reflection and discussion took place over an extended period of time (about
3 hours) with the intent of clarifying the need for proving to verify the conjectures,
and the goal of verifying the conjectures in general.
The students then worked in pairs to prove the first sentence (the “if”) for each
conjecture, and then individually they wrote a proof. The proving of the second
sentence (the “only if”) for each conjecture was done in a teacher-led discussion,
followed by individual proof writing. There was then a final class discussion, and
the students wrote reports on the entire activity at home.
As in Vicki Zack’s class, teaching time, problem solving, conjecturing and
explanations play significant roles in the Italian classes. As was noted above, many
hours (80+) were spent exploring sun shadows. This means that there was sufficient
time for students to explore significant problems and develop and refine conjectures.
Explanations and verification of conjectures were also stressed and explicitly
discussed as a necessary part of mathematical activity.
SUMMARY
While it is not possible to prescribe any teaching method on the basis of the existing
research on proof and proving, some suggestions can be made. The approaches
described above have features in common, that, given the range of contexts they
come from, indicates that these features are of value.
175

via free access
CHAPTER 9
First, note that they rely on conjectures generated by the students, and they rely
on the students to verify or explain their conjectures.
Second, there is no prescription of a certain form for proof-texts. Precision in
language is assumed to result from the students’ discussions.
Third, the focus is on the nature of the discourse, not on the content. Fawcett’s
students learned some geometry, but less than they might have otherwise, and
that did not concern him. The problems solved in the Debate context, in Zack’s
classroom and in the Italian classrooms did not focus on results required by the
curriculum.
Finally, the mathematical activity occurs in a social context in which there is an
expectation for explanations and in which accommodation is made for the time and
attention explaining deductively requires.
IDEAS FOR RESEARCH
The teaching experiments described above suggest characteristics that proof teaching
more generally could have. However, there are many questions remaining. Perhaps
most strikingly, there are no recent teaching experiments at the high school level
that have been described thoroughly. What approaches to teaching proof might be
successful in context where less time is spent on geometry? Could Fawcett’s
approach be applied in other areas of mathematics? There are also no studies in
early elementary schools. What approaches to teaching proof would establish the
basis for teaching proof in later grades? And clearly context is important to teaching.
In what ways are the approaches described in this chapter dependent on the
national and social contexts in which they are found?
176

via free access
PART 3
PROCESSES OF REASONING
AND ARGUMENTATION
The research reviewed in Part 2 is largely focussed on discrete entities: proof-texts,

and short episodes of reasoning. When longer instances of reasoning or argumentation
are discussed they are often described too generally to allow for comparison with
other instances. In this part of the book we will consider processes of reasoning and
of argumentation in such a way that comparisons become possible.
In Chapter 10 we look at argumentation processes and structures. We describe
the argumentation occurring during a classroom lesson through a representation
called an “argumentation structure”. Argumentation structures portray the big picture
of the entire proving process while keeping in sight individual arguments and their
place in the argumentation. Four argumentation structures that were observed in
different contexts are described and compared.
In Chapter 11 we look at patterns of reasoning, which are combinations of acts
of reasoning that occur during individual and small group mathematical activity.
Patterns of reasoning also portray a big picture, in this case reasoning over extended
times, without losing sight of the nature of the reasoning occurring from moment to
moment. Again, several patterns are described and contrasted. This allows us to
consider what it means to prove in a different light, as a pattern of needs, acts of
reasoning, and formulating of reasoning.
It might be suggested that the research we have reviewed in Part 2 provides
a sufficient basis to move from research on what occurs in classrooms and in
students’ thinking to research that explores new methods of teaching proof. In fact,
in Part 4 we take some small steps in that direction. However, we believe that there
is still a need to understand both more broadly and more deeply classroom proving
processes and students’ reasoning. Hence, our interest in argumentation structures
and patterns of reasoning.
This belief is based on a view of change as being always determined by what
is. In psychological terms this view is common to most contemporary theories of
learning. What a person can learn at any point is determined by what that person
knows at that point. In the context of proof this means that the ways of reasoning,
or patterns of reasoning, a student can learn are determined by the ways of reasoning
or patterns of reasoning that person already engages in. By better understanding
patterns of reasoning that occur in students’ mathematical activity now, we under-
stand better what must be the basis for teaching in the future.
Similarly, in sociological terms, we view the change that is possible for a social
entity, be it a classroom, education system or nation, as being determined by the
present state of that social system. Changes to how teachers teach proof must be
based on a detailed understanding of how teachers already teach proof, for that is
where the process of change must begin.
177

via free access
PART 3
For this reason we welcome studies like those of Herbst (2002ab) that explore
the history of the teaching of proof. By better understanding the changes in the past
that brought us to present practices, we understand those practices in a way that can
support future changes. Being aware of the consequences of past efforts of reform
also provides a useful caution against too hasty innovation.
We do not wish to claim that we are unique in being aware of the need to base
suggestions for change on careful research of existing conditions. Harel and Sowder,
for example, examined university students’ reasoning carefully before making sugges-
tions for teaching. Their papers from the late 1990s (e.g., Harel & Sowder, 1996,
1998) outline the theoretical framework of proof schemes they developed to describe
student’s reasoning, while more recent papers (Harel, 2001; Sowder & Harel, 2003)
have reported teaching experiments based on what they learned about students’ proof
schemes.
In keeping with our emphasis on the value of multiple perspectives, we propose
the methods of describing classroom proving processes and students’ patterns of
reasoning in Chapters 10 and 11 not as rivals to frameworks like Harel and Sowder’s,
but as alternative perspectives that capture some of what they miss, and no doubt
miss some of what they capture.
178

via free access
CHAPTER 10
ARGUMENTATION STRUCTURES
In this chapter we describe argumentation structures, which are the global, overall
structures of the arguments that occur in a classroom proving process. These
structures, first described by Knipping (2003b) allow one to examine the entirety of
the proving process, without losing sight of the details that make it up. Under-
standing the rationales and the contextual constraints that shape these argumentations
can help us to improve our efforts in teaching proof, by giving us greater insight into
the nature of existing proving processes in classrooms.
– What teaching goals seem to be indicated by the different structures identified?
– What constraints in schools or in teaching generally might influence the shapes
of these structures?
– How much variation would you expect to find between the structures identified
in different teachers’ classrooms compared to the variation you would find between
different lessons taught by the same teacher or between different national school
systems and cultures of teaching?
TOULMIN’S FUNCTIONAL MODEL AND ARGUMENTATION STRUCTURES
Analyses of argumentations in mathematics lessons have received increasing

attention in recent years (Knipping, 2003b; Krummheuer, 1995, 2007; Pedemonte,
2007) and have become a means to better understand proving processes in class. As
discussed in Chapter 8 the work of Toulmin has provided researchers in mathematics
education with a useful tool for research, including arguments in classrooms
(Knipping, 2003b; Krummheuer, 1995) and individual students’ proving processes
(Pedemonte, 2002).
Toulmin (1958) describes the basic structure of rational arguments as a linked
pair datum + conclusion (see Figure 17). This step might be challenged and so it
is often explicitly justified. A ‘warrant’ is given to establish the “bearing on the
conclusion of the data already produced” (p. 98). These warrants “act as bridges,
and authorize the sort of step to which our particular argument commits us” (p. 98).
While Toulmin acknowledges that the distinction between data and warrants may
not always be clear, their functions are distinct, “in one situation to convey a piece
of information, in another to authorise a step in an argument” (p. 99). In fact, the
same statement might serve as either datum or warrant or both at once, depending
on context (p. 99), but according to Toulmin the distinction between datum, warrant,
and the conclusion or claim provides the elements for the “skeleton of a pattern for
analyzing arguments” (p. 99).
179

via free access
CHAPTER 10
Figure 17. Toulmin model.
Toulmin adds several other elements to this skeleton, only one of which will be
discussed here. Both the datum and the warrant of an argument can be questioned.
If a datum requires support, a new argument in which it is the conclusion can be
developed. If a warrant is in doubt, a statement Toulmin calls a “backing” can be
offered to support it.
Toulmin’s functional model of argumentation has been a foundation of Knipping’s
work (2001, 2002, 2003ab, 2004, 2008). But she also pointed out the need for a
model that would also allow the structure of the argument as a whole to be laid out.
As Toulmin notes “an argument is like an organism. It has both a gross, anatomical
structure and a finer, as-it-were physiological one” (Toulmin, 1958, p. 94). Whereas
Toulmin’s aim was to explore the fine structure, Knipping attempted to extend the
Toulmin model and to provide a model that also allows one to describe the gross
structure, which she calls the global argument. The method she proposed for
reconstructing arguments in classrooms is presented in Knipping (2008). Very briefly,
the analyses of proving discourses use the Toulmin model in order to identify
individual steps from data to conclusion. As the conclusions of some steps are
recycled as data for others, these steps join up into argumentation streams (AS);
however, these streams are generally not linear chains of steps. Argumentation
streams themselves are interconnected in more complex ways and together form the
argumentation structure. The analysis proceeds from the fine structure captured in
individual steps to the global structure of the entire argumentation.
In the following four different types of argumentation structures that occurred in
classroom proving processes will be presented and discussed. The first two types
Knipping (2003ab) called the source-structure and the reservoir-structure. They
were observed in proving processes from German and French classrooms. The last
two are from a classroom in Canada. We refer to them as the spiral-structure and
the gathering-structure.
THE SOURCE-STRUCTURE
In proving discourses with a source like argumentation structure, arguments and

ideas arise from a variety of origins, like water welling up from many springs. The
structure has these characteristic features:
– Argumentation streams that do not connect to the main structure.
– Parallel arguments for the same conclusion.
180

via free access
– Argumentation steps that have more than one datum, each of which is the conclu-
sion of an argumentation stream.
– The presence of refutations in the argumentation structure.
The source-structure is also characterised by argumentation steps that lack explicit
warrants or data. While this also occurs in the other types of argumentation structure
we will examine, it is frequent in the source-structure.
The teacher encourages the students to formulate conjectures which are examined
together in class. In some cases this means that students propose conjectures which
are unconnected to the overall structure. More than one justification of a statement
is appreciated and encouraged by the teacher. This diversity of justifications results
in an argumentation structure with parallel streams in which intermediate statements
are justified in various ways. False conjectures are eventually refuted, but they are
valued as fruitful in the meantime.
In argumentations with a source-structure a funnelling effect becomes apparent.
Towards the end of the argumentation only one chain of statements is developed
in contrast to the beginning where many parallel arguments are considered. For
example, in Figure 18 only a single chain of arguments (AS-7) occurs in the second
half of the argumentation. The argumentation begins in a very open way, drawing
on many sources, but is funnelled towards one final conclusion. Thus a variety of
justifications all support the overall argument.
Example 1: The Source-Structure in Mr. Lüders’ Class

Our first example of the source-structure comes from a grade 9 mathematics class
in Germany where the teacher, Mr. Lüders, sought to develop a proof of the Pytha-
gorean Theorem together with the class (Knipping, 2003b). The argumentation
structure (Figure 18) includes the features typical of the source-structure.
– Argumentation streams that do not connect to the main structure (AS-6)
– Parallel arguments for the same conclusion (AS-3, AS-4 and AS-5).
– Argumentation steps that have more than one datum, each of which is the
conclusion of an argumentation stream (in AS-5 and AS-7).
– The presence of refutations in the argumentation structure (in AS-2 and AS-6).
Argumentation stream AS-6 is an example of two of these features: refutations,
and argumentation streams that do not connect to the overall structure. In it Sebastian
conjectures, without providing any supporting data or warrant, that the area of the
rectangles in the right hand figure in the proof diagram (Figure 19) is equal to the
area of the inner squares. This argument was refuted by the teacher and was never
integrated into the larger structure.
The parallel arguments are AS-3, AS-4, and AS-5. They all lead to the conclusion
that the angle γ of the inner quadrilateral (in the left hand diagram in Figure 19) is 90°
on the basis of both visual and conceptual arguments. In AS-3 there is data missing
and one step of AS-5 lacks a warrant.
We will have a closer look at argumentation stream AS-2 that precedes these
parallel arguments and which includes another refutation. This will provide an
illustration of the process by which the classroom discourse is represented in an
181

via free access
CHAPTER 10
Figure 18. Overall argumentation structure in the proving process in Mr. Lüders’ class.
Figure 19. Proof diagram from Lüders’ class.
argumentation structure diagram. For more details on this process see Knipping
(2008). AS-2 begins when the teacher asks why the inner quadrilateral in the left
hand figure is a square and Stefanie gives a reason, but an insufficient one.
35 Teacher: A square in a square. This is a square in a square, can you tell
me, why this is a square? … Why is this a square, why is this
a square? You can tell me anything. Stefanie!
38 Stefanie: Because it has four sides of equal length?
39 Teacher: Actually I believe you. So, if this is side b and this is side a
and … And how long is this side?
41 Student: c too.
42 Teacher: c too, right? I have always taken the same triangle. Your
reasons are fine. So, what else? Katrin, pay attention please. ...
Is Stefanie’s justification sufficient for proving That this is a
square?
182

via free access
45 Student: No.
46 Torben: It also has four right angles?
47 Teacher: Why does it have four right angles? … You claim that this is a
right angle?
52 Teacher: ... Again, why is this a right angle? Four equal sides, everybody
who has ever eaten salinos [a rhombus shaped liquorice] knows
this, are not enough to form a square. ... Eric, you start.
(Knipping, 2003b, p.155, ellipses in original represent pauses)
Figure 20 shows the structure of this argumentation stream. Note that the
elements of it do not occur in chronological order. The teacher’s argument that the
sides of the inner quadrilateral are equal to the hypotenuse c of the triangle with
sides a and b (39–42) comes after Stefanie’s argument that the quadrilateral in the
square is a square, because it has four sides of equal length (38) but in the structure
of the argument it precedes it because it provides data she needs for her argument.
In this way the statement “it has four sides of equal length” (38), which is assumed
by Stefanie is supported. Stefanie’s conclusion, however, is questioned: “Is Stefanie’s
justification sufficient for proving that this is a square?” (43) and finally refuted by
the teacher who provides a reason from an everyday context. “Four equal sides,
everybody who has ever eaten salinos knows this, are not enough to form a square.”
(52/53). Here a mathematical refutation is supported by a backing from an every-
day context. Torben’s question suggests the piece of data that is missing from
Stefanie’s argument: “It also has four right angles?” (46). The teacher, by asking
for a justification why the angle is a right angle, implicitly turns Torben’s question
into data.
D1: a and b, segments of the outer square, are sides of congruent triangles
C/D2: The sides of the inner figure are the sides c of the congruent triangles
C/D3: The inner figure has four equal sides of length c
R: Four equal sides is not a sufficient condition for a figure to be a square
W: Salinos are an example of a figure that have four equal sides but are not square
C4: The inner figure is a square
D4: The inner figure has four right angles
Figure 20. Argumentation stream AS-2 from Lüders’ class.
183

via free access
CHAPTER 10
Example 2: The Source-Structure in Nissen’s Class

The following example of the source-structure was observed in another German
grade 9 mathematics class, taught by Ms. Nissen (Knipping, 2003ab). To begin the
proving process she sketched a drawing (Figure 21) on the chalkboard and asked
her students to interpret it.
Eight different argumentation streams (AS-1 to AS-8) make up the global argu-
mentation structure (Figure 22). Again the typical features of the source-structure
are evident:
– Argumentation streams that do not connect to the main structure (AS-6)
– Parallel arguments for the same conclusion (AS-1 and AS-2).
sion of an argumentation stream (AS-8).
– The presence of refutations in the argumentation structure (AS-3, AS-6).
AS-1 and AS-2 are parallel argumentation streams for the conclusion that the
side of the outer square is c. In AS-1 the argument is based on the conclusion that
the inner quadrilateral is a square, with the drawing of the proof figure as a
warrant. In AS-2 it is argued that the triangles make up the outer shape, again based
on the drawing.
Visually, argumentation steps which have more than one datum, each of which
is the conclusion of an argumentation stream, show up as long verticals connecting
several data with a conclusion. For example the conclusions of AS-4, AS-5 and
AS-7 are data for the first step in AS-8.
There are two interesting refutations within this argumentation structure. In
AS-3 Maren, in the process of describing the area of the outer square c², assumes
that the area of the inner square is b². The teacher contradicts her, refuting Maren’s
suggestion visually. She then develops together with the class an argument that
the side length of the inner square is b–a (AS-5) and therefore the inner square’s
area must be (b–a)². This provides the data for a justification that c² consists of a
square with side b–a and four congruent right triangles, which is noted on the
blackboard as follows: c²=(b–a)²+4rwD.
Figure 21. Proof diagram from Nissen’s class.
184

via free access
Figure 22. The source-structure in Nissen’s class.
Sascha’s conjecture (AS-6) is the other example of a refutation. Sascha claims

that two of the triangles form a square which the teacher refutes by having the class
put together the cut-out triangles. However, she says “I really like your idea, I think
ideas that lead to the right result in detours are wonderful” giving value to Sascha’s
conjecture even though she has refuted it. This refutation provides a context for the
argument in AS-7 that two of the triangles form a rectangle.
THE RESERVOIR-STRUCTURE
Argumentations with a reservoir-structure flow towards intermediate target-conclu-

sions that structure the whole argumentation into parts that are distinct and self-
contained. The statements that mark the transition from the first to the second part
of the proving discourse (shown as rectangles) are like reservoirs that hold and
purify water before allowing it to flow on to the next stage. Most of the features listed
above as characteristic of the source-structure are missing in the reservoir-structure,
with the exception of argumentation steps which have more than one datum each of
which is the conclusion of an argumentation stream. Argumentation steps that lack
explicit warrants or data occur, but less often than in the source-structure.
The most important feature of the reservoir-structure, which distinguishes it
from a simple chain of deductive arguments, is that the reasoning sometimes
moves backwards in the logical structure and then forward again. Initial deductions
lead to desired conclusions that then demand further support by data. This need is
made explicit by identifying possible data that, if they could be established, would
lead to the desired conclusion (indicated by the dotted line in Figure 23). Once
185

via free access
CHAPTER 10
Figure 23. The reservoir-structure in Pascal’s class.
these data are confirmed further deductions lead reliably to the desired conclusion.
This characterises a self-contained argumentation-reservoir that flows both forward
towards, and backwards from, a target-conclusion.
Example 3: The Reservoir-Structure in Pascal’s class

This example comes from a French level 4 (age 13–14) classroom where the proof
of the Pythagorean Theorem is the topic. We have seen it before, in Chapter 6, where
we used it as an example of abductive reasoning. We will now consider how it fits
into the larger argumentation structure.
The class has concluded (in AS-1) that the inner quadrilateral of the proof
diagram (see Figure 24) is a rhombus. They make an abduction from the desired
result that ABCD is a square, the datum that ABCD is a rhombus, and the general
rule that if a rhombus has a right angle it is a square, to conclude that ABCD has a
right angle. This becomes the target-conclusion in the argumentation streams AS-2
and AS-3. The three streams AS-1, AS-2 and AS-3 form a reservoir in which the
argumentation remains until it is sufficiently clarified to proceed.
Figure 24. Diagram used in Pascal’s class for Pythagorean Theorem proof.
186

via free access
A closed structure can also be found in the second part of the process, formed
by AS-5, AS-6 and AS-7. In contrast to the reservoir in the first part, the argu-
mentation in the second part only flows forwards. In AS-5 it is justified that the
area of the outer square equals (a+b)²–2ab. The subsequent argument (AS-6)
restricts and directs the argumentation towards the final target-conclusion, that
a2+b2=c2 (AS-7).
AS-8 represents a variation on this theme. At the end of the overall argumenta-
tion a student asks for a justification of the statement (a+b)²=a²+2ba+b² which is
used as a warrant in the argumentation. The teacher responds to a student’s
question by reminding the class that the statement has been proven in the previous
lesson, but proves the statement again (AS-8), together with the students. In this
case a warrant is needed in AS-6, and this motivates the deduction of a suitable
warrant in AS-8.
Example 4: The Reservoir-Structure in Dupont’s class

The proving discourse of another French lesson, taught by Mr. Dupont, is another
example of a reservoir-structure (See Figure 25). As in Pascal’s class distinct argu-
mentation chains (AS-1, AS-2, AS-3) in the first part are organised by an abduction.
The first stream is a move forward (AS-1). It concludes that the inner quadrilateral in
the proof diagram is a rhombus. Then, as in Pascal, an abduction establishes a datum
as a target-conclusion. The datum ‘The angle SRU is 90°’ is then justified (in AS-2)
and finally used to prove that the rhombus is a square (AS-3). The argumentation
structure of the second part of the proving discourse is more straightforward.
Figure 25. Overall argumentation structure in Dupont’s class.
THE SPIRAL-STRUCTURE
We have applied Knipping’s method of analysing classroom proving processes to

data from Canadian classrooms, gathered as part of the RIDGE project (see http://
www.acadiau.ca/~dreid/RIDGE/index.html). We have identified two argumentation
structures in this context: the spiral-structure and the gathering-structure.
The spiral-structure shares some characteristics with Knipping’s source-structure.
– Argumentation streams that do not connect to the main structure.
– Parallel arguments for the same conclusion.
187

via free access
CHAPTER 10
sion of an argumentation stream.
– The presence of refutations in the argumentation structure.
Argumentation steps that lack explicit warrants or data occur less often than in the
source-structure.
The main distinction between the spiral-structure and the source-structure is the
location of the parallel arguments. Recall that in the source-structure the parallel
arguments occur at the beginning of the process, and that later there is a funnelling
into a single stream leading to the final conclusion. In the spiral-structure the final
conclusion is repeatedly the target of parallel argumentation-streams. The conclusion
is proven again and again, in different ways.
We have observed the spiral-structure in two cases, one of which we will use as
an example here. This example comes from Ms. James’ grade 9 (age 14–15 years)
classroom in Canada. The class was trying to explain why two diagonals that are
perpendicular and bisect each other define a rhombus. The students had discovered
and verified empirically that the quadrilateral produced is a rhombus using dynamic
geometry software and the proving process led by the teacher was framed as an
attempt to explain this finding using triangle congruence properties.
Figure 26 shows the argumentation structure for this proving process. It displays
several of the features discussed above:
– Argumentation streams that do not connect to the main structure (AS-C).
– Parallel arguments for the same conclusion (AS-B, AS-D, AS-E).
– Argumentation steps that have more than one datum, each of which is the
conclusion of an argumentation stream (within AS-A and the final conclusions
of AS-B and AS-E).
– The presence of refutations in the argumentation structure (AS-D).
Figure 26. Argumentation structure in James’ class for the rhombus proving process.
188

via free access
In the argumentation structure three parallel arguments AS-B, AS-D, and AS-E
lead to one conclusion, that the four sides are congruent, which acts as the datum
for the final conclusion that the quadrilateral is a rhombus. In AS-B the congruency
of the sides is shown by showing that the four triangles formed by the diagonals
are congruent. In AS-D a student offers an alternative argument, based on the
idea that the quadrilateral cannot be shown to be a square. This argument is
listened to attentively by the teacher, who eventually refutes it. This is similar to
the treatment of conjectures in the classes where the source-structure was obser-
ved. Finally, in AS-E the teacher offers an alternative argument based on using
the Pythagorean Theorem instead of triangle congruency to establish that the four
sides are equal. This argumentation stream is unusual because of the lack of
warrants.
In the same classroom we observed a similar structure when the class was
explaining why two congruent diagonals that bisect each other define a rectangle.
These structures observed in Canada are similar in many ways to the source-structure,
but they differ from it in an important way. In the source-structure the parallel
arguments occur early in the proving process. The teacher invites input at this
stage, but once the basis for the proof is established, the teacher guides the class
to the conclusion through a structure that no longer has parallel arguments. In the
spiral-structure, however, the conclusions to the parallel arguments are almost
the final conclusion in the entire structure. In fact, the three parallel arguments
could stand alone as proofs of the conclusion. Having proven the result in one
way, the teacher goes back and proves it again and again. For this reason we
describe these structures as spiral. Comparison of the argumentation structures
in the lessons taught by Mr. Lüders, Ms. Nissen and Ms. James reveals some
similarities, but also significant differences in the teaching approach in these
contexts.
THE GATHERING-STRUCTURE
In the gathering-structure the argumentation includes the gathering of a large amount

of data to support several related conclusions. New data is introduced as needed
rather than being given initially, and the conclusions are also not specified in advance.
Metaphorically, the class moves along, gathering interesting information as it goes.
The gathering-structure differs from the source-structure and the spiral-structure in
that it does not include parallel arguments for a single conclusion and argument-
tation streams that do not connect to the main structure. However, like the source-
structure there are many argumentation steps that lack explicit warrants or data.
The gathering-structure is unlike the reservoir-structure as it includes refutations
and lacks backwards reasoning.
Our example of a gathering structure (Figure 27) comes from the same Canadian
grade 9 classroom as our examples of the spiral-structure, above. The students
had explored independently, using dynamic geometry software, whether three side
lengths given to them determined a unique triangle, and whether any three side
lengths would determine a triangle. The argumentation structure represents the class
189

via free access
CHAPTER 10
Figure 27. Argumentation structure in James’ class for the side-side-side proving process.
discussion afterwards. In AS-A they conclude that the three given side lengths
determine a unique triangle. Within this stream it is conjectured that more than one
triangle is possible but this is refuted. This was intended to be the final conclusion,
so that it could be used as a basis for later arguments, but students’ conjectures
led to further discussions, involving additional empirical data and leading to other
unanticipated conclusions. While the first conclusion was being discussed a
conjecture was made that it is always possible to create a triangle given any three
side lengths. AS-B is the argumentation stream resulting from the argument for and
against this conjecture (marked by a white rectangle). The conjecture is refuted (by
means of additional data gathered and brought in through AS-C) and its negation
becomes the final conclusion for AS-B (indicated by the small black dot after the
white rectangle). Having arrived at the two conclusions that three sides sometimes
determine a unique triangle and sometimes do not determine a triangle at all, the
students gathered more data related to the question of when three sides determine a
triangle. In AS-D the students combine their conclusions from AS-A and AS-B
with additional empirical evidence to conclude the triangle inequality: that a unique
triangle is possible only if the sum of any two of the given side lengths is greater
than the third.
Even though the same class is involved, this argumentation structure is quite
different from the spiral-structure. In the gathering-structure shown in Figure 27
there are no parallel arguments (like AS-B, AS-D and AS-E in Figure 26) or discon-
nected streams (like AS-C in Figure 26) and very few warrants.
A similar gathering-structure was observed in another proving process in which
the students were determining whether three given angle measures would determine
a unique triangle, and whether any three angle measures would determine a tri-
angle. In both of these proving processes the students’ arguments were based
largely on the empirical evidence offered by the dynamic geometry software. They
lacked, at that point, sufficient prior knowledge to deduce their conclusions. We
hypothesise that the lack of parallel structures is related to the use of empirical
arguments, as for the most part evidence accumulates rather than arising in distinctly
new ways.
190

via free access
IDEAS FOR RESEARCH
Above we have described four types of argumentation structures: the source-

structure, the reservoir-structure, the spiral-structure and the gathering-structure.
These have been observed in different contexts: in different countries and in classes
focused on different topics. We believe analysis of argumentation structures can
reveal different classroom cultures and approaches to teaching proof that follow
their own peculiar rationales. Our examples here suggest that these classroom
cultures may be influenced by the larger cultures in which they are embedded, but
also by the nature of the mathematics being studied and the teacher’s goals. This
suggests two directions for future comparative research.
Sekiguchi (1991) and Herbst (2002ab) examine the origins and nature of classroom
proving cultures in the US, typified by the form of proofs accepted: two-column
proofs. This form was not the norm in the classrooms we described above. It would
be interesting to see if there is an argumentation structure associated with the use of
this form, or several related to the topic under consideration. Comparative analyses
of this type can deepen our understanding of how classroom proving processes can
be an obstacle or an opportunity for students to learn to reason mathematically and
to engage in proving.
It would be interesting as well to examine another context: classrooms that
explicitly espouse an inquiry mathematics approach. A comparison of the argument-
tation structures that occur in such classrooms versus those that occur in more
traditional classrooms would deepen our understanding of the differences in teaching
and learning that occurs in these contexts.
As we mentioned in Chapter 8, Krummheuer (1995, 2007) sees individual
learning in the classroom as dependent on the students’ participation in “collective
argumentation” (2007). Krummheuer understands participation in argumenta-
tions as “a pre-condition for the possibility to learn” (p. 62) and investigates class-
room situations and their potential for learning. The classroom episodes discussed
by Krummheuer often contain only “the minimal form of an argumentation”
(Krummheuer, 1995, p. 243), consisting simply of data, warrant, and conclusion
rather than the more complex streams and structures described here. This is not
surprising given the grade level Krummheuer is investigating and that in most of
the classrooms argumentation was not subject or goal of the lesson. The hypothesis
that argumentations are “a pre-condition for the possibility to learn” would be
interesting to investigate in more elaborated argumentations for which analysing
and comparing argumentation structures provides a useful tool.
Further research on argumentation structures that build on Krummheuer’s work
together with Knipping’s must take into account differences in their terminology,
especially when writing in German. What we have called here a “step” (following
Toulmin’s usage) is referred to by Knipping (2003b) as a Schritt but by Krummheuer
(2003, Krummheuer & Brandt, 2001) as a Strang. Krummheuer’s “minimal form”
differs from what we call a step as it must include the three elements, data, warrant,
conclusion, while our step might have the data or warrant missing, and might include
multiple sources of data. Knipping (2003b) uses Krummheuer’s word for a step,
Strang, to refer to what we have called here a “stream”. Krummheuer does not
191

via free access
CHAPTER 10
describe anything exactly like a stream, but uses Mehrgliedrige Argumentation

(2001, p. 36) or “chain of argumentations” (2007, p. 65) to refer to a stream like
AS-B in Figure 26 in which each datum is the conclusion of a prior step. Finally, our
“parallel arguments” are called a Parallelargumentation by Knipping (2003b) and
are similar to what Krummheuer calls an Argumentationszyklus (2003, p. 248) or
cycles of argumentation (2007, p. 75).
192

via free access
CHAPTER 11
PATTERNS OF REASONING
In this chapter we look at a way of describing and analysing patterns of reasoning

that makes comparisons possible, both in detail and overall. Patterns of reasoning
are combinations of acts of reasoning that occur during individual and small group
mathematical activity. Describing them involves a portrayal of the reasoning that
occurs over an extended time, without losing sight of the nature of the reasoning
occurring from moment to moment.
Much of the research reviewed in Part 2 focussed on discrete entities: proof-
texts, and short episodes of reasoning. Research that focuses on proof-texts, the
end products of reasoning, do not allow us to understand the process that led up
to the proof. When reasoning over longer periods is discussed it is often
summed up in one or two descriptors that do not convey the complexity of the
process, hide distinctions we believe are important, and make comparisons
difficult.
By describing patterns of reasoning it is possible to look at reasoning closely
and to offer descriptions at a high level that do not obscure important details. This
allows us to consider what it means to prove in a different light, as a pattern of
needs, acts of reasoning, and formulating of reasoning, and to explore what
patterns of reasoning count as stages in learning to prove and implications for
teaching.
We focus on the patterns of reasoning in individuals (or small groups reasoning
together) reconstructed on the basis of empirical evidence in order to offer a model
for the detailed analysis of students’ reasoning when engaged in mathematical
activity.
– How do patterns of reasoning relate to argumentation structures?
– Are some patterns more characteristic of mathematics than of other discip-
lines?
– What impact could research on patterns of reasoning have on teaching?
We will look at several examples and the patterns of reasoning in them, which
we will describe in the terms based on Lakatos’s (1976) description of proof-analysis.
Five patterns will be described which we will refer to as:
– Deduce-Conjecture-Test cycle
– Proof Analysis
– Scientific Verification
– Surrender
– Exception and Monster Barring
193

via free access
CHAPTER 11
DEDUCE-CONJECTURE-TEST CYCLE
The first pattern of reasoning we will consider begins with a deduction, which
guides the making of a conjecture, which is then tested. The results of the test are
then used as the basis for further deducing and the pattern proceeds cyclically (See
Figure 28).
Deducing
↗ ↘
Testing ← Conjecturing
Figure 28. The Deduce-Conjecture-Test cycle.
The Deduce-Conjecture-Test cycle is illustrated by the mathematical activity of

Sandy, a precocious sixth grader (c. 11 years old). He was interviewed regularly
over an extended period by several researchers (e.g., Pirie & Kieren 1992). In this
case three researchers (David, Tom and Brent) are observing him solve Arithmagon
puzzles (See Mason, Burton & Stacey, 1985, for one presentation of these puzzles
and Figure 14 in Chapter 6 for another).
In these transcripts an em dash (—) indicates a silence of about one second.
Ellipses (...) indicate omitted speech. Three question marks (???) indicate inaudible
speech or a guess at partly audible speech.
Episode 1: The Original Arithmagon

1 David The idea is to figure out what three numbers you can put at the
corners, so that these two add up to 27 these two add up to 18 and
these two add up to 11. ... [See Figure 29]
Figure 29. The first Arithmagon puzzle.
2 Sandy OK [ eight seven eight ??? ] — — — — — — — — [I duh that work

???]
3 Thirteen — — — — — — — uh, hmm — — —
194

via free access
4 You need a fairly low one here.

5 Lets try 5 [Wrote “5” in Figure 30] — — — Well —
Figure 30. Sandy’s first guess: 5.
6 David Hmm
7 Sandy [You???] just go through all [of them???]
[Five second exchange between researchers looking for an eraser, at
the end of which Sandy erases the “5” in Figure 30]
8 Sandy OK let’s see. Just go through them all.
9 Zero doesn’t work there — because — just think — — umm — —
10 David so you go-
11 Sandy Seventeen. yeah — One, Ten, — Eighteen [Wrote “10” and “17” off to
the side, then wrote “1” “10” and “17” in Figure 31]
Figure 31. Sandy’s solution to the first puzzle.
12 David That was quick. How’d you do that? — —

13 Tom Wow
Sandy I knew this number if you kept on guessing then that would equal, —
then that would — then that would make these two numbers and you
just had to make one that added up to twenty seven
14 David Hmm — —
15 Sandy So then I was guessing, zero doesn’t work — —
[David proposes Sandy make up a puzzle, and Tom asks a question]
195

via free access
CHAPTER 11
16 Tom How did you get that one up there? That’s what I’m interested in. You
had five up there before
17 Sandy Well I thought it, I knew it was a fairly low number — because that’s
almost these two together
Discussion of Episode 1
It is difficult to reconstruct Sandy’s reasoning as his speech is often inaudible and
he is speaking to himself rather than explaining to the observers. However, it seems
clear that in line 4 he is making a conjecture “You need a fairly low one here.” He
is referring to the upper corner of the triangle, where he then writes “5”. How does
Sandy know he needs a fairly low number there? His mention of “thirteen” (line 3)
his guess of five (line 5) and his later comments (line 13 and in Episode 2, below)
suggest a possible reasoning process. (To make our reconstructions of his reason-
ing clearer we will use the symbols Sandy later used in Figure 32, but we do not
claim he is thinking in terms of general unknowns at this point).
Figure 32. Sandy’s symbols for the unknowns and givens in a general Arithmagon.
Sandy knows (or could know) that:

y + z = 27 (Given)
x is positive (Assumption)
z > y because 18 > 11 (x + z = 18; x + y = 11)
From these statements he may have concluded that z is at least 13 and y is at most
13. He may have mentioned “thirteen” (line 3) because he was thinking about y and
z as two parts of 27, and 13 is about half of 27. If z is at least 13 and x + z = 18
then x is at most 5.
An alternative reconstruction would be that he reasons in a similar way to
conclude that x < y and the fact that 5 is about half of 11 leads him to conjecture
that x = 5. While we cannot know if either of these is an accurate reconstruction of
his reasoning, it is difficult to imagine how he could have come to the conclusion
that “You need a fairly low one here.” without engaging in something similar.
Hence, the beginning of his reasoning process is deducing (in this case limits on
the possible values of x) followed by conjecturing (a value for x). The next step is
testing his conjecture, which he does mentally about line 5.
196

via free access
After finding that five is not correct, and saying that to find the answer he will
just go through all of them (lines 7 and 8), Sandy’s next guess is zero. Why not
four? If he is going to “go through all of them” and has tried five, why switch to
looking at zero next? There are two obvious possibilities, either or both of which
may have influenced Sandy’s reasoning:
– If x is five then c turns out to be 19, which is far from 27, suggesting that 5 is
not close to the correct answer;
– Calculations with zero are easier than calculations with four.
We prefer the first possibility as it is more in keeping with his reasoning at other
times. If he has reasoned in this way (or in some similar way) then his reasoning
process returns to deducing after testing, to further limit the possible values of x.
The cycle then begins again. This deducing is followed by conjecturing and testing:
“Zero doesn’t work there” (line 9). He tantalisingly continues “because” but does not
give his reason. He could easily have calculated mentally 11+18=29 and so
concluded that zero is incorrect. In so doing he is likely to have observed that 29 is
almost the correct value of 27.
This provides the basis for another deduction, that 0 is close to the correct value for
x, leading to another conjecture, x = 1, which turns out to be correct (line 11). We will
see later that there is some reason to believe that his final test also leads to deducing.
When Sandy is asked to explain how he found the answer so quickly, he replies
“I knew this number [x] if you kept on guessing then that would ... make these two
numbers [y and z] and you just had to make one that added up to twenty seven”
(line 13). Here he clearly articulates that he understands that the value of x
determines the values of the other two unknowns, but he uses this to constrain his
guessing, not to find the answer directly.
When he explains how he knew x was a fairly low number (line 17 “because that’s
almost these two together”) he seems to refer to knowledge he now has (that 11+18 is
almost 27) but which probably he did not have when he first said “You need a fairly
low one here” in line 4. If he was aware that 11+18 is not 27, then why did he guess
zero later (in line 9)? This illustrates an important point about Sandy’s reasoning:
Sandy can be articulate about his thought process, but the reasoning he describes may
not be the reasoning that actually led him to the conclusion in question. In this case we
believe his reasoning was something like our reconstruction above, which does not
depend on knowing that 11+18 is not 27. However, Sandy was probably not aware of
his reasoning; it is “unformulated” (see Part 2, Introduction). Throughout his reaso-
ning is motivated by a need to explore, to discover what the answer to the puzzle is.
Sandy’s pattern of reasoning can also be modelled linearly:
Ded.→Conjecture: 5→Test↵
[Ded.]
However, it is more revealing to model it cyclically as we have in Figure 28.
197

via free access
CHAPTER 11
PROOF ANALYSIS
The testing that Sandy did in Episode 1 led him to reject his first two conjectures,
but that did not affect the validity of the deducing he had done, as his deducing
only provided limits to the possible values of x. The next pattern of reasoning we
will describe involves a counterexample to a general statement, which we believe
Sandy arrived at through partly or wholly deductive reasoning.
In Episode 2, Sandy’s approach to a new puzzle is quite different from his
approach in Episode 1. He immediately calculates a value for one of the missing
numbers by adding two of the known numbers, subtracting the third and dividing
the result by 2. Why does he do this? Two possibilities seem likely.
We believe that when he found that x=1 in the first puzzle, he noticed that
increasing x from zero to one had the effect of decreasing c from 29 to 27. In other
words the change in c is double the change in x. Having observed this interesting
fact he reasoned abductively that if this were a general rule it would explain this
interesting special case.
However, it is also possible that he reasoned entirely deductively at this point.
He had worked with the relation x+y=a on several occasions when testing supposed
values of x. It is easy to deduce from this relation that if the value of x increases by
one, then the value of y decreases by one. Similarly, if the value of x increases by
one, then the value of z decreases by one. So, if the value of x increases by one, then
the value of y+z decreases by two, which means the value of c decreases by two.
Whichever way he reasoned, he could conclude that to change c by some
amount one should change x by half of that amount. This leads to his procedure.
When Sandy applied his general procedure for solving Arithmagon puzzles to a
specific case in Episode 2, it did not work. We call his pattern of reasoning in
reaction to this “Proof Analysis” as it is analogous to Lakatos’s (1976).
Episode 2: Sandy Makes Up One of his Own.

18 Tom Ok so why don’t you make up another one
19 Sandy Ok — —
20 Tom Put your own ones on the side
21 Sandy Thirteen — — — — — — — — — Now add these up thirteen
[times???] — — gives thirty four
22 — — which is [higher than ???] forty two — given — —
23 eight — and eight — which would give me four
24 — [just think ???] — — hmm — — — that’s eight
25 David hmm
26 Sandy — ummm — — — [I see ???] — — — [long this ???]
27 David So why did you guess four?
28 Sandy Um, I don’t know. I was just trying something
29 David OK
30 Sandy — — — — seventeen — nine [Wrote “9” and “17” as if he was
going to add them but didn’t write the answer] — [???] twenty two,
forty two —
198

via free access
Figure 33. Sandy’s puzzle.
31 Maybe it’s two one — [Wrote 2 at the top and then wrote 19 and 11
at the other corners.] — — —
32 Are there only certain combinations that work on this?
33 Tom You mean on- of numbers here?
34 Sandy our numbers here
35 Tom we don’t know [laughter] is the- is the- answer
36 Tom It should, it should work that one
37 David all the ones we’ve ever tried have worked
38 Tom Yeah
39 Sandy — — [square the line ???] — — — — — — thirteen and four would
be nine, twenty one, and then that — — that number’s too high —
and those ones are pretty high — — — — don’t- no you can’t do it
40 unless you put minus four on this — — — — —
41 Tom ah
42 David hmm
43 Sandy — [erasing?] [???] — — fourteen — — — — twenty five, That’s it.
— Seventeen, Twenty five — — — Forty two [Calculated 17+25 on
paper and wrote “-4” “17” and “25” at the corners of the triangle]
Discussion of Episode 2
In lines 21–23 Sandy applies his method to his 13-21-42 puzzle. He first found the
value of c if x is zero (in other words he adds 13 and 21). He then finds the
difference between this value and the desired value of c (he subtracts 34 from 42).
This is the amount by which he wants to adjust c, so he must adjust x by half this
amount, so x must be four. So far the pattern of this reasoning consists of deducing
and conjecturing (his general procedure), and specialising (to apply it to his puzzle).
Unfortunately, x is not four. His initial reaction to this counterexample (in lines
24 to 31) leads us to suspect he arrived at his procedure via abductive reasoning
rather than deductive reasoning, and that it has the epistemic value of a conjecture
rather than that of a generalisation (see Chapter 6).
When he found that four does not work, he checked his arithmetic (adding 17 and
9 again in line 30) and then returned to his pattern of making conjectures (2, and
possibly 1, in line 31), testing them and deducing from the result. Recall that the
199

via free access
CHAPTER 11
result of deductive reasoning is more certain than the result of abductive reasoning,
so if Sandy had arrived at his procedure deductively, he might have been more
reluctant to give up on it and return to testing conjectures. However, his lack of
confidence in it might also be related to the unformulated nature of his reasoning.
Whatever reasoning Sandy used to arrive at his procedure he does not explain why
he started with four when asked (in line 27).
Sandy’s cycle of deducing, conjecturing and testing tells him that 2 is too large,
and he may also have tested one and zero mentally and come to the same
conclusion. He next questions whether the puzzle might be impossible (in lines 32
and 40). This makes it clear that he had assumed that x must be positive. He can
revive his confidence in his general procedure if this puzzle is in fact impossible. In
that case it changes from being a counterexample to being an exception. This is
what Lakatos calls “exception barring”: defining a restricted domain in which the
conjecture holds and which excludes the known counterexamples (1976, p. 26).
Sandy keeps his procedure and the deductions supporting it unchanged, but notes
that some puzzles are impossible. The counterexample he encountered does not
invalidate his procedure because it is an exception to the rule.
David and Tom strongly suggest (in lines 36 and 37) that exceptions should not
exist, however, and Sandy now looks for the cause of the problem in his (as yet
unstated) assumptions. He becomes aware of his assumption that x must be positive
and now that he is aware of it he can explore the effects of removing it (line 40).
Formally, one might say he has revised his procedure from x = |(a+b) – c| ÷ 2 to
x = ((a+b) – c) ÷ 2 by removing the constraint that x must be positive.
We call this pattern of reasoning “Proof Analysis” because it includes this step
of locating a faulty assumption in the reasoning and revising the conclusion
accordingly. Lakatos (1976) describes an analogous process in his theoretical recons-
truction of the collective reasoning process of mathematicians over an extended
period of time. Lakatos proposes that mathematical activity involves a process of
proofs and refutations that he calls “proof-analysis.”
There is a simple pattern of mathematical discovery — or of the growth of
informal mathematical theories. It consists of the following stages:
(1) Primitive conjecture.
(2) Proof (a rough thought-experiment or argument, decomposing the primi-
tive conjecture into sub-conjectures or lemmas).
(3) ‘Global’ counterexamples (counterexamples to the primitive conjecture)
emerge.
(4) Proof re-examined: the ‘guilty lemma’ to which the global counterexample
is a ‘local’ counterexample is spotted. This guilty lemma may have pre-
viously remained ‘hidden’ or may have been misidentified. Now it is
made explicit, and built into the primitive conjecture as a condition. The
theorem — the improved conjecture — supersedes the primitive conjecture
with the new proof-generated concepts as its paramount new feature.
These four stages constitute the essential kernel of proof analysis. (pp. 127–128)
200

via free access
Deducing
↗ ↘
Conjecturing → Testing
↗
Exception
Deducing → Conjecturing → Specialising → CE* →
Barring
↘
Lemma
incorporation
→ Generalising
*Counterexample
Figure 34. Sandy’s reasoning, including proof analysis.
Here Sandy seems to have followed an analogous process (see Figure 34). One
difference however, is that in Lakatos’s proof-analysis the process begins with a
conjecture and we believe that Sandy arrived at his procedure through partly or
wholly deductive reasoning. So in his reasoning the “proof ’ came before the conjec-
ture, although we doubt that Sandy was aware of these deductions.
Lakatos’s third stage is encountering a counterexample, which Sandy did
when he found that four is not the answer to his puzzle. After attempting to bar the
counterexample as an exception, Sandy turns to examining the reasoning leading to
the conjecture and identifying the “guilty lemma” which accounts for the error.
This is what Sandy does in line 39. The guilty lemma here is the assumption that x,
y and z are all positive. Once this assumption is removed, his method immediately
produces the correct answer.
It is worth recalling that Lakatos sees proof-analysis as a cycle, that the final act
of generalising which we have shown as the end of Sandy’s pattern of reasoning,
could lead to further specialising, counterexamples, and improving of the conjecture.
The pattern of reasoning in Sandy’s solution to his puzzle extends the pattern in
his solution to the initial puzzle. However, the two patterns need not be linked. One
could have a Deduce-Conjecture-Test cycle without it leading into Proof Analysis,
and one could have Proof Analysis that began with a conjecture emerging from
another pattern of reasoning.
SCIENTIFIC VERIFICATION
Reid (2002a) describes three other examples of patterns of reasoning which we call
Scientific Verification, Surrender, and Exception and Monster Barring. Scientific
Verification is similar to what Polya (1968) calls “verification of a consequence”
(vol. 2, p. 3). It consists of five elements: the observation of a pattern, conjecturing
that the pattern applies generally, testing the conjecture, generalising the conjecture,
and finally using the generalisation as the basis for simple deductions about other
aspects of the situation (See Figure 35).
This pattern is illustrated by the transcript in Chapter 6 showing inductive reasoning
in students’ mathematical activity. Recall that in that transcript Will observes a pattern
in his solution to the 4 by 4 Count the Squares problem (see Figure 7 in Chapter 6).
201

via free access
CHAPTER 11
Pattern Observing → Conjecturing → Testing → Generalising → Deducing
Figure 35. Scientific Verification.
He observes that the quantities of some sizes of squares are related to the sizes of
other squares in a systematic way. He then conjectures that a similar pattern will
occur with other Count the Squares problems. He tests his conjecture by specialising
it to the 5 by 5 case and comparing the quantities predicted by his conjecture to
those he counts in the grid. His test confirms his conjecture and he generalises
“The chances are if it works for those then it works for the rest of them.” Will then
continues:
14 Will So because of this if you find out only half of it you can calculate the
rest and just add them up...Or actually I realized something. You don’t
have to know the actual amounts... If you only know the sizes if you
continue this pattern then — [he paused to think] — Wait! You don’t
even have to know the number of squares, as long as you know the
sizes ... because this is equal to this...
Here he uses his generalisation to deduce two things: “So because of this if you
find out only half of it you can calculate the rest and just add them up” and “You
don’t even have to know the number of squares, as long as you know the sizes. …
because this is equal to this”.
Scientific Verification differs from the deduce-conjecture-test cycle and Proof
Analysis in that it begins with pattern observing rather than deducing. Like those
patterns it includes testing, but unlike them the testing must confirm the conjecture.
When the testing results in a counterexample to the conjecture different patterns
arise, including Surrender, and Exception and Monster Barring.
SURRENDER
One response to a counterexample is to reject the conjecture entirely and to replace

it with its negation. This is analogous to what Lakatos (1976, p. 13) calls “Surrender”
when discussing mathematicians’ historical reactions to counterexamples and so
we have adopted the same term to refer to this pattern of reasoning. This pattern
of reasoning is diagrammed in Figure 36. (We have modified Reid’s, 2002a, termi-
nology slightly.)
Reid provides two examples of this pattern from the mathematical activity of
students attempting to solve the Count the Squares problem. In one (Reid’s Case 2)
the motivation for the reasoning is to find a counterexample in order to refute it. In
other words, it fits the general structure of a proof by contradiction, although the
proof is almost entirely unformulated.
Pattern Observing → Conjecturing → Testing → Counterexample → Negation
Figure 36. Surrender.
202

via free access
We will include the transcript for Reid’s second illustration here as it includes an
explicit conjecture and a situation where the intent is not to reject the conjecture
from the outset. The episode comes from the mathematical activity of Walt, Ryan,
Mona, and Sue as a group of four. They are discussing the 5 by 5 Count the Squares
problem. (See Chapter 6 and Zack, 1997, 1998, 1999ab for other episodes and inter-
pretations involving the same children.)
1 Walt Can we just try really quickly a 6 by 6 because I mean maybe it’s
coincidence but a 4 by 4 it’s a multiple of five
[Sue says she has done the 6 by 6, but then realizes she has not.]
2 Walt Because, cause, I’m thinking because a 4 by 4 it’s a multiple of five,
30. the 5 by 5, it’s a multiple of five. 55.
3 So are they all multiples of five? Just going up somehow? Like by 25.
4 So the next one might be — I know, I know I’m wild but uh — the
next one might be 80. I’m guessing. I think the 6 by 6…my hypo-
thesis is 80.
[After a brief discussion of the meaning of “hypothesis” which is a
new word for most of them, they count the squares for the 6 by 6
case.]
5 Walt So that’s, 91?
6 Mona We must have a problem somewhere-
7 Walt Yeah
8 Mona because it has to be a multiple of five.
9 I am almost positive it’s a multiple of five.
10 Walt Yeah
11 Sue Maybe it’s just a coincidence that the two we did are multiples of
five, maybe they’re not all.
12 Walt No. I don’t think so
13 Mona Neither do I. I’ve got a feeling.
14 Walt 4 by 4s Let’s try that 4 by 4 one more time, because there might be
10. —
15 See if there are 10. No that would be just 92. So we went wrong
somewhere.
[They count the squares in the 6 by 6 case again, and get 91 again.]
16 Walt So either my hypothesis is wrong or we’ve made a mistake some-
where along the way.
[They count the squares again.]
17 Walt So if they’re not all — there’s gotta be some coincidence in this
18 Sue Maybe your hip- guess …
19 Walt Maybe it’s wrong, but — is there any pattern between 30 55 91?
20 Mona I’m sure it has to be 90. There’s something wrong
21 Walt 90 or 95
[They count the squares again].
22 Walt So if it is 91— …
23 Sue I think your hypothesis … your guess, is wrong
203

via free access
CHAPTER 11
As with the Scientific Verification pattern, the reasoning here begins with pattern
observing (“a 4 by 4 it’s a multiple of five, 30. The 5 by 5, it’s a multiple of five,”
line 2) leading to a conjecture (“So are they all multiples of five?” line 3), motivated
by a need to explore. Walt’s expressing himself in a question suggests this general
rule has the epistemic value of a conjecture, not a generalisation. Out of a need to
verify, the conjecture was then tested, which involved specialising from the
conjecture (“the next one might be 80,” line 4), and comparing with an additional
case. This sequence, from observing a pattern to testing, went very quickly in
Walt’s mind as the first thing he says about it is to propose testing it in the 6 by 6
case (line 1). He then articulates his thinking for the others (lines 2–4).
Their counting of the squares in the 6 by 6 case brought them to a counter-
example (“91?”, line 5). The students’ reactions indicate that the conjecture had a
different epistemic value for each of them. Walt’s questioning tone suggests that he
had expected it to be confirmed. Mona feels even more strongly. She says explicitly
“it has to be a multiple of five” showing that she is ready to generalise (line 9).
Sue on the other hand seems ready to accept the counterexample as negating the
conjecture. They are not all multiples of five “maybe they’re not all” (line 11). Her
reasoning has already progressed through the entire Surrender pattern.
Walt and Mona, however, are not yet prepared to surrender, and they check the
6 by 6 case again, and again get 91. Walt sums up the situation “So either my
hypothesis is wrong or we’ve made a mistake somewhere along the way” (line 16)
He is succinctly describing two possibilities created whenever a counterexample is
discovered, that the conjecture is wrong or there is an error in the specific case which
is observed. This is a point that is sometimes missed in philosophical discussions
of reasoning in mathematics and science. Lakatos, for example, does not mention
the possibility of mathematical counterexamples arising from faulty observations
or calculations, and this is one of Chalmers’ (1982) criticisms of Popper’s falsifica-
tionist philosophy of science.
After the third counting of the squares in the 6 by 6 grid again results in 91, Walt’s
position begins to change. He admits the possibility “if they’re not all” (line 17),
voicing the negation of his hypothesis. For Mona, on the other hand, Walt’s “hypo-
thesis” is still a generalisation. She repeats her conclusion that there must be an
error in their counting (lines 6, 20). The three identical results of counting the 6 by
6 squares have not convinced her that the answer is 91. After one more counting of
the squares again results in 91, Sue’s position “your guess, is wrong” (line 23) seems
to be accepted by the group, bringing them all to negating the conjecture, based on
the counterexample. All their reasoning since Walt first observed his pattern up to
this point has been motivated by a need to verify the conjecture and the number of
squares in the 6 by 6 grid.
EXCEPTION AND MONSTER BARRING

Walt, Mona, Sue and Ryan continued their explorations and their subsequent mathe-
matical activity illustrates two other possible reactions to counterexamples: treating
them as exceptions or as “monsters”. We have already seen an example of Exception
Barring in the mathematical activity of Sandy, the Exception Barring pattern in the
204

via free access
Gen1→Ded1
↗
PO1→Conj→Test→CE1 Gen2→Ded2
↘ ↗
Neg →PO2 →CE2 →MB
↘
CE3 →PO3 →EB→Gen3→Sp
PO1: Pattern Observing – “it’s a multiple of five” – line 2
Conj: Conjecturing – “are they all multiples of five?” – line 3
Test: Testing – “Can we just try really quickly a 6 by 6” – line 1
CE1: Counterexample – “91” – lines 5, 16, 22
Gen1: Generalising – “it has to be a multiple of five” – lines 8, 9, 20
Ded1: Deducing – “We must have a problem somewhere” – lines 6, 15, 20
Neg: Negation - “I think your hypothesis … your guess, is wrong” – lines 11, 23
PO2: Pattern Observing – “that is a multiple of five” – lines 27, 32
Gen2: Generalising – it has to be a multiple of five (implicit) – line 28
Ded2: Deducing – “so we must have been wrong for the 6 by 6” – line 28
CE2: Counterexample – “not one” – line 33
MB: Monster Barring – “1 is ... your starting number” – line 38
CE3: Counterexample – there are 14 squares in a 3 by 3 grid (implicit) – – about line 40
PO3: Pattern Observing – The 6 by 6 and the 3 by 3 are not multiples of 5 (implicit) – about
line 40
EB: Exception Barring – “Multiples of three” – line 42
Gen3: Generalising – “Multiples of three won’t be a multiple of five” – line 42
Sp: Specialisation – “so a 9 by 9 won’t be a multiple of five,” – line 47
Figure 37. Exception and Monster Barring.
mathematical activity of Sandy, above. The related reaction, monster barring, was
also observed by Lakatos in the historical record of mathematics. It involves declaring
the counterexample to be inadmissible in some way without allowing that the
conjecture might have exceptions.
The Monster Barring and Exception Barring patterns that occur in the following
episode relate to different conjectures and counterexample and the reasoning builds
on what came before. Hence our diagram (Figure 37) shows not only the pattern
leading to the Monster Barring and Exception Barring but also the Surrender
pattern that occurred earlier.
The transcript resumes as they calculate the answer for the 7 by 7 problem.
24 Mona I got 140

25 Ryan Which is a multiple of five
26 Walt I got 140,
27 W&M and that is a multiple of five,
28 Mona so we must have been wrong for the 6 by 6
[They discuss how tired they feel, and seem uncertain what to do
next.]
205

via free access
CHAPTER 11
30 Walt Do you want to do a 1 by 1? That’s easy, there’s one, 1 by 1 …

31 Mona What if we had a 2 by 2? …
32 Walt There’s only five, and that’s a multiple of five! … and one is a
multiple of five!
33 Mona You’re right. That’s so freaky. No it’s not one.
34 Walt One, yeah
35 Mona No
36 Walt But you can go one, five, When you count by fives you go one, five, …
37 Mona No, you go zero, five, ten …
38 Walt You have to have 1 because 1 is, you know, your starting number
39 Mona maybe it’s like even or odd numbers that are multiples of five
40 Walt And 6 is, yeah, but 4 by 4 is a multiple of five still
[Walt speculates about multiples of four, while Mona tries to show
something to Ryan. She is interrupted by a question. ]
41 VZ … [to Mona] Didn’t you do a 3 by 3?
42 Walt Multiples of three won’t be a multiple of five —
43 Mona What?
44 VZ What was the answer for the 3 by 3?
45 Mona [works out, with Ryan, the answer for the 3 by 3] 14
46 VZ So the answer is fourteen, is that a multiple of five?
46 Walt No, so everything that’s a multiple of three, so a 9 by 9 won’t be a
multiple of five
Discussion
Three related patterns of reasoning can be seen here. The first is Scientific Verifi-
cation, when Mona returns to her generalising of Walt’s hypothesis (Gen2, line 28)
as a result of the pattern observing at line 27.
The second is Monster Barring. Their further pattern observing (line 32) leads to
another counterexample: there is only one square in a 1 by 1 grid. One is not a
multiple of five (as Walt initially claims in line 32) but he gives it a special status
which means it does not count as a counterexample. Monster Barring involves
declaring the counterexample to be inadmissible in some way (Lakatos, 1976, p. 23).
Walt is asserting that one is not a typical number to which the usual meaning of
“multiple of five” would apply. Instead it is the “starting number” that “you have to
have” (line 38).
The third pattern begins with further pattern observing, in this case involving a
pattern in the counterexamples (other than the 1 by 1 case that has already been
dealt with by Monster Barring). This pattern allows for Exception Barring. Walt
revises his hypothesis by describing a set of exceptions to which it does not apply.
“Multiples of three won’t be a multiple of five” (line 42). He generalises this
revised hypothesis and then uses it to make a prediction by specialising “so a 9 by 9
won’t be a multiple of five” (line 47). Note that this is stated very differently from
his initial hypothesis “So are they all multiples of five?” (line 3). The reasoning
process has transformed his “hypothesis” from a conjecture into a generalisation,
206

via free access
from which he is confident to specialise other values. In the remainder of his work
with Count the Squares problems, Walt used this generalisation as a way of checking
calculations when counting the number of squares in larger grids.
Three different needs are addressed through these three different patterns. Mona’s
Scientific Verification is motivated by a need to verify Walt’s conjecture. Each
additional multiple of five they observe adds to her confidence that the counter-
example they have discovered is an error. Walt’s Monster Barring is motivated by a
desire to explain why the value one is not a counterexample. His Exception Barring,
however, is motivated by the need to explore further to arrive at a restricted
conjecture that excludes all the known counter examples as exceptions.
SUMMARY
In this chapter we have described five patterns of reasoning that have been
observed in empirical work:
– Deduce-Conjecture-Test cycle
– Proof Analysis
– Scientific Verification
– Surrender
– Exception and Monster Barring
These types share a number of characteristics, but also differ in significant ways.
The starting point of the pattern is one notable difference: the first two patterns
start with deducing while the last three start with observing a pattern.
Testing, that is predicting a value based on a conjecture or a deduction, is one
common feature. However, the results of testing and the reactions to those results
differ. In the Deduce-Conjecture-Test cycle a value is predicted on the basis of a
deduction and if it fails the testing then the way in which it fails becomes the basis
for further deducing. When the value is correct information from the testing also
informs further deducing, of a general procedure. In Proof Analysis testing results
in a counterexample to a conjecture that arose through deducing, and that counter-
example leads to a revisiting of the hidden assumptions made while deducing,
revealing one that accounts for the counterexample. The conjecture is then revised
accordingly. In Scientific Verification the testing confirms a conjecture that is then
generalised; that is, its epistemic value changes from a plausible conjecture to an
accepted general rule. In Surrender the testing refutes a conjecture that is then
negated; that is, its epistemic value changes from a plausible conjecture to a known
falsehood. Finally, in Exception and Monster Barring the testing results in a
counterexample, but the counterexample itself is rejected either by considering it a
special case (Monster Barring) or by modifying the conjecture to exclude a class of
cases including the counterexample (Exception Barring).
The needs motivating the reasoning in these patterns also differ. In the Deduce-
Conjecture-Test cycle and in Exception Barring the need is to explore. In Monster
Barring the need is to explain. In Scientific Verification and Surrender the need is
to verify. And in Proof Analysis the initial need is to verify, but this changes to a
need to explore when the counterexample emerges.
207

via free access
CHAPTER 11
Finally, a significant difference between these patterns of reasoning is the degree

to which they might be unique to mathematics. Reid (2002a) comments that the
patterns he describes (Scientific Verification, Surrender and Exception and Monster
Barring) are only partially mathematical, because they lack the expectation for
deductive reasoning that can been seen in Sandy’s patterns of reasoning (the Deduce-
Conjecture-Test cycle and Proof Analysis). It is clear that Scientific Verification and
Surrender are not patterns unique to mathematical reasoning as they form the core of
scientific reasoning in general. On the other hand Proof Analysis requires a deductive
basis for conjectures and that makes it more clearly a mathematical pattern. This
suggests that in teaching proof it is important to include contexts in which the
Proof Analysis pattern is likely to occur.
IDEAS FOR RESEARCH
The patterns described above are hardly an exhaustive catalogue of what can occur
in students’ mathematical activity. Additional patterns will no doubt be identified in
future research focussed on the description of such patterns. Such research will have
to address the interrelation between needs and reasoning, and also the formulation
of reasoning and the formality of the proofs produced, to provide a more complete
picture of the patterns of reasoning described. Also, further analysis of what makes
a pattern of reasoning “mathematical” or “scientific” will require additional empirical
research and theoretical developments.
Finally, for those interested in proof, an important question to consider is what
patterns of reasoning lead to proving. Among those offered here only Sandy’s
employs deductive reasoning in a significant way, and it still lacks the formalisa-
tion needed to produce a proof.
208

via free access
PART 4
CONCLUSIONS
The two chapters in this part draw conclusions from the research described in previous
chapters. Chapter 12 focusses on implications for teaching, and Chapter 13 on future
research directions.
All of the research on proof and proving has implications for teaching, however
those implication are not always direct or obvious. In Chapter 12 we discuss possi-
bilities for teaching proof, keeping in mind that it is not possible, given the present
state of research, to be prescriptive about teaching. Nevertheless, there are themes
that emerge. First we consider how different researcher perspectives and meanings
for proof suggest different foci for teaching. Then we reflect on the problem of
determining what must be included and what can be omitted in preformal and
semi-formal proofs. We discuss two concepts that offer ways of thinking about the
necessary informality of students’ proofs. Starting from an assumption that teaching
must begin with students’ prior understandings and that changing teaching can only
begin from present practices, we then consider the implications research has for
teaching and changing teaching. Finally, we revisit the teaching experiments described
in Chapter 9 and their implications.
The large amount of research that has been done on proof in recent years provides
some implications for teaching, but it is also clear that there is still much to be learned.
Chapter 13 summarises interesting research questions arising from the research.
Questions related to teaching proof are considered first, followed by questions related
to learning to prove. Conceptual questions are outlined in the final section.
209

via free access
via free access
CHAPTER 12
IMPLICATIONS FOR TEACHING
This book has focussed on research, all of which has some implication for teaching,
but it is not always clear what exactly that implication is. And implications are a
long way from prescriptions. Though a large amount of research has been done on
proof in recent years, there is still much to be learned, and so no definitive
statements can yet be made on how proof should be taught. With this in mind, this
chapter will discuss possibilities for teaching proof, founded on research but
influenced also by our own experiences as teachers and researchers.
Some implications come primarily from theoretical considerations and we
will discuss these first. They include the obvious point that how you teach proof
depends on what you mean by proof, and clearly the differences identified in
Chapters 2, 3 and 7 have significant implications for teaching. Likewise, how you
teach proof depends on what you think proofs are for, and so the discussion in
Chapter 5 of the role of proof will be referred to in considering possibilities for
teaching proof.
Another theoretical point with implications for teaching is that of formality. In
Chapter 1 we noted that the proofs of professional mathematicians are often only
semi-formal; that is, they leave out steps and make use of unstated assumptions. In
Chapters 3 and 7 we discussed the work of the preformalists who claim that teaching
should include proofs even less formal than those accepted by mathematicians. Here
we will discuss some implications for teaching, and two ideas related to formality,
the “tool-box” and “local organisation”.
We then turn to implications based primarily on empirical research. In Chapter 4
we reviewed findings from empirical research on proof and proving, much of
which supported a general conclusion that students cannot write correct proofs and
do not understand what a correct proof is. Here we will discuss this conclusion in
the context of the approaches that have been used since the 1960s to teach proof.
Much of the work we included in Chapters 6, 8, 10 and 11 involves describing the
nature of students’ reasoning and argumentation in existing classroom contexts. In
this chapter we will consider this as a starting point for teaching. Finally, the
teaching experiments described in Chapter 9 share some common features that are
suggestive, which we will review here.
WHAT IS PROOF AND WHAT IS IT FOR?
How you teach proof depends on what you mean by “proof ” and what you think
proofs are for. The diversity of usages of the word “proof ” described in Chapter 2
leads to a considerable variety of suggestions for ways of teaching proof. This might
211

via free access
CHAPTER 12
suggest that there are many ways of teaching proof, but it would be more
accurate to say there are many things taught under the heading of “teaching
proof”.
One usage of “proof ” is to mean “a convincing argument” but as we noted in
Chapter 2, the audience to be convinced can vary. The focus might be on helping
students to produce arguments that are convincing, and specifically arguments that
are convincing to mathematicians. Herbst (2002b) describes this as the focus of
American mathematics education in what he calls “the era of Originals”. Of course,
one might be more concerned that students be convinced themselves by the proofs
they read, rather than able to produce proofs convincing to others. In that case, like
Fischbein (1982), you might feel that “teaching proof ” involves developing an
“intuition” that “will enable the pupil not only to understand a formal proof but
also to believe (fully, sympathetically, intuitively) in the a priori universality of the
theorem guaranteed by the respective proof ” (p. 17).
If “teaching proof ” focuses on “proof ” as a proof-text then it might resemble
what Herbst (2002b) describes in “the era of Texts” when students were expected
to read and reproduce proofs of significant results in mathematics. A more recent
variant of this idea uses proof-texts to help students understand mathematical ideas.
Movshovits-Hadar (1988ab) advocates presenting theorems in the most surprising
way possible to create a need for an explanation, and then offering proof-texts that
explain the surprising result.
If you see “teaching proof” as referring chiefly to teaching students to produce
proof-texts involving only deductive reasoning, then you might adopt teaching
practices similar to those advocated by Duval and Egret (1989, 1993; Duval 1991;
Egret & Duval, 1989). They propose having students come to a conjecture in a
“heuristic phase” which also includes identifying the key ideas in their conjecture.
This is followed by a distinct phase of “deductive organisation” in which the students
develop a graphical representation of their conjecture, breaking apart its antecedents
and consequents and citing theorems that connect them. Once this graphical represen-
tation is complete the students use it to produce a proof-text.
On the other hand, if you see the production of proof-texts as being dependent
on prior (often non-deductive) reasoning in the course of developing a conjecture
(as is suggested by the idea of cognitive unity) then your teaching approach might
involve much more extensive exploration of conjectures. Multiple conjectures are
made, and refined through class discussion. The impossibility of verifying general
statements empirically is discussed and then students attempt to produce proof-texts to
verify their conjectures, supported by group and class discussions. (For examples,
see Boero & Garuti, 1994; Boero, Chiapini, Garuti & Sibilla, 1995; Boero, Garuti &
Mariotti, 1996; Boero, Garuti, Lemut et al., 1996).
FORMALITY
As we noted in Chapters 1 and 2, the proofs of professional mathematicians are not

formal in the sense of the formalists. Instead they are semi-formal: steps and justifi-
cations are left out, and the starting point is rarely a set of axioms and more often a
212

via free access
set of results generally accepted by the community. And in Chapters 3 and 7 we

discussed the work of the preformalists who advocate including proofs less than
semi-formal in teaching. This raises a problem for teachers and students for whom
it is difficult to know what steps are required and what can be omitted, and what
assumptions one can make without justification. In this section we will look at this
problem through two ideas: “the tool-box” and “local organisation”.
The Tool-Box
Semi-formal and preformal proofs make reference to, or assume without stating,
other theorems or assumptions. They are taken for granted. Even in the work of
professional mathematicians there are theorems that are used without their proofs
being read, and even without any source being known.
Within any field, there are certain theorems and certain techniques that are
generally known and generally accepted. When you write a paper, you refer
to these without proof. You look at other papers in the field, and you see
what facts they quote without proof, and what they cite in their bibliography.
You learn from other people some idea of the proofs. Then you’re free to
quote the same theorem and cite the same citations. You don’t necessarily
have to read the full papers or books that are in your bibliography. Many of
the things that are generally known are things for which there may be no
known written source. As long as people in the field are comfortable that
the idea works, it doesn’t need to have a formal written source. (Thurston,
1995, p. 33)
Netz (1999) discusses the omission of references to theorems and assumptions in
classical Greek proofs. He calls the set of theorems and assumptions that can be
used without comment the “tool-box”. For example, Archimedes can assert that
two segments are of the same length because they are radii of the same circle. He
does not have to even make any reference to this justification. He simply states that
they are the same length and leaves it to the reader to figure out why (Netz, 1999,
p. 172). As his readers were all members of a cultural community for whom the
same tool-box was taken for granted, this was acceptable.
Netz argues that the results proven in Euclid’s Elements constitute most of
the classical Greek tool-box. One might imagine that the study of the Elements was
undertaken as the beginning of a scholar’s mathematical education and so thereafter
mathematicians could assume that anything in Euclid could be used without
comment or reference. However, Netz suggests that the contents of the tool-box
could become known by noticing what was omitted in arguments:
The very fact that an argument was made, without any intuitive or diagram-
matic support for that argument, must have signalled for the audience that the
argument was sanctioned by the Elements. Once this is the expectation, the
need to refer explicitly to the Elements declines, which would in turn support
the same tendency. (p. 232)
213

via free access
CHAPTER 12
In other words, the process might resemble that described by Thurston, above; the
reader comes to know, by observing what things are taken for granted in proofs,
what is in the tool-box, without ever being told explicitly. And the more this
happens the larger the tool-box grows, allowing more references to be omitted.
Netz’ concept of the tool-box is an important one for teaching proof. Proofs in
school mathematics are also semi-formal (at best) even when structures like two
column proofs attempt to impose a rule that every statement must be justified. One
approach sometimes taken is to inform the students that they must forget everything
they already know and start from the axioms and definitions given in the textbook.
But experience tells us that the students do not forget everything they know, and
the axioms and definitions given in the textbook are rarely (never) complete
themselves. The result is that the class pretends to base its arguments on the given
axioms and definitions, while being guided by their prior knowledge and adding to
the tool-box through observing what is omitted in the teacher’s and textbook’s
proofs. In this case the toolbox is supposed to be limited to what the textbook allows,
but is actually larger.
Another approach is for a teacher to start presenting proofs without establishing
what is in the tool-box, so that, as Netz says happened for the Greeks, the contents
of the tool-box become known to the students through the making of arguments
without stating their justifications. If a proof is based on an assumption that the
measures of angles can be added, but this is never stated or justified, then it must
be part of the tool-box.
The idea of the tool-box applies to semi-formal proofs in the context of a coherent
system of interrelated theorems, as in a traditional high school geometry course.
However, as the preformalists note, it is possible to explore proofs earlier and in less
structured contexts. In such contexts the idea of local organisation become useful.
Local Organisation
Freudenthal (1971, 1973) claims that proving must begin with what he calls “local
organisation” as opposed to the “global organisation” of an axiomatic system.
In a globally organised system the definition of parallelogram would be part of
the tool-box and would either be explicitly taught or would become known through
its use in proofs. Freudenthal describes another approach:
If the child knows what a rhomb or a parallelogramme is, it may visually
discover properties of these shapes. There are a lot of them, and in the class
discussion children will recite them. [Here Freudenthal lists some properties.]
There are a host of visual properties which ask for organization. Here starts
deductivity; rather than being imposed it unfolds from local germs. The
properties of the parallelogramme become deductively interrelated, and finally
one emerges (maybe different ones for different children) from which the
others can be deducted. This property can be taken as a definition, and now
it can be understood why a square and a rhomb should be considered as
parallelogrammes. (1971, p. 424)
214

via free access
So rather than a definition being given at the outset, the students’ proving determines
which property is a definition and which ones are consequences of it. There is a
local organisation of mathematical knowledge, but no global system into which
that knowledge fits. This is necessary, Freudenthal states, because, “a student who
never exercized organizing a subject matter on local levels will not succeed on the
global one” (p. 426). It is at the local level that proving and defining are learned,
before being used (perhaps much later) to define and prove in an axiomatic system.
Freudenthal’s implications for teaching are clear:
In general, what we do if we create and if we apply mathematics, is an activity
of local organization. Beginners in mathematics cannot do even more than that.
Every teacher knows that most students can produce and understand only short
deduction chains. They cannot grasp long proofs as a whole, and still less can
they view [a] substantial part of mathematics as a deductive system. (p. 431)
Local organisation and use of an implicit tool-box raise the question for teaching of
what must be made explicit, what requires proof, and what can be left implicit.
Jahnke (1978, p. 213) raises this point and suggests that Freudenthal’s examples
offer a resolution to the “problem of the missing axioms”.
Freudenthal [1973] behandelt das Problem der fehlenden Axiomatik allgemein
unter Hinweis auf die Tätigkeit des Schülers als dem obersten regulierenden
Prinzip. Und in der Tat kann man sagen, dass es beim Problem der fehlenden
Axiomatik um die Herstellung einer angemessenen Balance von Begründung
und Anwendung in der Schülertätigkeit geht. [Freudenthal addresses the
problem of the missing axioms generally by referring to the activity of the
student as the supreme regulative principle. And in fact, one can say that
the problem of the missing axioms is the establishment of a reasonable balance
between justification and application in the students’ activity.] (p. 213)
In other words the decision as to what to leave implicit, what to make explicit and
what to prove must be made in order to develop students’ understanding of mathe-
matical concepts and ability to apply them. In cases where the proof brings a new
understanding of the concepts involved, the proof is useful. Assumptions that have
unexpected implications will need to be made explicit in exploring those implications.
Assumptions that would seem obvious and trivial to the students if made explicit
can safely be left implicit.
The idea of the tool-box of statements that can be assumed without being stated,
and the idea of local organisation of mathematical knowledge outside of an axiomatic
system differ markedly from the assumption of the main teaching approaches
adopted in the late twentieth century. We will discuss these in the next section,
along with the poor results they seem to have produced.
RESULTS FROM NEW MATH AND TWO-COLUMN PROOF TEACHING
The research findings in Chapter 4 come primarily from the US and UK and date
back to the 1960s. They generally support the conclusion that the teaching methods
215

via free access
CHAPTER 12
in use when the studies were conducted were not successful in teaching proof and
proving to most students. Therefore, an examination of those teaching methods
might indicate problematic aspects of them that contributed to students’ poor under-
standing.
The earliest studies (Reynolds, 1967, Bell, 1976, and Williams, 1979) were
conducted in the 1960s and early 1970s, at a time when teaching was strongly
influenced by the New Math movement. An emphasis on proof was a key aspect of
New Math teaching and this emphasis was carried into the early grades. By the late
1970s there was general agreement that there had been problems with this
approach, even among supporters of the New Math. Here is how three mathematics
educators of the time looked back on the teaching approaches and emphasis on
axiomatic proof at this time:
As an overreaction to the mechanical memorization of algorithm steps, we
sought to break the algorithm down into its component parts to relate it to our
newly identified mathematical structure and a new emphasis on the notion
of proof. In our zeal to relate these algorithms to the rest of mathematical
structure, we wrote, and led the children to write, lengthy step-by-step proofs
of the various whole-number algorithms. Certainly, one can hardly argue with
the effort to give meaning to these computations. Unfortunately, these efforts
were carried too far, and, as a result, everyday, whole-number algorithms
sometimes became pages of carefully developed axiomatic proof. It takes
only a little hindsight to recognize that these carefully detailed axiomatic proofs
obscured the meaning and the purpose of the algorithm. (Hill, Rouse & Wesson,
1979, p. 78)
The students participating in the research of Reynolds, Bell and Williams would
have been exposed to this approach to proof for at least some of their school years,
however many of them continued to accept or use examples as verification of a
mathematical statement, many could not write a correct proof and some rejected
a deductive proof as a verification. This suggests that the formal, axiomatic approach
to teaching proof adopted in the New Math was not a success.
In the years immediately following the generally rejection of the New Math in
the 1970s a variety of approaches were taken to proof. In some places a back-to-
the-basics movement took hold, in which there was a de-emphasis on proof and
renewed focus on knowledge of facts, rules and procedures. In other places proofs
were taught, but primarily in the context of a year long geometry course with a
focus on two-column proofs. In still other places a discovery approach to mathe-
matics was adopted in which students investigated and discovered mathematical
statements, but did not prove them. The research reviewed in Chapter 4 does not
usually describe the curriculum in use, but the results suggest that none of these
approaches was significantly better than the New Math.
It is not surprising that the approach of not teaching proof adopted by the back-
to-the-basics movement did not result in more students coming to understand proof.
Herbst (2002a) describes some aspects of teaching two-column proofs that may
contribute to students not learning proof well through this approach. He describes
216

via free access
how the form of the two-column proof and the teaching practices associated with it
serve to make the task of writing a proof easier for the students to do and for the
teacher to teach. However, they also lead to the teacher providing significant
guidance to the students resulting in a division of labour that could be described by
saying the teacher proves and the students write down the proof. Hoyles (1997)
describes the effects a discovery based curriculum had on students’ approaches to
proof in the UK. In the 1995 UK National Curriculum, proof is placed at the end of
an investigation process, and associated with a high level of attainment. This means
that “the majority of students will engage in data generation, pattern recognition,
and inductive methods while only a minority, at levels 7 or 8, are expected to prove
their conjectures in any formal sense” (p. 9). Hoyles relates this focus on investi-
gations to the strong preference for empirical arguments she found in her research
(see Chapter 4).
It is interesting to note that the approaches to proof advocated in the New Math
reforms were based primarily on theoretical considerations without taking into
consideration classroom practices or conditions. In contrast, two-column proofs as
a support for teaching developed in the context of classrooms as teachers struggled
to find a way to teach proof that was possible given their circumstances. The ultimate
lack of success of these two approaches suggests that neither a purely theoretical
nor a purely pragmatic basis for a teaching approach is sufficient.
For those seeking new ways of teaching proof these examples of unsuccessful
approaches are useful. However, to imagine another approach requires more than
simply doing things differently. It requires as a starting point a thorough understan-
ding of how students reason and argue in classrooms. In the next section we will
turn to these topics.
STARTING WHERE STUDENTS AND TEACHERS ARE
Much of the work we included in Chapters 6, 8, 10 and 11 involves describing the

nature of students’ reasoning and argumentation in existing classroom contexts. In
this chapter we will consider this as a starting point for teaching. Such descriptions
are needed if one believes that learning is based on prior learning. This is true both
for individuals and for communities. For individuals this assumption is common to
most contemporary theories of learning. In the context of learning proof this
means that the ways of reasoning a student can learn are determined by the ways of
reasoning that person already engages in and has experienced. A better understanding
and description of the reasoning that occurs in students’ mathematical activity at
present establishes the basis for teaching in the future. Similarly, for communities
(in this case classrooms) the changes that are possible are determined by the present
state of that community. Changes to how teachers teach proof must be based on a
detailed understanding of how teachers now teach proof, and the context of that
teaching, for that is where the process of change must begin.
The main implications for teaching arising from the descriptions of reasoning
and patterns of reasoning in Chapters 6 and 11 are related to simultaneously expan-
ding and making more precise our ideas of what students can do and could be
217

via free access
CHAPTER 12
taught to do. For example, by looking carefully at the reasoning of elementary

school students (as Zack has done) we can see that they can reason deductively in
appropriate contexts, a finding that is supported by research in psychology
(Stylianides & Stylianides, 2008). What seems to be an issue is the formulation of
that reasoning and extending that reasoning to more abstract contexts, which
suggests that these be the focus of proof teaching at the elementary school level.
The examination of types and roles of reasoning in Chapter 6 suggests that descrip-
tions of mathematical reasoning (in curriculum documents, for example) should not
be limited to inductive discovery and deductive verification. The roles of reasoning
by analogy and abductive reasoning need also to be included. Having more precise
descriptions of reasoning is the basis for a more sophisticated approach to defining
the goals of teaching. For example, the Proof Analysis pattern of reasoning
(described in Chapter 11) is more clearly mathematical than some other patterns.
This suggests that in teaching proof it is important to include contexts in which the
Proof Analysis pattern is likely to occur.
The descriptions of argumentation and argumentation structures in Chapters 8
and 10 also have implications for teaching. First, they allow for more clarity when
classroom argumentation is being discussed. Being aware of the key differences in
usages of the word “argumentation” allow one to reconcile contradictory statements.
It is less of an issue to read that argumentation is the necessary basis for learning
proof and that argumentation is an obstacle to learning proof, if one understands
that “argumentation” is meant differently in those two statements. And greater
clarity allows not only discussions of argumentation but also documents that support
teaching to be clearer. While more research is needed, Krummheuer’s work and the
work of Boero, Douek, Mariotti, Pedemonte, Bartolini Bussi and their colleagues
provide some strong indications of ways classroom argumentation shapes and
supports students’ learning of proof, which are reflected in some of the teaching
experiments described in Chapter 9. The idea of argumentation structures in class-
room discussions leading to proofs adds a further level of description, allowing
different teaching goals to be described in terms of their effect on the argumentation
structure. It would be interesting to turn this around and to see if descriptions of
argumentation structures can be used in teacher development as a focus for discussion
and setting of teaching goals.
Most of the teaching experiments described in Chapter 9 share some common

features that are suggestive, most of which we summarised at the end of that chapter.
They include students proving their own conjectures, limited attention to form and
content versus discourse, an expectation for explanation and allocation of sufficient
time to develop explanations. Given the range of contexts they come from, we
believe those commonalties indicate features of importance. Here we will add to
them by considering them in relation to the implications for teaching from other
chapters that we introduced above.
218

via free access
Fawcett’s teaching began with an exploration of the nature of definitions and

his students’ construction of a “Theory of Space” was based on definitions and
undefined terms they identified before proving any theorems. This approach is
axiomatic rather than being based on exploring the local organisation of concepts.
The students were explicitly developing a tool-box to use in their proving and their
notebooks became the concrete expression of this tool-box. However, once a
statement entered into the tool-box it was not fixed. Through a process similar to
that described by Lakatos (1976) definitions and assumptions were revisited and
refined. Fawcett’s approach could be described as explicitly filling the tool-box,
starting with the most basic propositions, but through a quasi-empirical process.
In the debate approach proposed by the Grenoble/Lyon group, in contrast, there
is no explicit tool-box. The students make use of anything they know or believe
when trying to convince their peers of the correctness of their solutions to the
problems posed. In the debate some of the things are questioned and some are not.
Unquestioned assertions could be thought of as being part of a implicit tool-box,
but this implicitness and the focus of the reasoning around a specific problem
suggests that it makes more sense to think of what the students are doing in terms
of local organisation. The reasoning around each problem has a structure including
assumptions, but these assumptions do not necessarily carry over to another problem.
They are locally valid, but not global.
The situation in Zack’s classroom is similar, as the Problem of the Week tasks
are largely independent of each other and of the mathematics content being taught
at other times. There are exceptions, however. Sometimes a concept (e.g., square
number) became important in the solution of a problem and Zack took some time to
explicitly discuss the definition, so that it became part of the implicit tool-box
applicable in later problem solving. Also, because some Problem of the Week tasks
are similar to others (there are four that involve summing a series) procedures used
in one problem became part of the tool-box for the solution of other problems.
Largely, however, the nature of arguments in Zack’s classroom was one of local
organisation.
In the Italian teaching experiments considerable time was spent exploring a
single context, that of sun shadows. This means that it was possible for proofs to
refer back to prior results. Embedding proofs in a “reference theory” is seen as
essential to proving in Mariotti’s research perspective (see Chapter 3) and so it is
not surprising that teaching experiments developed from that perspective allow for
the development of a tool-box over time.
In these four teaching experiments we can see that the ideas of local organisation
and the tool-box provide a way to characterise and discuss teaching. They also
show that even in situations that are primarily based on local organisation, a tool-
box can be developed, although it might not be.
SUMMARY
An important theme of this book is the diversity of meanings given to “proof ” and
the range of roles it is given. While there is no definitive position on these points, it
219

via free access
CHAPTER 12
is clear that different meanings and roles for proof lead to different teaching methods.
There are, however, some implications for teaching that are sensible from a number
of positions.
The results of teaching proof in an axiomatic context in the New Math and with
a focus on form through two column proofs did not lead to student learning. This
may be because excessive formality ignores the semi-formal nature of mathema-
ticians’ proofs and the usefulness of preformal proofs in schools. Teaching with
preformal proofs must necessarily involve only local organisation of mathematical
knowledge. As proofs become more formal and as the structures they are embedded
in become more explicit, a tool-box of accepted results must be developed to support
further proving.
Teaching experiments involving students solving problems, conjecturing and
explaining and verifying their conjectures seem to provide good contexts for proving.
Careful examination and description of students’ patterns of reasoning and argumen-
tation structures in these and other classrooms should permit research to explore in
detail how different meanings and roles for proof and different levels of formality,
along with other factors, influence teaching.
220

via free access
CHAPTER 13
DIRECTIONS FOR RESEARCH
In the previous chapter we discussed some implications of past research for

teaching proof. Here we will outline some areas where further research is needed.
As we noted in the introduction, this book is intended to provide a synthesis of the
existing research literature on teaching and learning proof and proving. We hope
that researchers and students will find this book helpful in providing a thorough
background in the research literature to date, and this chapter useful as a guide
toward interesting research questions.
Each chapter in Parts 2 and 3 ended with a section describing ideas for research.
Here we will explore common themes that emerge from these sections. We will
group them according to whether they are primarily related to teaching, learning, or
conceptual issues.
TEACHING PROOF
While Chapter 12 suggested some aspects of teaching proof that seem promising,
more research remains to be done. In this section we will propose some research
questions related to teaching proof. The first concerns the extent to which what is
taught in schools can or should reflect practices of professional mathematicians.
This is related to the question of why proof should be taught in schools and to
the question of formulation. For professional mathematicians making reasoning
explicit is an essential part of practice, but should this be the case for school
students? How aware must students become of their own reasoning? And to
what extent does the teaching of proof require entering into the practices of
professional mathematics? What else is required? Would awareness of the
argumentation structures in their own classrooms help teachers to teach proof
more successfully?
Professional Mathematics and School Mathematics

An assumption is often made that the proving practices of mathematicians and the
role of proof in professional mathematics should have some bearing on the teaching
of proof in schools. Exactly what bearing professional practice should have,
however, is not clear. In fact, given that many students will have no contact with
mathematics after they leave school, it might be argued that the practices of those
who use mathematics (e.g., scientists, tradespeople, engineers, medics, etc.) should
be more significant. From that perspective a role such as developing logical thinking
might be more important than verification or explanation of mathematical statements.
221

via free access
CHAPTER 13
On the other hand, school mathematics could be seen as similar to subjects in

which students are prepared not to practice in the field, but rather to appreciate the
products of it. In that case a focus on proof reading rather than writing, and a
greater emphasis on roles such as aesthetics might be called for. This is both a
possible research question and a possible policy debate, concerning the purpose of
teaching mathematics and teaching proof.
Formulation
There exists a body of research in both psychology and mathematics education on
the kinds of reasoning students of different ages can use and choose to use; however,
questions remain. One key point that has received little attention is the process of
formulating reasoning. Formulation is a central concern in teaching proving as it
is necessary for the production of a proof-text based on prior reasoning. But how
much formulation is necessary? This question is related to the ideas of the tool-box
and local organisation (see Chapter 12). In teaching proof, local organisation of
results might precede the establishment of explicit axioms and definitions, which,
once established might again fade from view as they become part of the tool-box.
Or the assumptions used in reasoning could be left unstated and become part of the
tool-box without ever being questioned. The advantages of making assumptions
explicit, and the question of which assumptions must be made explicit, requires
further research.
Teacher Development
Many of the results in this book have implications for teacher development, although
more research is desirable in this area as well. It would be interesting, for example,
to see if descriptions of argumentation structures can be used in teacher development
as a focus for discussion and setting of teaching goals. More fundamentally, research
must address the question of what teachers should know about proof in order to
successfully teach it. How should teachers be taught to teach proof? Do teachers
currently in schools have sufficient understanding of proof to teach differently? If
not then students’ understanding of proof might be best improved by addressing
teachers’ understanding of proof first.
Argumentation Structures
Two-column proofs are the norm in many US classrooms and some in the UK;
however, this is not so in the countries where the research outlined in Chapter 10
took place. It would be interesting to see if there is an argumentation structure
associated with the use of this form, or several related to the topic under conside-
ration. Comparative analyses of this type can deepen our understanding of how
classroom proving processes can be an obstacle or an opportunity for students to
learn to reason mathematically and to engage in proving. It would also be interesting
to examine argumentation structures in classrooms that explicitly espouse an inquiry
222

via free access
mathematics approach. A comparison of the argumentation structures that occur in

such classrooms versus those that occur in more traditional classrooms would
deepen our understanding of the differences in teaching and learning that occurs in
these contexts.
STUDENTS’ UNDERSTANDINGS OF PROOF
Much of the existing research focusses on students’ understandings of proof and

ability to prove. There remains much work to be done in this area, however.
Interesting questions arise in connection with national and temporal differences in
understanding proof, attitudes towards proofs and counterexamples, and students’
motivation to prove.
National Differences
As Hoyles (1997) notes there is a “huge variation in when proof is introduced and
how it is treated in different countries” (p. 7). One would expect this to have
some impact on students’ learning of proof. The TIMSS data in Chapter 4 indicate
that this variation is considerable, but does not reveal details. Research on this
question is especially important as much of the research literature comes from
a few, predominantly English speaking countries. Applying this research to other
countries, given the variations in teaching approaches between them, is probably
foolish.
The comments of Sekiguchi and Miyazaki (2000) also suggest another direction
for future research. Comparative studies on argumentation are rare, and Sekiguchi
and Miyazaki offer us insight into another perspective that sheds light on aspects of
argumentation that might go unnoticed because they are generally assumed by
Western researchers (see Chapter 8). These comparisons could be extended to other
languages and cultures.
Changes Over Time

Are students less able to prove now than in the past? As noted in Chapter 4,
Reynolds (1967) and Healy and Hoyles (2000) both conducted large scale written
assessments in the UK, but separated by thirty years. Their results suggest that
there is not much difference over time in the level of students’ understanding of proof
and proving, except for a small decline in proof writing skill. In 1995 the London
Mathematical Society published a report that identified “serious problems perceived
by those in higher education”. They included “a changed perception of what mathe-
matics is – in particular the essential place within it of precision and proof.” (p. 2).
Has this perception actually changed over time? Among which students? Investigating
changes in proof ability and attitudes over time would be an interesting area for
research. Detailed examination of the assessments used would be required, and
ways to identify the effects of curriculum changes, social changes and other factors
would have to be developed.
223

via free access
CHAPTER 13
Rejection of Proofs and Counterexamples
Many students do not accept deductive proofs as verification. However, more research
is needed on the circumstances under which students reject deductive proofs, and
their reasons for doing so. Students’ attitudes towards counterexamples also deserve
additional study. Some of the research cited in Chapters 4 and 11 suggests that
some students do not consider counterexamples sufficient to reject a general state-
ment. However, other students seek out counterexamples as a way of testing
statements. The circumstances leading to these two approaches require clarification.
The Role of Proof and Motivation to Prove

More research is necessary to examine the effects of giving a specific role to
proof on students’ need to prove or their acceptance of proofs. While Mudaly and
de Villiers (2000) have looked at explanation as a motivation and report that proofs
are accepted by students seeking an explanation, other roles remain to be explored,
and given the complexity of teaching proof the roles of verification, explanation
and communication must be examined further.
CONCEPTIONAL ISSUES
A number of research questions are related to what we think proof is, what reasoning
it involves and how that reasoning is related to the role of proof and argumentation
in general.
Reasoning and Roles

The traditional stereotype of discovery through inductive reasoning and verification
through deductive reasoning has been shown to be too limited to describe mathe-
matical reasoning (see Chapter 6). But it is not yet clear what kinds of reasoning
can (theoretically or empirically) address a need to explain, explore, verify and so
on. A related question is how students might develop specific emotional reactions
to certain kinds of reasoning, for example, having a feeling of certainty related to
deductive reasoning and not to other kinds of reasoning. Identification of patterns
of reasoning beyond those described in Chapter 11 might aid in coming to a better
understanding in this area.
Abduction and Analogy

Abductive reasoning and reasoning by analogy have a role in exploration and
discovery, but these are both areas in which research is ongoing (see, for example,
the work of Hofstadter, 1995, 2001) and mathematics education may turn out to be
a good context for such research. And if reasoning by analogy is the basis for all
human thought (as Lakoff and Johnson, 1980 have argued) then how can other
224

via free access
kinds of reasoning, especially deductive reasoning, have developed from it? Mathe-
matical activity, because it has a special place for deductive reasoning, may be a
context in which this question can be explored.
Different Kinds of Reasoning and Arguments

There are many open research questions related to whether one kind of reasoning
aims or interferes with the use of another kind in the same context, and similarly,
whether facility with some types of arguments interferes with the use of other types
of arguments (see Chapters 6 and 7). How abductive reasoning and deductive
reasoning are related through cognitive unity is one area that has received consi-
derable attention, but other combinations have not, for example, reasoning by
analogy and deductive reasoning. This question may be related to the question of
how needs and reasoning are linked (see above). Peirce’s description of the roles of
different types of reasoning: abductive reasoning explains and explores; inductive
and deductive reasoning verify (see Chapter 6) offers one possibility, but if the role
of deductive reasoning in a different context is to explore instead of to verify, then
what roles can other kinds of reasoning play?
Proof and Argumentation

The relationship between argumentation and proof is far from clear. As we noted in
Chapter 8, some confusion is related to the use of different terminology, but there
remain genuine questions. For example, how is the social process of argumentation
related to an individual’s reasoning when reading or writing a proof? Is classroom
argumentation a good teaching context for proof, and in what contexts? There has
been a good theoretical groundwork laid for research on these questions but now
more empirical work must be done to apply and develop that groundwork.
CONCLUSION
There has been a vast research literature on proof and proving produced, especially
over the past two decades. However, this research has been conducted from
disparate research perspectives making use of different meanings of basic terms,
and ascribing a range of roles to proof. Balacheff has referred to this situation as a
deadlock.
My claim is that this is a deadlock for the whole field. Unless we have
clarified precisely what this deadlock is like and how it limits our capacity to
share research outcomes, it will be hardly possible to make significant progress
in the field. (2002/2004, p. 4 in 2004)
In addition to this confusion at the theoretical level, there are also stumbling blocks
in making use of the results of empirical research. Work done in specific contexts in
specific times has been taken to apply generally, without taking into account the
limitations resulting from significant curricular and cultural differences between
225

via free access
CHAPTER 13
countries and over time. Many researchers have also been too quick to accept the
conclusions of empirical research in spite of plausible alternative explanations for
research findings. It is hard to say to what extent this situation has hindered
research, but it has certainly contributed to confusion among readers of the research
literature.
In this book we have endeavoured to overcome the theoretical deadlock
identified by Balacheff by shedding light on the diversity of approaches to research
on proof and proving. By providing clarification of researchers’ perspectives, bridges
between perspectives, and explication of important theoretical ideas, we have
provided the basis for researchers to work with an understanding that a diversity of
research perspectives exists. In addition, by structuring empirical research results,
deriving teaching implications from them and making suggestions for future
research we have established a basis for that research, and for significant progress
in the field of teaching and learning proof and proving.
226

via free access
REFERENCES
Aigner, M., & Ziegler, G. (2004). Proofs from The Book (3rd ed.). Berlin: Springer.
Alberta Education. (1991). Math 30 teacher resource manual. Edmonton, AB: Author.
Alcock, L. (2004). Uses of example objects in proving. In M. J. Høines & A. B. Fuglestad (Eds.),
Proceedings of the twenty-eighth conference of the International Group for the Psychology of
Mathematics Education (Vol. 2, pp. 17–24). Bergen, Norway: Bergen University College.
Alibert, D. (1988). Codidactic system in the course of mathematics: How to introduce it. In A. Borbàs
(Ed.), Proceedings of the twelfth international conference of the International Group for the Psychology
of Mathematics Education (Vol. 1, pp. 109–116). Veszprém, Hungary: Ferenc Genzwein, OOK.
Anglin, W. (1994). Mathematics: A concise history and philosophy. New York: Springer.
Arnauld, A., & Nicole, P. (1965). La logiques ou l’art de penser. Paris: PUF. (Originally work published
1662)
Arsac, G. (2007). Origin of mathematical proof: History and epistemology. In P. Boero (Ed.), Theorems
in school: From history, epistemology and cognition to classroom practice (pp. 27–42). Rotterdam,
The Netherlands: Sense Publishers.
Arsac, G., Balacheff, N., & Mante, M. (1992). Teacher’s role and reproducibility of didactical situations.
Educational Studies in Mathematics, 23(1), 5–29.
Arsac, G., Chapiron, G., Colonna, A., Germain, G., Guichard, Y., & Mante, M. (1992). L’initiation au
raisonnement déductif au collège. Lyon: IREM de Lyon and PUL.
Arsac, G., Germain, G., & Mante, M. (1991). Problème ouvert et situation-problème. Lyon: IREM.
Arzarello, F., Micheletti, C., Olivero, F., & Robutti, O. (1998a). A model for analysing the transition to
formal proofs in geometry. In A. Olivier & K. Newstead (Eds.), Proceedings of the twenty-second
annual conference of the International Group for the Psychology of Mathematics Education (Vol. 2,
pp. 24–31). Stellenbosch, South Africa.
Arzarello, F., Micheletti, C., Olivero, F., & Robutti, O. (1998b). Dragging in Cabri and modalities of
transition from conjectures to proofs in geometry. In A. Olivier & K. Newstead (Eds.), Proceedings
of the twenty-second annual conference of the International Group for the Psychology of Mathematics
Education (Vol. 2, pp. 32–39). Stellenbosch, South Africa.
Balacheff, N. (1982). Preuve et démonstration en mathématiques au collège. Recherches en Didactique
des Mathématiques, 3(3), 261–304.
Balacheff, N. (1987). Processus de preuve et situations de validation. Educational Studies in Mathematics,
18, 147–176.
Balacheff, N. (1988a). Aspects of proof in pupils’ practice of school mathematics. In D. Pimm (Ed.),
Mathematics, teachers and children (pp. 316–230). London: Hodder and Stoughton.
Balacheff, N. (1988b). Une étude des processus de preuve en mathématique chez des élèves de Collège.
Thèse de doctorat d’état sciences. Grenoble: Université Joseph Fourier.
Balacheff, N. (1991a). The benefits and limits of social interaction: The case of mathematical proof. In
A. Bishop, S. Mellin-Olson, & J. van Doormolen (Eds.), Mathematical knowledge: Its growth
through teaching (pp. 175–192). Boston: Kluwer Academic.
Balacheff, N. (1991b). Treatment of refutations: Aspects of the complexity of a constructivist approach
of mathematics learning. In E. von Glasersfeld (Ed.), Radical constructivism in mathematics education
(pp. 89–110). Dordrecht, The Netherlands: Kluwer Academic Publisher.
Balacheff, N. (1999). Is argumentation an obstacle? Invitation to a debate (V. Warfield, Trans.). Retrieved
from http://www.lettredelapreuve.it/Newsletter/990506Theme/990506ThemeUK.html (Original work at
http://www.lettredelapreuve.it/Newsletter/990506Theme/990506ThemeFR.html)
Balacheff, N. (2002/2004). The researcher epistemology: A deadlock for educational research on
proof. In Fou Lai Lin (Ed.), Proceedings of 2002 international conference on mathematics:
Understanding proving and proving to understand (pp. 23–44). Taipei: NSC and NTNU. Reprinted
227

via free access
REFERENCES
in Les cashiers du laboratoire Leibniz, no 109, August 2004. Retrieved from http://www-leibniz.
imag.fr/NEWLEIBNIZ/LesCahiers/Cahier109/ResumCahier109.html
Balacheff, N. (2008). The role of the researcher’s epistemology in mathematics education: An essay on
the case of proof. ZDM The International Journal on Mathematics Education, 40, 501–512. doi:
10.1007/s11858-008-0103-2.
Barbin, E. (1996). Can proof be taught? In E. Barbin & R. Douady (Eds.), Teaching mathematics: The
relationship between knowledge, curriculum and practice (pp. 195–210). Pont-à-Mousson: Topiques
éditions.
Barkai, R., Tsamir, P., Tirosh, D., & Dreyfus, T. (2002). Proving or refuting arithmetic claims: The case
of elementary school teachers. In A. Cockburn & E. Nardi (Eds.), Proceedings of the twenty-sixth
pp. 57–64). Norwich, UK.
Barnes, J. (Ed.). (1984). The complete works of Aristotle: The revised Oxford translation. Princeton, NJ:
Princeton University Press.
Bartolini Bussi, M. (2000). Early approach to mathematical ideas related to proof making. Contribution
to: P. Boero, G. Harel, C. Maher, M. Miyazaki (organisers), Proof and proving in mathematics
education. ICME9 TSG 12. Tokyo/Makuhari, Japan. Retrieved from http://www.lettredelapreuve.it/
ICME9TG12/ICME9TG12Contributions/BartoliniBussiICME00.html
Bell, A. (1976). A study of pupils’ proof-explanations in mathematical situations. Educational Studies in
Mathematics, 7(1–2), 23–40.
Benis-Sinaceur, H. (2000). The nature of progress in mathematics: The significance of analogy. In
E. Grosholz & H. Breger (Eds.), The growth of mathematical knowledge (pp. 281–293). Dordrecht,
The Netherlands: Kluwer.
Blum, W. (1998). On the role of “Grundvorstellungen” for reality-related proofs: Examples and
reflections. In P. Galbraith, W. Blum, & I. Huntley (Eds.), Mathematical modelling: Teaching and
assessment in a technology-rich world (pp. 63–74). Chichester: Horwood Publishing.
Blum, W., & Kirsch, A. (1979). Zur Konzeption des Analysisunterrichts in Grundkursen. Der Mathematik-
unterricht, 25(3), 6–24.
Blum, W., & Kirsch, A. (1989). Warum haben nicht-triviale Lösungen von keine Nullstellen?
Beobachtungen und Bemerkungen zum inhaltlich-anschaulichen Beweisen. In H. Kautschitsch &
W. Metzler (Eds.), Anschauliches Beweisen (pp. 199–209). Vienna: Hölder-Pichler-Tempsky.
Blum, W., & Kirsch, A. (1991). Preformal proving: Examples and reflections. Educational Studies in
Mathematics, 22, 183–203.
Boero, P. (1999). Argumentation and mathematical proof: A complex, productive, unavoidable relationship
in mathematics and mathematics education. Retrieved from http://www.lettredelapreuve.it/Newsletter/
990708Theme/990708ThemeUK.html
Boero, P., Douek, N., & Garuti, R. (2003). Children’s conceptions of infinity of numbers in a fifth grade
classroom discussion context. In N. Pateman, B. Dougherty, & J. Zilliox (Eds.), Proceedings of the
twenty-seventh annual conference of the International Group for the Psychology of Mathematics
Education (Vol. 2, pp. 121–128). Honolulu, HI: University of Hawaii.
Boero, P., & Garuti, R. (1994). Approaching rational geometry: From physical relationships to
conditional statements. In I. Hirabayashi, N. Nohda, K. Shigematsu, & F.-L. Lin (Eds.), Proceedings
of the seventeenth annual conference of the International Group for the Psychology of Mathematics
Education (Vol. 2, pp. 96–103). Tsukuba, Japan.
Boero, P., Chiapini, G., Garuti, R., & Sibilla, A. (1995). Towards statements and proofs in elementary
arithmetic: An exploratory study about the role of teachers and the behaviour of students. In L. Meira &
D. Carraher (Eds.), Proceedings of the nineteenth annual conference of the International Group for
the Psychology of Mathematics Education (Vol. 3, pp. 129–136). Recife, Brazil.
Boero, P., Garuti, R., & Lemut, E. (1999). About the generation of conditionality of statements and its
links with proving. In O. Zaslavsky (Ed.), Proceedings of the twenty-third conference of the
International Group for the Psychology of Mathematics Education (Vol. 2, pp. 137–144). Haifa,
Israel: PME.
228

via free access
REFERENCES
Boero, P., Garuti, R., & Mariotti, M. A. (1996). Some dynamic mental processes underlying producing
and proving conjectures. In L. Puig & A. Gutierrez (Eds.), Proceedings of the twentieth conference
of the International Group for the Psychology of Mathematics Education (Vol. 2, pp. 121–128).
Valencia, Spain: University of Valencia.
Boero, P., Garuti, R., Lemut, E., & Mariotti, M. A. (1996). Challenging the traditional school approach
to theorems: A hypothesis about the cognitive unity of theorems. In L. Puig & A. Gutierrez (Eds.),
Proceedings of the twentieth conference of the International Group for the Psychology of Mathematics
Education (Vol. 2, pp. 113–120). Valencia, Spain: University of Valencia.
Bonfantini, M., & Proni, G. (1983). To guess or not to guess. In U. Eco & T. Sebeok (Eds.), The sign of
three: Dupin, Holmes, Peirce (pp. 119–134). Bloomington, IN: Indiana University Press.
Borel, J. M. (1983). [Referenced by Duval, 1990, but not in his reference list. We have been unable to
locate the original source.]
Branford, B. (1908). A study of mathematical education: Including the teaching of arithmetic. Oxford:
Clarendon Press.
Branford, B. (1913). Betrachtungen über mathematische Erziehung vom Kindergarten bis zur Universität.
(R. Schimmack & H. Weinreich, Trans.). Leipzig & Berlin: Teubner. (Original work published
1908).
Brown, J. R. (1999). Philosophy of mathematics: An introduction to the world of proofs and pictures.
London: Routledge.
Bruner, J. S. (1986). Actual minds, possible worlds. Cambridge, MA: Harvard University Press.
Burton, L., & Morgan, C. (2000). Mathematicians writing. Journal for Research in Mathematics Education,
31(4), 429–452. doi: 10.2307/749652.
Cabassut, R. (2005). Argumentation and proof in examples taken from French and German textbooks. In
M. Bosch (Ed.), Proceedings of the fourth congress of the European Society for Research in Mathematics
Education. Sant Feliu de Guixols, Spain. Retrieved from http://ermeweb.free.fr/CERME4/
CERME4_WG4.pdf#page=9
Carroll, L. (1958). Symbolic logic. New York: Dover. (Original work published 1897)
Chalmers, A. F. (1982). What is this thing called science? St. Lucia, Queensland, Australia: University
of Queensland Press.
Chazan, D. (1993). High school geometry students’ justification for their views of empirical evidence
and mathematical proof. Educational Studies in Mathematics, 24(4), 359–387.
Cifarelli, V., & Sáenz-Ludlow, A. (1996). Abductive processes and mathematics learning. In
E. Jakubowski, D. Watkins, & H. Biske (Eds.), Proceedings of the eighteenth annual meeting of
the North American chapter of the International Group for the Psychology of Mathematics Education
(Vol. I, pp. 161–166). Columbus, OH: ERIC Clearinghouse for Science, Mathematics, and Environ-
mental Education.
Cobb, P., Yackel, E., & Wood, T. (1989). Young children’s emotional acts while doing mathematical
problem solving. In D. McLeod & V. Adams (Eds.), Affect and mathematical problem solving:
A new perspective (pp. 117–148). New York: Springer-Verlag.
Coe, R., & Ruthven, K. (1994). Proof practice and constructs. British Educational Research Journal,
20, 41–53.
Davis, P. J., & Hersh, R. (1981). The mathematical experience. Boston: Birkhäuser.
Davis, P. J., & Hersh, R. (1983). The mathematical experience. Great Britain: Pelican Books. (UK
edition of David & Hersh, 1981)
de Villiers, M. (1990). The role and function of proof in mathematics. Pythagoras, 24, 17–24.
de Villiers, M. (1991a). Pupils’ needs for conviction and explanation within the context of geometry. In
F. Furinghetti (Ed.), Proceedings of the fifteenth conference of the International Group for the
Psychology of Mathematics Education (Vol. 1, pp. 255–262). Assisi, Italy.
de Villiers, M. (1991b). Pupils’ need for conviction and explanation within the context of geometry.
Pythagoras, 26, 18–27.
de Villiers, M. (1999). Rethinking proof with the Geometer’s Sketchpad. Emeryville, CA: Key Curriculum
Press.
229

via free access
REFERENCES
Descartes, R. (1993). Discourse on method and meditations on first philosophy (D. Cress, Trans.).
Indianapolis, IN: Hackett Publishing. (Original work published 1637)
Dhombres, J. (1993). Is one proof enough? Travels with a mathematician of the baroque period.
Douek, N. (1998). Some remarks about argumentation and mathematical proof and their educational
implications. In I. Schwank (Ed), Proceedings of the first congress of the European Society for
Research in Mathematics Education (Vol. 1, pp. 125–139). Retrieved from http://www.fmd.uni-
osnabrueck.de/ebooks/erme/cerme1-proceedings/cerme1-proceedings-1-v1-0-2.pdf
Douek, N. (1999a). Argumentation and conceptualisation in context: A case study on sunshadows in
primary school. Educational Studies in Mathematics, 39, 89–110.
Douek, N. (1999b). Argumentative aspects of proving: Analysis of some undergraduate mathematics
students’ performances. In O. Zaslavsky (Ed.), Proceedings of the twenty-third conference of the
International Group for the Psychology of Mathematics Education (Vol. 2, pp. 273–288). Haifa,
Israel: PME.
Douek, N. (2002). Context, complexity and argumentation. In A. Cockburn & E. Nardi (Eds.), Proceedings
of the twenty-sixth annual conference of the International Group for the Psychology of Mathematics
Education (Vol. 2, pp. 297–304). Norwich: PME.
Douek, N. (2005). Argumentation and development of mathematical knowledge. In A. Chronaki &
I. M. Christiansen. (Eds.), Challenging perspectives on mathematics classroom communication
(pp. 145–172). Greenwich, CN: Information Age Publishing.
Douek, N. (2007). Some remarks about argumentation and proof. In P. Boero (Ed.), Theorems in school:
From history, epistemology and cognition to classroom practice (pp. 163–181). Rotterdam, The
Netherlands: Sense Publishers.
Duval, R. (1990). Pour une approche cognitive de l’argumentation. Annales de Didactique et de Sciences
Cognitives, 3, 195–221. Strasbourg: IREM de Strasbourg.
Duval, R. (1991). Structure du raisonnement déductif et apprentissage de la démonstration. Educational
Studies in Mathematics, 22(3), 233–261.
Duval, R. (1992–1993). Argumenter, démontrer, expliquer: continuité ou rupture cognitive? Petit x, 31,
37–61.
Duval, R. (1999). Questioning argumentation (V. Warfield, Trans.). Retrieved from http://www.lettre
delapreuve.it/Newsletter/991112Theme/991112ThemeUK.html (Original at http://www.Lettredela
preuve.it/ Newsletter/991112Theme/991112ThemeFR.html)
Duval, R. (2007). Cognitive functioning and the understanding of mathematical processes of proof.
In P. Boero (Ed.), Theorems in school: From history, epistemology and cognition to classroom
practice (pp. 137–161). Rotterdam, The Netherlands: Sense Publishers.
Duval, R., & Egret, M.-A. (1989). L’organisation déductive du discours: Interaction entre structure
profonde et structure de surface. Annales de Didactique et de Sciences Cognitives, 2, 25–40.
Duval, R., & Egret, M.-A. (1993). Introduction à la démonstration et apprentissage du raisonnement
déductif. Repère-IREM, 12, 114–140.
Eco, U. (1983). Horns, hooves, insteps: Some hypotheses on three types of abduction. In U. Eco &
T. Sebeok (Eds.), The sign of three: Dupin, Holmes Peirce (pp. 198–220). Bloomington, IN: Indiana
University Press.
Egret, M.-A., & Duval, R. (1989). Comment une classe de quatrième a pris conscience de ce qu’est une
démarche de démonstration. Annales de Didactique et de Sciences Cognitives, 2, 65–89.
English, L. (1997). Mathematical reasoning: Analogies, metaphors and images. Mahwah, NJ: Lawrence
Erlbaum Associates.
Ernest, P. (1991). The philosophy of mathematics education. London: Falmer Press.
Ernest, P. (1998). Social constructivism as a philosophy of mathematics. Albany, NY: State University of
New York Press.
Fawcett, H. (1938). The nature of proof: A description and evaluation of certain procedures used in
a senior high school to develop an understanding of the nature of proof (NCTM yearbook 1938).
New York: Teachers’ College, Columbia University.
230

via free access
REFERENCES
Fischbein, E. (1982). Intuition and proof. For the Learning of Mathematics, 3(2), 9-18.
Fischbein, E. (1999). Intuitions and schemata in mathematical reasoning. Educational Studies in
Mathematics, 38(1–3), 11–50.
Fischbein, E., & Kedem, I. (1982). Proof and certitude in the development of mathematical thinking. In
A. Vermandel (Ed.), Proceedings of the sixth annual conference of the International Group for the
Psychology of Mathematics Education (pp. 128–131). Antwerp.
Freudenthal, H. (1971). Geometry between the devil and the deep sea. Educational Studies in Mathematics,
3(3–4), 413–435.
Freudenthal, H. (1973). Mathematics as an educational task. Dordrecht, The Netherlands: Reidel.
Freudenthal, H. (1980). Weeding and sowing: Preface to a science of mathematical education. Dordrecht,
The Netherlands: Reidel.
Galbraith, P. L. (1981). Aspects of proving: A clinical investigation of process. Educational Studies in
Mathematics, 12(1), 1–28.
Galotti, K., Komatsu, L., & Voelz, S. (1997). Children’s differential performance on deductive and inductive
syllogisms. Developmental Psychology, 33(1), 70–78.
Garuti, R., Boero, P., & Lemut, E. (1998). Cognitive unity of theorems and difficulty of proof. In A. Olivier &
K. Newstead (Eds.), Proceedings of the twenty-second annual conference of the International Group
for the Psychology of Mathematics Education (Vol. 2, pp. 345–352). Stellenbosch, South Africa.
Godino, J., & Recio, A. (1997). Meaning of proofs in mathematics education. In E. Pekhonen (Ed.),
Proceedings of the twenty-first conference of the International Group for the Psychology of Mathematics
Education (Vol. 2, pp. 313–320). Lahti, Finland.
Hanna, G. (1983). Rigorous proof in mathematics education. Toronto, ON: OISE Press.
Hanna, G. (1989). Proofs that prove and proofs that explain. In G. Vergnaud, J. Rogalski, & M. Artigue
(Eds.), Proceedings of the thirteenth international conference on the Psychology of Mathematics
Education (Vol. 2, pp. 45–51). Paris.
Hanna, G. (1990). Some pedagogical aspects of proof. Interchange, 21(1), 6–13.
Hanna, G. (1991). Mathematical proof. In D. Tall (Ed.), Advanced mathematical thinking (pp. 54–61).
Dordrecht, The Netherlands: Kluwer.
Hanna, G. (1996). The ongoing value of proof. In L. Puig & A. Gutierrez (Eds.), Proceedings of the
twentieth conference of the International Group for the Psychology of Mathematics Education (Vol. 1,
pp. 21–34). Valencia.
Hanna, G. (2000). Proof, explanation and exploration: An overview. Educational Studies in Mathematics,
44(1–2), 5–23.
Hanna, G., & Barbeau, E. (2002). What is proof ? In B. Baigrie (Ed.), History of modern science and
mathematics (4 Vols.), (Vol. 1, pp. 36–48). New York: Charles Scribner’s Sons.
Hanna, G., & Jahnke, H. N. (1993). Proof and application. Educational Studies in Mathematics, 24(4),
421–438.
Hanna, G., & Jahnke, H. N. (1996). Proof and proving. In A. Bishop, M. A. Clements, C. Keitel,
J. Kilpatrick, & C. Laborde (Eds.), International handbook of mathematics education (pp. 877–908).
Dordrecht, The Netherlands: Kluwer.
Hanna, G., & Jahnke, H. N. (2002a). Another approach to proof. Zentralblatt für Didaktik der Mathematik,
34(1), 1–8.
Hanna, G., & Jahnke, H. N. (2002b). Arguments from physics in mathematical proofs: An educational
perspective. For the Learning of Mathematics, 22(3), 38–45.
Hannaford, C. (1998). Mathematics teaching is democratic education. Zentralblatt für Didaktik der
Mathematik, 30(6), 181–187.
Harel, G. (2001). The development of mathematical induction as a proof scheme: A model for DNR-
based instruction. In S. Campbell & R. Zazkis (Eds.), The learning and teaching of number theory
(pp. 185–212). Dordrecht, The Netherlands: Kluwer.
Harel, G. (2007). Students’ proof schemes revisited. In P. Boero (Ed.), Theorems in school: From
history, epistemology and cognition to classroom practice (pp. 65–78). Rotterdam, The Netherlands:
Sense Publishers.
231

via free access
REFERENCES
Harel, G., & Fuller, E. (2009). Current contributions toward comprehensive perspectives on the learning
and teaching of proof. In D. Stylianou, M. Blanton, & E. Knuth (Eds.), Teaching and learning proof
across the grades: A K-16 perspective (pp. 355–370). New York: Routledge.
Harel, G., & Sowder, L. (1996). Classifying processes of proving. In L. Puig & A. Gutierrez (Eds.),
Proceedings of the twentieth conference of the International Group for the Psychology of Mathematics
Education (Vol. 3, pp. 59–65). Valencia.
Harel, G., & Sowder, L. (1998). Students’ proof schemes: Results from exploratory studies. In A. Schoenfeld,
J. Kaput, & E. Dubinsky (Eds.), Research in collegiate mathematics education III (Issues in Mathematics
Education, Vol. 7, pp. 234–282). Providence, RI: American Mathematical Society.
Harel, G., & Sowder, L. (2007). Toward comprehensive perspectives on the learning and teaching of
proof. In F. Lester (Ed.), Second handbook of research on mathematics teaching and learning
(pp. 805–842). Reston, VA: National Council of Teachers of Mathematics.
Healy, C. (1993). Build-a-Book geometry: A story of student discovery. Berkeley, CA: Key Curriculum
Press.
Healy, L., & Hoyles, C. (1998). Justifying and proving in school mathematics. Summary of the results
from a survey of the proof conceptions of students in the UK. Research Report. Mathematical
Sciences, Institute of Education, University of London.
Healy, L., & Hoyles, C. (1999). Students’ performance in proving: Competence or curriculum? In
I. Schwank (Ed.), Proceedings of the first congress of the European Society for Research in
Mathematics Education (Vol. 1, pp. 153–167). Retrieved from http://www.fmd.uni-osnabrueck.de/
ebooks/erme/cerme1-proceedings/papers/g1-healy-et-al.pdf
Healy, L., & Hoyles, C. (2000). A study of proof conceptions in algebra. Journal for Research in
Mathematics Education, 31(4), 396–428.
Heath, T. L. (1956). Euclid: The thirteen books of the elements (Vols. 1–3). New York: Dover.
Herbst, P. (2002a). Engaging students in proving: A double bind on the teacher. Journal for Research in
Mathematics Education, 33, 176–203.
Herbst, P. (2002b). Establishing a custom of proving in American school geometry: Evolution of the
two-column proof in the early twentieth century. Educational Studies in Mathematics, 49, 283–312.
Herbst, P. (2004). Interactions with diagrams and the making of reasoned conjectures in geometry.
Zentralblatt für Didaktik der Mathematik, 36(5), 129–139.
Herbst, P., & Balacheff, N. (2009). Proving and knowing in public: The nature of proof in a classroom.
In D. Stylianou, M. Blanton, & E. Knuth (Eds.), Teaching and learning proof across the grades:
A K-16 perspective (pp. 40–63). New York: Routledge.
Herbst, P., Chen, C., Weiss, M., & González, G. (2009). “Doing proofs” in geometry classrooms. In
D. Stylianou, M. Blanton, & E. Knuth (Eds.), Teaching and learning proof across the grades: A K-
16 perspective (pp. 250–268). New York: Routledge.
Hersh, R. (1993). Proving is convincing and explaining. Educational Studies in Mathematics, 24(4),
389–399.
Hersh, R. (2009). What I would like my students to already know about proof. In D. Stylianou, M. Blanton, &
E. Knuth (Eds.), Teaching and learning proof across the grades: A K-16 perspective (pp. 17–20).
New York: Routledge.
Hilbert, D. (1921). The foundations of geometry (E. J. Townsend, Trans., 2nd ed.). Chicago: Open
Court. (Original work published in 1899)
Hill, J., Rouse, W., & Wesson, J. (1979). Mathematics education: Reactionary regression or responsible
reform? Elementary School Journal, 80(2), 76–79.
Hofstadter, D. (1995). Fluid concepts & creative analogies: Computer models of the fundamental
mechanisms of thought (together with the Fluid Analogies Research Group). New York: Basic Books.
Hofstadter, D. (2001). Analogy as the core of cognition. In D. Gentner, K. Holyoak, & B. Kokinov
(Eds.), The analogical mind: Perspectives from cognitive science (pp. 499–538). Cambridge, MA:
The MIT Press.
Hoyles, C. (1997). The curricular shaping of students’ approaches to proof. For the Learning of
232

via free access
REFERENCES
Hoyles, C., & Küchemann, D. (2002). Students’ understandings of logical implication. Educational
Jahnke, H. N. (1978). Zum Verhältnis von Wissensentwicklung und Begründung in der Mathematik-
Beweisen als didaktisches Problem (IDM Materialien und Studien, Vol. 10). Bielefeld: Universität
Bielefeld.
James, T., & Tyack, D. (1983). Learning from past efforts to reform the high school. Phi Delta Kappan,
64, 400–406.
Johnson, W. E. (1924). Logic, Part III: The logical foundation of science. Cambridge: Cambridge University
Press.
Jones, F. B. (1977). The Moore method. The American Mathematical Monthly, 84(4), 273–278.
Jones, R. B. (1996). The formalisation of mathematics. Retrieved from http://www.rbjones.com/
rbjpub/maths/math006.htm
Jones, R. B. (1997). A short history of rigour in mathematics. Retrieved from http://www.rbjones.com/
rbjpub/maths/math003.htm
Joseph, G. G. (1992). Different ways of knowing: Contrasting styles of argument in India and the West.
In D. Robitaille, D. Wheeler, & C. Kieran (Eds.), Selected lectures from the 7th international congress
on mathematical education. Quebec: Université Laval.
Joseph, G. G. (1994). Different ways of knowing: Contrasting styles of argument in Indian and Greek
mathematical traditions. In P. Ernest (Ed.), Mathematics, education and philosophy: An international
perspective (pp. 194–207). London and Washington, DC: The Falmer Press.
Kant, I. (1927). Critique of pure reason (M. Müller, Trans.). New York: Macmillan. (Original work published
1781)
Kleiner, I. (1991). Rigor and proof in mathematics: A historical perspective. Mathematics Magazine,
64(5), 291–314.
Kline, M. (1962). Mathematics: A cultural approach. Reading, MA: Addison-Wesley.
Kline, M. (1972). Mathematical thought from ancient to modern times. New York: Oxford University
Press.
Kline, M. (1980). Mathematics: The loss of certainty. New York: Oxford University Press.
Knipping, C. (2001). Towards a comparative analysis of proof teaching. In M. van den Heuvel-Panhuizen
(Ed.), Proceedings of the twenty-fifth conference of the International Group for the Psychology of
Mathematics Education (Vol. 3, pp. 249–256). Utrecht, The Netherlands.
Knipping, C. (2002). Proof and proving processes: Teaching geometry in France and Germany. In
H.-G. Weigand, et al. (Eds.), Developments in mathematics education in German-speaking countries.
Selected papers from the annual conference on Didactics of Mathematics Bern 1999 (pp. 44–54).
Hildesheim: Verlag Franzbecker.
Knipping, C. (2003a). Argumentation structures in classroom proving situations. In M. A. Mariotti (Ed.),
Proceedings of the third conference of the European Society in Mathematics Education (unpaginated).
Retrieved from http://ermeweb.free.fr/CERME3/Groups/TG4/TG4_Knipping_cerme3.pdf
Knipping, C. (2003b). Beweisprozesse in der Unterrichtspraxis: Vergleichende Analysen von Mathematik-
unterricht in Deutschland und Frankreich. Hildesheim: Franzbecker Verlag.
Knipping, C. (2004). Argumentations in proving discourses in mathematics classrooms. In G. Törner,
et al. (Eds.), Developments in mathematics education in German-speaking countries. Selected
papers from the annual conference on Didactics of Mathematics, Ludwigsburg, March 5–9, 2001 (pp.
73–84). Hildesheim: Verlag Franzbecker.
Knipping, C. (2005). Challenges in teaching proofs. In H.-W. Henn & G. Kaiser (Eds.), Mathematik-
unterricht im Spannungsfeld von Evolution und Evaluation: Festschrift für Werner Blum (pp. 165–174).
Berlin: Franzbecker.
Knipping, C. (2008). A method for revealing structures of argumentations in classroom proving processes.
ZDM The International Journal on Mathematics Education, 40(3), 427–441. doi: 10.1007/s11858-
008-0095-y
Knorr, W. (1975). The evolution of the Euclidean elements. Dordrecht, The Netherlands: Reidel.
233

via free access
REFERENCES
Knuth, E. (2002a). Secondary school mathematics teachers’ conceptions of proof. Journal of Research
in Mathematics Education, 33(5), 379–405.
Knuth, E. (2002b). Teachers’ conceptions of proof in the context of secondary school mathematics.
Journal of Mathematics Teacher Education, 5, 61–88.
Krummheuer, G. (1995). The ethnography of argumentation. In P. Cobb & H. Bauersfeld (Eds.), The
emergence of mathematical meaning: Interaction in classroom cultures (pp. 229–269). Hillsdale,
NJ: Lawrence Erlbaum.
Krummheuer, G. (2003). Argumentationsanalyse in der mathematikdidaktischen Unterrichtsforschung.
Zentralblatt für Didaktik der Mathematik, 35(6), 247–256.
Krummheuer, G. (2007). Argumentation and participation in the primary mathematics classroom: Two
episodes and related theoretical abductions. Journal of Mathematical Behavior, 26(1), 60–82.
Krummheuer, G., & Brandt, B. (2001). Paraphrase und Traduktion. Partizipationstheoretische Elemente
einer Interaktionstheorie des Mathematiklernens in der Grundschule. Weinheim: Beltz-Deutscher
Studienverlag.
Lakatos, I. (1961). Essays in the logic of mathematical discovery. Unpublished Ph.D. dissertation:
University of Cambridge. (Edited and reprinted as Proofs and Refutations in Lakatos, 1963–64,
1976.)
Lakatos, I. (1963–64). Proofs and refutations. The British Journal for the Philosophy of Science, 14, 1–25,
120–139, 221–245, 296–342. (Edited and reprinted in Lakatos, 1976)
Lakatos, I. (1976). Proofs and refutations. Princeton, NJ: Princeton University Press.
Lakatos, I. (1978). Mathematics, science and epistemology. Cambridge, NJ: Cambridge University Press.
Lakatos, I. (1984). Preuves et réfutations. Essai sur la logique de la découverte mathématique.
(N. Balacheff & J. M. Laborde Trans.). Paris: Edition Hermann. (Original work published in 1976)
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.
Lampert, M. (1990). When the problem is not the question and the solution is not the answer:
Mathematical knowing and teaching. American Educational Research Journal, 27(1), 29–63.
London Mathematical Society/Institute for Mathematics and its Applications/Royal Statistical Society.
(1995). Tackling the mathematics problem. London: LMS/IMA/RSS.
Lovell, K. (1971). The development of the concept of mathematical proof in abler pupils. In M. Rosskopf,
L. Steffe, & S. Taback (Eds.), Piagetian cognitive-development research and mathematical education
(pp. 66–80). Washington, DC: National Council of Teachers of Mathematics.
MacKernan, J. (1996). What’s the point of proof ? In D. Ball (Ed.), Teaching proof. Special issue,
Mathematics teaching, 155, 14–20.
Maher, C., & Martino, A. (1996a). The development of the idea of mathematical proof: A 5-year study.
Journal for Research in Mathematics Education, 27(2), 194–214.
Maher, C., & Martino, A. (1996b). Young children invent methods of proof: The “Gang of Four”. In
P. Steffe, L. P. Steffe, P. Cobb, B. Greer, & J. Goldin (Eds.), Theories of mathematical learning
(pp. 431–444). Hillsdale, NJ: Lawrence Erlbaum Associates.
Manin, Y. (1977). A course in mathematical logic. New York: Springer Verlag.
Mariotti, M. A. (1997). Justifying and proving in geometry: The mediation of a microworld, Revised
and extended version of the version published. In J. Novotna (Ed.), Proceedings of the second
European conference on mathematical education (pp. 21–26). Prague: Prometheus. Retrieved from
http://www.lettredelapreuve.it/resumes/mariotti/mariotti97a/mariotti97a.html
Mariotti, M. A. (2000). Introduction to proof: The mediation of a dynamic software environment.
Educational Studies in Mathematics, 44(1–3), 25–53.
Mariotti, M. A. (2006). Proof and proving in mathematics education. In A. Gutierrez & P. Boero (Eds.),
Handbook of research on the psychology of mathematics education: Past, present and future
(pp. 173–204). Rotterdam, The Netherlands: Sense.
Mariotti, M. A., Bartolini Bussi, M., Boero, P., Ferri, F., & Garuti, R. (1997). Approaching geometry
theorems in contexts: From history and epistemology to cognition. In E. Pekhonen (Ed.), Proceedings
of the twenty-first conference of the International Group for the Psychology of Mathematics
234

via free access
REFERENCES
Martin, G., & Harel, G. (1989). Proof frames of preservice elementary teachers. Journal for Research in
Mathematics Education, 20(1), 41–51.
Martin, M. O., & Kelly, D. L. (Eds.). (1998). TIMSS technical report Vol. 3: Implementation and
analysis (Final year of secondary school). Chestnut Hill, MA: Boston College. Retrieved from
http://timss.bc.edu/timss1995i/TechVol3.html
Martzloff, J.-C. (1997). A history of Chinese mathematics. Berlin: Springer.
Martzloff, J.-C. (1989). Quelques exemples de démonstration en mathématiques chinoises. In M. Henry/
Commission inter-IREM Histoire et Épistémologie des Mathématiques (Ed.), La Démonstration
Mathématique dans l’Histoire: Actes du 7e colloque inter-IREM “épistémologie et histoire des
mathématiques”. Lyon: IREM de Lyon.
Mason, J. (1996). Abduction at the heart of mathematical being. In E. Gray (Ed.), Thinking about
mathematics & music of the spheres: Papers presented for the inaugural lecture of Professor David
Tall (pp. 34–40). Coventry: Mathematics Education Research Centre.
Mason, J., & Pimm, D. (1984). Generic examples: Seeing the general in the particular. Educational
Mason, J., Burton, L., & Stacey, K. (1985). Thinking mathematically (Rev. ed.). New York: Addison
Wesley.
McCrone, S., & Martin, T. (2009). Formal proof in high school geometry: Student perceptions of
structure, validity and purpose. In D. Stylianou, M. Blanton, & E. Knuth (Eds.), Teaching and
learning proof across the grades: A K-16 perspective (pp. 204–221). New York: Routledge.
Mikami, Y. (1913). The development of mathematics in China and Japan. Leipzig: B. G. Teubner.
Mill, J. S. (1884). A system of logic ratiocinative and inductive: Being a connected view of the
principles of evidence and the methods of scientific investigation. New York: Harper.
Monks, K. (2002). Introduction to formal proofs: A toy proof system. Retrieved from http://math.
scranton.edu/monks/courses/math448/ToyProofsV2.html
Moore, R. (1990). College students’ difficulties in learning to do mathematical proofs. Unpublished
doctoral dissertation, University of Georgia.
Movshovits-Hadar, N. (1988a). School mathematics theorems: An endless source of surprise. For the
Learning of Mathematics, 8(3), 34–40.
Movshovits-Hadar, N. (1988b). Stimulating presentation of theorems followed by responsive proofs.
For the Learning of Mathematics, 8(2), 12–30.
Mudaly, V., & de Villiers, M. (2000). Learners’ needs for conviction and explanation within the context
of dynamic geometry. Pythagoras, 52, 20–23.
Nakayama, O. (1989). “Bokashi” no shinri [psychology of “ambiguity”]. Osaka: Sogensha. (in Japanese)
National Council of Teachers of Mathematics. (1989). Curriculum and evaluation standards for school
mathematics. Reston, VA: Author.
National Council of Teachers of Mathematics. (1991). Professional standards for teaching mathematics.
Reston, VA: Author.
National Council of Teachers of Mathematics. (2000). Principles and standards for school mathematics.
Reston, VA: Author.
National Institute of Standards and Technology [NIST]. (2005, February 7). Experiments prove
existence of atomic chain ‘Anchors’. Science Daily. Retrieved from http://www.sciencedaily.com/
releases/2005/02/050204214602.htm
Needham, J. (1959). Science and civilisation in China (Vol. 3). Cambridge: Cambridge University Press.
Netz, R. (1998). Greek mathematical diagrams: Their use and their meaning. For the Learning of
Netz, R. (1999). The shaping of deduction in Greek mathematics: A study in cognitive history. Cambridge,
UK: Cambridge University Press.
Oliveira e Silva, T. (2010). Goldbach conjecture verification. Retrieved from http://www.ieeta.
pt/~tos/goldbach.html
Pedemonte, B. (2002). Etude didactique et cognitive des rapports de l’argumentation et de la demonstration
en mathématiques. Unpublished doctoral thesis, Université Joseph Fourier, Grenoble.
235

via free access
REFERENCES
Pedemonte, B. (2003). What kind of proof can be constructed following an abductive argumentation? In
M. A. Mariotti (Ed.), Proceedings of the third conference of the European Society for Research in
Mathematics Education (unpaginated). Retrieved from http://ermeweb.free.fr/CERME3/Groups/
TG4/TG4_Pedemonte_cerme3.pdf
Pedemonte, B. (2007). How can the relationship between argumentation and proof be analysed?
Educational Studies in Mathematics, 66, 23–41.
Peirce, C. S. (1867). On the natural classification of arguments. Presented 9 April 1867 to the American
Academy of Arts and Sciences. Proceedings of the American Academy of Arts and Sciences, 7,
261–287. (Compiled in Peirce, C. S., 1931–58, 2.461–516; Collected in Peirce, C. S., 1982-,
Document P31, Vol. 2, pp. 23–46).
Peirce, C. S. (1878, August). Deduction, induction, and hypothesis. Popular Science Monthly, 13, 470–482.
(Compiled in Peirce, C. S., 1931–58, 2.619–644; Collected in Peirce, C. S., 1982–, Vol. 3, pp. 323–338).
Peirce, C. S. (1902). Minute logic (an uncompleted book). (Compiled in Peirce, C. S., 1931–58, 2.1–202).
Peirce, C. S. (1903). Lectures on pragmatism. (Compiled in Peirce, C. S., 1931–58, 5.14–212).
Peirce, C. S. (1908, October). A neglected argument for the reality of God. The Hibbert Journal, 7, 90–112.
(Compiled in Peirce, C. S., 1931–58, 6.452–485).
Peirce, C. S. (1931–58). Collected papers. C. Hartshorne & P. Weiss (Eds.), Vols. 1–6, A. W. Burks
(Ed.), Vols. 7–8. Cambridge, MA: Harvard University Press. [references including “CP” indicate the
volume number and paragraph number in this work]
Peirce, C. S. (1955). Philosophical writings of Peirce (J. Bülchler, Ed.). New York: Dover.
Peirce, C. S. (1982–). Writings of Charles S. Peirce. A chronological edition. Bloomington, IN: Indiana
University Press. [references including “CE” indicate the volume and page in this work]
Perelman, C. (1970). Le champ de l’argumentation. Bruxelles: Presses Universitaires.
Piaget, J. (1928). Judgement and reasoning in the child (2002 printing). London: Routledge.
Pimm, D. (1987). Speaking mathematically. Communication in the mathematics classroom. London:
Routledge and Kegan Paul.
Pirie, S., & Kieren, T. (1992). Watching Sandy’s understanding grow. Journal of Mathematical
Behavior, 11(3), 243–257.
Polya, G. (1968). Mathematics and plausible reasoning (2nd ed.). Princeton, NJ: Princeton University Press.
Porteous, K. (1990). What do children really believe? Educational Studies in Mathematics, 21(6), 589–598.
Raman, M. (2002). Proof and justification in collegiate calculus. Unpublished doctoral dissertation,
University of California, Berkeley, CA.
Raman, M. (2003). Key Ideas: What are they and how can they help us understand how people view
proof? Educational Studies in Mathematics, 52(3), 319–325.
Recio, A., & Godino, J. (2001). Institutional and personal meanings of mathematical proof. Educational
Reid, D. (1995a). Proving to explain. In L. Meira & D. Carraher (Eds.), Proceedings of the nineteenth
pp. 137–143). Recife, Brazil.
Reid, D. (1995b). The need to prove. Unpublished doctoral dissertation, University of Alberta, Department
of Secondary Education.
Reid, D. (1996a). Describing reasoning. Poster presentation. Eighth International Congress on Mathematical
Education, Sevilla.
Reid, D. (1996b). The role of proving: Students and mathematicians. In M. de Villiers & F. Furinghetti
(Eds.), Proofs and proving: Why when and how? Proceedings of topic group 8, eighth international
congress on mathematical education. Cetralhil, South Africa: Association for Mathematics Education
of South Africa.
Reid, D. (1997). Constraints and opportunities in teaching proving. In E. Pehkonnen (Ed.), Proceedings
of the twentieth-first annual conference of the International Group for the Psychology of Mathematics
Reid, D. (2002a). Conjectures and refutations in grade 5 mathematics. Journal for Research in Mathematics
Education, 33(1), 5–29.
236

via free access
REFERENCES
Reid, D. (2002b). Elements in accepting an explanation. Journal of Mathematical Behavior, 20(4), 527–547.
Reid, D. (2003). Forms and uses of abduction. In M. A. Mariotti (Ed.), Proceedings of the third
conference of the European Society in Mathematics Education (unpaginated). Retrieved from
http://ermeweb.free.fr/CERME3/Groups/TG4/TG4_Reid_cerme3.pdf
Reid, D. (2005). The meaning of proof in mathematics education. In M. Bosch (Ed.), Proceedings of the
fourth congress of the European Society for Research in Mathematics Education (pp. 458–468).
Retrieved from http://ermeweb.free.fr/CERME4/CERME4_WG4.pdf
Reid, D., & Zack, V. (2009). Aspects of teaching proving in upper elementary school. In D. Stylianou,
M. Blanton, & E. Knuth (Eds.), Teaching and learning proof across the grades: A K-16 perspective
(pp. 133–146). New York: Routledge.
Reynolds, J. (1967). The development of the concept of proof in grammar school pupils. Unpublished
doctoral thesis, University of Nottingham, Nottingham, England.
Rota, G.-C. (1997). The phenomenology of mathematical beauty. Synthese, 111, 171–182.
Russell, B. (1937). The principles of mathematics (2nd ed.). New York: Norton. (First edition published
1903)
Schifter, D. (2009). Representation-based proofs in elementary grades. In D. Stylianou, M. Blanton, &
E. Knuth (Eds.), Teaching and learning proof across the grades: A K-16 perspective (pp. 71–86).
New York: Routledge.
Schoenfeld, A. (1989). Explorations of students’ mathematical beliefs and behavior. Journal for Research
in Mathematics Education, 20(4), 338–355.
Segal, J. (2000). Learning about mathematical proof: Conviction and validity. Journal of Mathematical
Behavior, 18, 191–210.
Sekiguchi, Y. (1991). An investigation on proofs and refutations in the mathematics classroom (Doctoral
dissertation, University of Georgia, 1991). Dissertation Abstracts International, 52, 03A.
Sekiguchi, Y., & Miyazaki, M. (2000). Argumentation and mathematical proof in japan. The Proof
Newsletter. Retrieved from http://www.lettredelapreuve.it/Newsletter/000102Theme/000102Theme
UK.html
Semadeni, Z. (1984). Action proofs in primary mathematics teaching and in teacher training. For the
Learning of Mathematics, 4(1), 32–34.
Senk, S. (1985). How well do students write geometry proofs? Mathematics Teacher, 78(6), 448–456.
Sfard, A. (1997). Commentary: On metaphorical roots of conceptual growth. In L. English (Ed.),
Mathematical reasoning: Analogies, metaphors and images (pp. 339–372). Mahwah, NJ: Lawrence
Erlbaum.
Simon, M. (1989). Intuitive understanding in geometry: The third leg. School Science and Mathematics,
89, 373–379.
Simon, M. (1996). Beyond inductive and deductive reasoning: The search for a sense of knowing.
Siu, M.-K. (1993). Proof and pedagogy in ancient China: Examples from Liu Hui’s Commentary on Jiu
Zhang Suan Shu. Educational Studies in Mathematics, 24(4), 345–357.
Smith, E., & Henderson, K. (1959). Proof. In H. Fawcett, A. Hach, C. Junge, H. Syer, H. van Engen, &
P. Jones (Eds.), The growth of mathematical ideas K-12 twenty-fourth yearbook (pp. 111–181).
Washington, DC: NCTM.
Smorynski, C. (2008). History of mathematics: A supplement. New York: Springer.
Sowder, L., & Harel, G. (1998). Types of students’ justifications. Mathematics Teacher, 91(8), 670–675.
Sowder, L., & Harel, G. (2003). Case studies of mathematics majors’ proof understanding, production,
and appreciation. Canadian Journal of Science, Mathematics and Technology Education, 3(2), 251–267.
Stanic, G. (1986). The growing crisis in mathematics education in the early twentieth century. Journal
for Research in Mathematics Education, 17(3), 190–205.
Steiner, M. (1978). Mathematical explanation. Philosophical Studies, 34, 135–151.
Stylianides, G., & Stylianides, A. (2008). Proof in school mathematics: Insights from psychological
research into students’ ability for deductive reasoning. Mathematics Thinking and Learning, 10(2),
103–133.
237

via free access
REFERENCES
Stylianides, G., & Stylianides, A. (2009). Facilitating the transition from empirical arguments to proof.
Stylianou, D., Blanton, M., & Knuth, E. (Eds.). (2009). Teaching and learning proof across the grades:
A K-16 perspective. New York: Routledge.
Tall, D. (1995). Cognitive development, representations & proof, justifying and proving in school
mathematics. Institute of Education, London, 27–38.
Third International Mathematics and Science Study [TIMSS]. (1998a). Data almanacs for achievement
items: Percent correct on the advanced mathematics items: Final year of secondary school.
Retrieved from http://timss.bc.edu/timss1995i/TIMSSPDF/pc_alm95/m12alm95.pdf
Third International Mathematics and Science Study [TIMSS]. (1998b). Data almanacs for achievement
items: Percent of responses by math item categories: Final year of secondary school. Retrieved from
http://timss.bc.edu/timss1995i/TIMSSPDF/dr_alm95/cism1295.pdf
Third International Mathematics and Science Study [TIMSS]. (1998c). TIMSS mathematics and science
items: Released set for population 3 (final year of secondary school), advanced mathematics items.
Retrieved from http://timss.bc.edu/timss1995i/TIMSSPDF/CitemAdM.pdf
Thurston, W. (1995). On proof and progress in mathematics. For the Learning of Mathematics, 15(1),
29–37. (Originally published in Bulletin of the American Mathematical Society, n.s., 30(2), 161–177)
Toulmin, S. E. (1958). The uses of argument. Cambridge: Cambridge University Press.
Uhlig, F. (2002). The role of proof in comprehending and teaching elementary linear algebra.
van Dormolen, J. (1977). Learning to understand what giving a proof really means. Educational Studies
in Mathematics, 8(1), 17–34.
Vinner, S. (1983). The notion of proof: Some aspects of students’ view at the senior high level. In
R. Hershkowitz (Ed.), Proceeding of the seventh international conference of the International Group
for the Psychology of Mathematics Education (pp. 289–294). Rehovot, Israel.
Volmink, J. D. (1990). The nature and role of proof in mathematics education. Pythagoras, 23, 7–10.
von Wright, G. H. (1965). The logical problem of induction. Oxford: Basil Blackwell.
Weber, K. (2002). Beyond proving and explaining: Proofs that justify the use of definitions and
axiomatic structures and proofs that illustrate technique. For the Learning of Mathematics, 22(3),
14–17.
Wheeler, D. (1990). Aspects of mathematical proof. Interchange, 21(1), 1–5.
Wilder, R. (1981). Mathematics as a cultural system. Oxford: Pergamon Press.
Williams, E. R. (1979). An investigation of senior high school students’ understanding of the nature of
mathematical proof. Unpublished doctoral dissertation. University of Alberta, Edmonton.
Wittmann, E. C., & Müller, G. (1988). Wann ist ein beweis ein beweis? In P. Bender (Ed.), Mathematik-
didaktik: Theorie und Praxis (pp. 237–257). Berlin: Cornelsen.
Wittmann, E. C., & Müller, G. (1990). When is a proof a proof? Bulletin de la Société Mathématique
Belge Ser. A., 42(1), 15–42.
Wood, T. (1999). Creating a context for argument in mathematics class young children’s concepts of
shape. Journal for Research in Mathematics Education, 30(2), 171–191.
Yackel, E. (1992). The evolution of second grade children’s understanding of what constitutes an
explanation in a mathematics class. Paper presented at the Seventh International Congress of
Mathematics Education, Quebec City.
Yackel, E. (2001). Explanation, justification and argumentation in mathematics classrooms. In M. van
den Heuvel-Panhuizen (Ed.), Proceedings of the twenty-fifth international conference on the
Psychology of Mathematics Education (Vol. 1, pp. 9–23). Utrecht.
Yackel, E., & Cobb, P. (1996). Sociomathematical norms, argumentation, and autonomy in mathematics.
Yackel, E., Cobb, P., & Wood, T. (1991). Small-group interactions as a source of learning opportunities
in second-grade mathematics. Journal for Research in Mathematics Education, 22, 390–408.
Zack, V. (1991). It was the worst of times: Learning about the Holocaust through literature. Language
Arts, 68(1), 42–48.
238

via free access
REFERENCES
Zack, V. (1997). “You have to prove us wrong”: Proof at the elementary school level. In E. Pehkonen
(Ed.), Proceedings of the twenty-first conference of the International Group for the Psychology of
Mathematics Education (Vol. 4, pp. 291–298). Lahti, Finland.
Zack, V. (1998, March 31–April 4). Coherence in five fifth grade boys’ argumentation about proof:
A sociolinguistic study of the role of repetition and logical structure in the boys’ talk. NCTM
Research Presession paper and symposium presentation, National Council of Teachers of Mathematics
(NCTM), 76th annual meeting, Washington DC.
Zack, V. (1999a). Everyday and mathematical language in children’s argumentation about proof.
Presented as part of Reform in the mathematics classroom: A focus on genre and discourse. Annual
meeting of the American Educational Research Association, Montréal.
Zack, V. (1999b). Everyday and mathematical language in children’s argumentation about proof. In
L. Burton (Ed.), Culture and the mathematics classroom. Special issue, Educational Review, 51(2),
129–146.
Zack, V. (2002). Learning from learners: Robust counterarguments in fifth graders’ talk about reasoning
and proving. In A. D. Cockburn & E. Nardi (Eds.), Proceedings of the twenty-sixth international
conference of the International Group for the Psychology of Mathematics Education (Vol. 4,
pp. 433–441). Norwich, UK: Program Committee.
Zack, V., & Graves, B. (2001). Making mathematical meaning through dialogue: “Once you think of it,
the z minus three seems pretty weird.” In C. Kieran, E. Forman, & A. Sfard (Eds.), Bridging the
individual and the social: Discursive approaches to research in mathematics education. Special
Issue, Educational Studies in Mathematics, 46(1–3), 229–271.
239

via free access
via free access
AUTHOR INDEX
A Chalmers, A., 204

Aigner, M., 76, 140 Chapiron, G., 169, 170
Alcock, L., 67 Chazan, D., 60, 62, 63, 71, 81, 96, 97
Alibert, D., 80 Chen, C., 80
Anglin, W., 3–6, 10, 16 Chiapini, G., 174, 212
Arnauld, A., 73–75 Cifarelli, V., 108
Arsac, G., 4, 160, 169–171
Cobb, P., 162, 163
Arzarello, F., 101, 108, 109 Coe, R., 60, 68
Colonna, A., 169, 170
B
Balacheff, N., 25, 31–33, 38, 46–49,
53–55, 60, 68, 72, 90, 129, 131, D
132, 134, 142, 145, 153, 157, Davis, P.J., 9, 16, 21, 22, 24, 26, 27,
158, 163, 170, 171, 225. See also 30, 73, 75, 78
Arsac, G.; Herbst, P.; Lakatos, I. de Villiers, M., 50, 74, 76–79, 81,
Barbeau, E., 4, 6, 10, 20 82, 127, 224. See also
Barbin, E., 73, 74 Mudaly, V.
Barkai, R., 60, 64, 68–70 Descartes, R., 6–8, 74
Barnes, J., 89 Dhombres J., 121
Bartolini Bussi, M., 49, 50, 54, 160, Douek, N., 26, 30, 158–160, 163,
218 218. See also Boero, P.
Bell, A., 60, 68, 69, 71, 74, 76, 131, Dreyfus, T., 60, 64, 68–70
134, 142–145, 151, 216 Duval, R., 30, 35, 44–46, 53, 54, 74,
Benis-Sinaceur, H., 110 75, 90, 153, 155–160, 163, 164,
Blanton, M., 54, 85 212. See also Egret, M.-A.
Blum, W., 35, 43, 44, 129, 137, 138,
142, 145 E
Boero, P., 49, 50, 54, 108, 109, 159, Eco, U., 101–104, 107, 109, 126
160, 163, 173–175, 212, 218. See Egret, M.-A., 212
also Garuti, R.; Mariotti, M.A.
English, L., 110
Bonfantini, M., 101
Ernest, P., 37, 39, 40, 48
Borel, J.M., 44, 45
Brandt, B., 191
Branford, B., 42, 44, 129, 138, 142, F
146 Fawcett, H., 32, 40–43, 48, 52–54,
Brown, J.R., 13 81, 82, 165–169, 171, 172,
Bruner, J.S., 123 176, 219
Burton, L., 30, 54, 74, 194 Ferri, F., 49, 50, 54
Fischbein, E., 28, 39, 40, 53, 60, 63,
C 71, 72, 74, 128, 152, 157
Cabassut, R., 109, 164 Freudenthal, H., 135, 148, 214, 215
Carroll, L., 84 Fuller, E., xiv, 54
241

via free access
AUTHOR INDEX
G Jones, R.B., 3, 10
Galbraith, P.L., 60, 63, 64, 68 Joseph, G.G., 15, 16
Galotti, K., 151
Garuti, R., 49, 50, 54, 108, 109, 159, K
160, 173–175, 212 Kant, I., 7, 8, 37, 145, 146
Germain, G., 160, 169, 170 Kedem, I., 39, 53, 60, 63, 71, 72, 74
Godino, J., 26, 27, 60, 69 Kelly, D.L., 68
González, G., 80 Kieren, T., 194
Graves, B., 173 Kirsch, A., 35, 43, 44, 129, 137, 142,
Guichard, Y., 169, 170 145
Kleiner, I., 3, 4, 7
H Kline, M., 3, 4, 7–9, 15
Hanna, G., 4, 6, 10, 20, 27, 28, 35, Knipping, C., xiii, 31, 104, 145, 164,
50–54, 78, 81, 121–123 179–184, 187, 191, 192
Hannaford, C., 4 Knorr, W., 20
Harel, G., xiv, 31, 51–54, 60, 61, 63, Knuth, E., 54, 60, 61, 63, 65, 66, 79,
65, 66, 72, 123, 124, 128, 133, 80, 85, 140. See also Stylianou, D.
135, 146–151, 178. See also Komatsu, L., 151
Martin, G.; Sowder, L. Krummheuer, G., 109, 161–163,
Healy, C., 167–168 179, 191, 192, 218
Healy, L., 29, 30, 53, 54, 59–67, 69, Küchemann, D., 27, 28
71, 80, 139, 140, 223
Heath, T.L., 4, 5, 11, 17–19 L
Henderson, K., 90, 92, 108, 110, Lakatos, I., 9, 22–24, 26, 38, 39, 46,
123, 131–133, 150, 151 47, 76, 78, 94, 160, 171, 193,
Herbst, P., 14, 33, 80, 139, 165, 166, 198, 200–202, 204–206, 219
178, 191, 212, 216 Lakoff, G., 128, 164, 224
Hersh, R., 9, 16, 21, 22, 24, 26, 27, Lampert, M., 171–173
30, 73, 75, 78, 81, 158. See also Lemut, E., 108, 109, 159, 160,
Davis, P.J. 173–175, 212
Hilbert, D., 7, 8, 18 Lovell, K., 60
Hill, J., 216
Hofstadter, D., 127, 224 M
Hoyles, C., 27–30, 53, 54, 59–67, MacKernan, J., 30
69, 71, 80, 139, 140, 223. See Maher, C., 133, 173
also Healy, L. Manin, Y., 30
Mante, M., 160, 169–171
J Mariotti, M.A., 35, 49, 50, 52–54,
Jahnke, H.N., 35, 50, 51, 54, 78, 109, 159, 160, 173, 174, 212,
121, 122, 215. See also 218, 219. See also Boero, P.
Hanna, G. Martin, G., 60, 61, 63, 65, 66
James, T., 169, 188–190 Martin, M.O., 68
Johnson, M., 128, 164, 224 Martin, T., 80
Johnson, W.E., 91 Martino, A., 133, 173
Jones, F.B., 168 Martzloff, J.-C., 12–15, 18–20
242

via free access
AUTHOR INDEX
Mason, J., 30, 74, 101, 102, 108, 128, 147, 168, 172, 173, 201–203,
134, 194 208
McCrone, S., 80 Reynolds, J., 59–61, 69, 71, 216,
Micheletti, C., 101, 108, 109 223
Mikami, Y., 11 Robutti, O., 101, 108, 109
Mill, J. S., 89, 91 Rota, G.-C., 77
Miyazaki, M., 164, 223 Rouse, W., 216
Monks, K., 142 Russell, B., 7, 8, 10, 16
Moore, R., 9, 168 Ruthven, K., 60, 68
Morgan, C., 54
Movshovits-Hadar, N., 212 S
Mudaly, V., 81, 82, 224 Sáenz-Ludlow, A., 108
Müller, G., 43, 129, 142 Schifter, D., 137
Schoenfeld, A., 63
N Segal, J., 140
Nakayama, O., 164 Sekiguchi, Y., 164, 191, 223
Needham, J., 11
Semadeni, Z., 43, 44, 135, 136
Netz, R., 19, 20, 213, 214
Senk, S., 30, 59, 60, 69
Nicole, P., 73–75, 81
Sfard, A., 110
O Sibilla, A., 174, 212
Oliveira e Silva, T., 94 Simon, M., 124, 125, 130, 147
Olivero, F., 101, 108, 109 Siu, M.-K., 12, 13
Smith, E., 90, 92, 108, 110, 123,
P 131–133, 150, 151
Pedemonte, B., 109, 160, 163, 179, Smorynski, C., 11
218 Sowder, L., 31, 51–54, 72, 123,
Peirce, C.S., 84, 87, 90, 100–109, 124, 128, 133, 135, 146–151,
126, 225 178. See also Harel, G.
Perelman, C., 154, 155, 157, 161, Stacey, K., 30, 74, 194
163 Stanic, G., 169
Piaget, J., 127 Steiner, M., 75
Pimm, D., 54, 134 Stylianides, A., 85, 127, 151, 218
Pirie, S., 194 Stylianides, G., 85, 127, 151, 218
Polya, G., 90–92, 95, 108, 110, 113, Stylianou, D., 54, 85
115, 118, 201
Porteous, K., 60, 64, 67, 68
T
Proni, G., 101
Tall, D., 124, 130, 137, 139, 141
R Thurston, W., 75, 77, 213, 214
Raman, M., 31 Tirosh, D., 60, 64, 68–70
Recio, A., 26, 27, 60, 69 Toulmin, S.E., 154, 158, 161–164,
Reid, D., xiii, 25, 31, 52, 53, 57, 179–180, 191
58, 76, 80, 83, 87, 91, 95, 98, Tsamir, P., 60, 64, 68–70
105, 111, 115, 116, 123, 124, 127, Tyack, D., 169
243

via free access
AUTHOR INDEX
U Wheeler, D., 80
Uhlig, F., 28, 29 Wilder, R., 3
Williams, E.R., 60, 61, 63, 71, 216
V Wittmann, E.C., 43, 129, 142
van Dormolen, J., 142, 144 Wood, T., 162, 163
Vinner, S., 67
Voelz, S., 151 Y
Volmink, J.D., 30, 77 Yackel, E., 162, 163
von Wright, G.H., 89–91
Z
W Zack, V., 85, 165, 172, 173, 175,
Weber, K., 109 176, 203, 218, 219. See also
Weiss, M., 80 Reid, D.
Wesson, J., 216 Ziegler, G., 76, 140
244

via free access
SUBJECT INDEX
A centre of gravity, 31, 32, 121, 122

a priorism (philosophy of chains of deductions, 27, 50, 85,
mathematics), 37–40, 43, 45 126, 150, 215
abductive reasoning, 57, 83, 84, Chinese mathematics, 4, 8, 10–15,
99–109, 123, 127, 186, 199, 200, 18, 20, 24
218, 224, 225 Chino and Tressie (David’s sister’s
absolute certainty, 43, 79 dogs), 83, 84
abstraction, 4, 83, 112, 117, 120, 151 cognitive unity, 108, 157, 159–161,
action proofs, 43, 135–137 212, 225
aesthetics, 74, 77, 82, 222 collective argumentation (Knipping,
algebraic form, 66, 108 Krummheuer), 31, 161, 191
algebraic manipulations, 124, 168 colligation (pattern observing), 91,
anschaulich, 43, 129, 145, 146 126
Anschauung, 145, 146 common notions (assumptions), 5,
antecedent, 23, 85, 150, 151, 212 16–18, 20
Aquinas, St. Thomas, 5, 8 completeness, 9, 78, 141
Arabic mathematics, 10, 11 condition (definition of ), 83
Aristotle, 4, 6, 8, 18, 88, 89 Confucian Analects, 14
Arithmagon puzzles, 194, 198 conjecturing, 92, 94, 95, 98, 103,
arithmetic, 7, 21, 69, 78, 108, 199 108, 109, 126, 159, 160, 163,
assessments, large scale written, 71, 172, 173, 175, 194, 196, 197,
223 199–202, 205, 220
axiomatic structure, 5, 109 consequent, 14, 23, 85, 103, 150,
axiomatisation, 7, 76 151, 212
axioms, 4–7, 9, 10, 15, 16, 20, 22–24, consistency, 7, 9, 78
27, 36–40, 51, 76, 78, 98, 109, content, 21, 35, 39, 45, 46, 51, 80,
149, 168, 170, 211, 214, 215, 222 155, 159, 165, 167, 169, 172,
176, 213, 214, 218, 219
B contrapositive, 150, 151
back-to-the-basics, 216 converse, 61, 65
backing (of an argument), 164, 180, convincing an enemy, 30, 74
183 convincing arguments, 30, 31, 33,
Beweis (German word for proof ), 52, 146
33, 146 correct proofs, 57, 59, 65, 68–70,
blindness, sudden, 102 221
Bourbaki, 10, 46 Count the Squares problem, 85, 86,
Build-a-Book geometry, 167 94, 95, 103, 112, 114, 201–203,
207
C counterexample, 22–26, 38, 46–48,
calculus, 7, 9, 21, 78 57, 59, 62–64, 69, 70, 72, 79,
Carroll, Lewis, 84 81, 89, 93, 98, 131, 133, 134,
245

via free access
SUBJECT INDEX
143–145, 150, 163, 198–202, empirical arguments, 57, 59, 61, 62,
204–207, 223, 224 65, 67, 68, 96, 130, 131, 133,
global (Lakatos), 22, 23, 200 134, 142, 143, 150, 151, 159,
local (Lakatos), 22, 23, 200 190, 217
crucial experiment (type of Empirical enactive proofs, 142
argument), 68, 129–132, 143–145 epistemic value, 74, 75, 91, 104,
curriculum, 10, 68, 71, 72, 79, 124, 199, 204, 207
141, 168, 169, 172, 176, 216–218, epistemologies of proof, 53, 54
223 Euclid, 3–11, 13, 15–24, 42, 73, 75,
165, 213
D myth, 16, 18, 24
debate (approach to teaching proof ), postulates, 7
165, 169–173, 219, 222 proof of the infinitude of primes,
deductive method, 4, 6, 10, 16, 18, 135
37–39, 47, 48 Euclidean geometry, 7, 21, 125, 139,
deductive system, 76, 215 149, 168
definitions, 5, 6, 10, 16, 18, 20–24, Euclidean methodology, 22–24
37–39, 41, 49, 75, 76, 78, 109, Euler, Leonhard, 115, 118
149, 154, 158, 159, 162, 163, examples
166, 168, 214, 219, 222 non-representational, 130, 131,
démonstration (French word for 133, 143
mathematical proof ), 32, 33, representational, 131, 133, 135,
44, 47–49, 129, 156, 157, 159, 143
169 exception barring (Lakatos), 200,
demonstration (to mean 205–207
mathematical proof ), 6, 32, 33, existential statements, 70, 71
41, 74, 166 experimental proofs, 129
demonstrative reasoning (Polya), 110
explanatory proofs, 28, 51, 75
Descartes, René, 6–8, 74
diagrams, 13, 14, 17–21, 24, 40–42, extending a pattern (type of
63, 66, 80, 104–106, 113, 115, Empirical argument), 130, 131,
137, 139, 181, 182, 184, 186, 143, 150
187, 205 extrapolation (Bell’s terms for
discourse, 12, 15, 18, 30–34, 37, 41, simple enumeration), 131, 144
43, 47–54, 77, 139, 154, 158,
163, 176, 180, 181, 185, 187, F
218 fallibilism (philosophy of
Ducrot, Oswald, 154 mathematics), 37, 38, 40, 43, 45,
48, 50–52
E familiarity of the methods used in a
eduction (predicting), 91, 126 proof, 66
Egyptian mathematics, 3, 4, 10, 11, first-order logic, 21, 22
15 flawed deductive proofs, 57, 59,
Elements (Euclid’s), 4–6, 8–11, 13, 64–65
16–19, 22, 42, 165, 213 Football field problem, 103, 107
246

via free access
SUBJECT INDEX
form (of a proof or argument), 29–33, H

35, 40, 45–47, 51, 65, 66, 129–131, Handshake problem, 103
141–143, 146, 157, 159, 191 heuristic argumentation (Duval),
form, logical, 9, 14, 100, 102–104 156, 157
formal language, 9, 21, 22, 26, 27, Hilbert, David, 7, 18
153 Hindu mathematics, 15, 18
formal proof, 9, 26, 27, 39, 40, 44, history of mathematics, 1, 3, 8, 10–11,
51, 141–143, 149, 157, 159, 212 15, 22, 24, 76, 98, 121, 160
formal proofs (Lakatos), 9
formalism (philosophy of I
mathematics), 7–10, 20, 22, 36, indirect proofs, 148, 150, 151
40, 43, 46, 52, 58, 98, 129, 130 induction, problem of, 91
formality, 8, 9, 21, 37, 40, 41, 57, inductive reasoning, 79, 83, 88–92,
58, 127, 151, 208, 211–213, 220 94, 96–99, 103, 122, 123, 126,
formulaic proofs, 58, 80, 124 127, 151, 201, 224
formulated reasoning, 4, 42, 57–58, infinitude of primes, 19, 135, 140,
87, 112, 147 143, 151
Foundations of Geometry (Hilbert), inhaltlich-anschaulich proofs, 129,
18 145, 146
foundations of mathematics, 9, 10, inquiry mathematics approach, 191,
18, 21, 26, 27, 39 222–223
Frege, Gottlob, 7, 10 intellectual proofs, 135, 145
intuition (Fischbein), 40, 128, 212
G intuitional proofs, 43, 138, 146
Gauss, Carl Friedrich, 7, 149 intuitionism (philosophy of
generic example, 20, 65, 66, 76, 129, mathematics), 36
131, 135–137, 144, 145, 148, 167
generic figure, 138, 139 J
geometric arguments, 131, 143 justification, 5, 7, 12, 18, 38, 42, 50,
geometry 64, 70, 74, 133, 135, 143, 155,
dynamic, 62, 81, 188–190 160, 162, 173, 181–184, 187,
Euclidean, 7, 21, 125, 139, 148, 212–215
149, 168
non-Euclidean, 7, 37, 76, 149 K
global argument, 180, 184 Kant, Immanuel, 7, 37, 145, 146
global counterexample (Lakatos), Kepler, Johannes, 78
22, 23, 200 kinds or types (type of Empirical
global organisation (Freudenthal), argument), 131, 132, 143, 144
214, 215 Klein, Felix, 3, 4, 7
Gödel, Kurt, 9
Goldbach conjecture, 94, 131, 143 L
Greek mathematics, 3, 4, 7, 9–11, Leibniz, Gottfried, 7, 9
18–20, 83, 84, 213, 214 Liu Hui, 12–18
guessing, 106, 108, 195, 197, 203 local counterexample (Lakatos), 22,
guilty lemma (Lakatos), 23, 200, 201 23, 200
247

via free access
SUBJECT INDEX
local organisation (Freudenthal), narrative reasoning (Bruner), 123

148, 214–215 necessary inference (deduction), 150
logic, 1, 4–7, 9, 13–15, 26, 27, 31, negation, 150, 151, 190, 202, 204,
37–41, 44, 45, 49–51, 57, 59, 65, 205
73, 74, 79, 80, 82–84, 89, 91, 98, New Math, 10, 215–217, 220
100, 102–104, 109, 115, 121, Newton, Isaac, 5, 7, 78, 100
124, 127, 138, 139, 141, 147, Newton’s law of gravity, 32, 78,
150, 154, 156, 158, 160, 163, 121, 122
164, 171, 185, 221 Newtonian physics, 100
deductive, 50, 170
first-order, 21, 22 O
symbolic, 10, 84, 85 odd numbers, 76, 86, 87, 92, 95, 98,
logical coherence, 57, 59 111, 112, 116, 125, 126, 206
logical form, 9, 14, 100, 102–104 ostension (Balacheff ), 145
logical thinking, 79, 80, 82, 221 Othello, 25
logical validity, 49
logicism (philosophy of P
mathematics), 7 parallel arguments, 121, 180, 181,
184, 187–190, 192
M parallel postulate, 7
manipulative proofs, 124, 141, 143 particular proofs (employing generic
mathematical emotional orientation, examples), 66
128 pattern observing (type of inductive
mathematical induction. See reasoning), 91, 92, 94, 95, 126,
reasoning by recurrence 202, 204–206
mechanical reasoning, 9, 21, 58, 124, Peano, Giuseppe, 7, 8, 99
141 Peirce, Charles Sanders, 84, 87, 90,
Mesopotamian mathematics, 3, 11, 15 100–109, 126, 225
metamathematics, 21, 26 Perelman, Chaïm, 154, 155, 157,
metaphor, 87, 110, 128, 158, 164, 189 161, 163
method of proof, 39, 167 perpendicular bisector, 119–121
Mill, John Stuart, 89, 91 philosophy of mathematics, 36, 37,
modus ponens, 31, 85, 126, 150 39–46, 48–52
modus tollens, 85, 126 physical manipulation, 16, 18
monster barring (Lakatos), 193, 201, Piaget, Jean, 127
202, 204–208 Plato, 4, 18, 36
Moore method, 168 Platonism (philosophy of
multiple perspectives, xiv, 55, 178 mathematics), 36
plausible conjecture (epistemic
N value), 207
naive conjecture (Lakatos), 23 plausible reasoning, 45, 90, 207
naïve empiricism (Balacheff ), 68, Popper’s falsificationist philosophy
129, 131, 144, 145 of science, 204
narrative proofs, 66, 123, 131, 140, Posterior Analytics (Aristotle), 4, 6,
143, 144 8, 89
248

via free access
SUBJECT INDEX
postulates, 5, 7, 9, 16–18, 20, 22, 36, proofs and refutations (Lakatos’s

39, 42–44, 49, 133 cycle of ), 38, 39, 46, 47, 200
pragmatic arguments, 27, 90, 145 proofs by contradiction, 69, 78, 133,
pragmatic methods, 27, 28 135, 148, 150, 151, 202
pragmatic reasoning (Balacheff ), 21, proofs by counterexample, 78, 145
90 proofs by exhaustion, 69, 89, 93,
predicting (type of inductive 131, 133, 143–145, 148
reasoning), 91, 92, 94, 126, 206, proofs that prove (Hanna), 51
207 proofs
premise (of an argument), 35, 36, 38, formal, 40, 51, 141–143, 149,
43, 79, 87, 90, 102, 104, 106, 157, 159
129, 130, 139, 154 formal (Lakatos), 9
preuve (French for proof), 32, 33, pre-formal (Lakatos), 9
46, 47, 49, 157 preformal (preformalists), 9, 35,
prime numbers, 20, 131, 134, 135, 39, 43, 44, 52, 53, 129, 130, 142,
140, 143, 149, 151 144–146, 151, 209, 211, 213,
probare (Latin root of prove), 25 214, 220
Problem of the Week tasks, 172, 219 semi-formal, 9, 22, 26, 27, 37,
problem solving, 57, 80, 95, 164, 39, 41, 43, 45, 49, 50, 52, 58,
172, 174, 175, 219 209, 211–214, 220
process of generation of proving discourse, 31, 37, 180, 185,
conditionality (PGC), 109 187
professional mathematics, 27, 30, Pythagorean theorem, 12, 13, 103,
47, 82, 109, 123, 129, 221–222 104, 113, 136, 181, 186, 189
proof schemes, 31, 51, 146–151,
178 Q
analytical, 52, 146, 147 qualified reader, 21, 24
authoritarian, 123, 146 quasi-empiricist (philosophy of
contextual, 147–149 mathematics), 38, 46–48
generic, 135, 148, 149
interiorised, 147 R
perceptual, 130, 131, 133, 143, reality oriented proofs, 137
147, 149 reasoning by analogy, 57, 83, 84, 99,
ritual, 124, 146 110–123, 125–128, 147, 218,
symbolic, 124, 146 224, 225
transformational, 52, 147–149 reasoning by recurrence,
proof writing, 71, 175, 223 mathematical induction (MI), 76,
proof-analysis, Lakatos’s process of, 98, 99, 126, 147, 150
22–24, 26, 38, 39, 78, 193, 200, reasoning from the known to the
201 unknown, 90
proof-generated concept, 23, 200 reasoning
proof-texts, 29–31, 33, 34, 37, 39, abductive (see abductive
41, 43–46, 52–54, 117, 118, 120, reasoning)
121, 146, 147, 150, 176, 177, avoiding, 57, 123–126, 146,
193, 212, 222 150
249

via free access
SUBJECT INDEX
inductive (see inductive social-constructivist (philosophy of

reasoning) mathematics), 39, 46, 48–52
unformulated, 57, 197, 200, 202 sociomathematical norms, 162
reductio ad absurdum, 133, 150–151 specialisation (type of deductive
reference theory, 49, 50, 219 reasoning), 85, 93, 96, 98, 107,
reference to authority (to verify), 57, 113, 114, 122, 126, 205
123–124, 171 specific cases, 16, 88, 89, 91–93,
refutation, 38, 39, 46, 47, 57, 59, 99–101, 113
63–64, 133, 171, 181, 183–185, square numbers, 78, 85, 219
188, 189, 200 standard view (of the history of
representation-based proofs, 137 proof ), 3–11, 16, 19–24
representations, 44, 45, 126, 130–143, students
146, 151, 155, 174, 177, 212 elementary school, 59, 64–66, 70,
retransmission of falsity (Lakatos), 85, 218
22–24, 38, 39, 48 secondary school, 59, 60, 62, 64,
rhetorical argumentation (Duval), 65, 67, 68, 80, 81
156, 157 university, 51, 59, 60, 66, 80,
rhombus, 104, 105, 133, 143, 183, 178
186–189 syllogisms, 14, 84, 85, 100–102,
rigour, 7–10, 15, 16, 20, 24, 39, 124, 104, 109
138, 159 symbolic proofs, 66, 124, 139–140,
ritualistic arguments, 66, 67 143, 146
rules of inference, 6, 36, 37, 41, symbols, 7–10, 21, 27, 84, 124, 130,
49–51 131, 138–141, 143, 147, 196
rules, general, 84, 85, 88, 89, 98, 99, systematisation, 35, 74, 76, 79, 80
101, 102, 106–109, 150, 186,
198, 204, 207 T
Russell, Bertrand, 7, 10, 16 teacher-game, 80, 81
teachers
S future elementary school, 59,
Schorle, 137, 138 64
scientific proofs, (preformalist inservice secondary school, 59
definition), 129, 138 inservice upper elementary
semi-formal proofs, 9, 22, 26, 27, school, 59
37, 39, 41, 43–45, 49, 50, 58, tetrahedron, 119–122
209, 211, 213, 214, 220 textbooks, 5, 13, 14, 19, 22, 96, 97,
set theory, 7, 9, 78 123, 146, 164, 165, 167, 168,
simple enumeration (type of 172, 214
Empirical argument), 130, 131, theorem credits (Thurston), 74, 77
143, 144, 150 thought-experiment, 9, 22, 129, 144,
social discourse, 30, 33, 37, 41, 43, 145, 174, 200
49 TIMSS, 59, 69, 71, 223
social norms, 162 tool-box (Netz), 211, 213–214, 219,
social proof, 27 220, 222
250

via free access
SUBJECT INDEX
Toulmin, Stephen, 154, 158, 161–164, V

179, 180, 191 valid mathematical proofs, 43, 64
transfer (of reasoning to non- validity, 26, 27, 36, 39, 41, 44, 45,
mathematical domains), 43, 166, 49–51, 53, 61, 62, 79, 108, 114,
169 145, 159, 166, 170, 198
transformational reasoning (Simon), verification
124–127, 147 deductive, 66, 118, 127, 218
triangle angle sum, 42, 139, 143, empirical, 66
149 visual arguments, 20
truth value, 35, 45, 70 visual proofs, 13, 136, 137
two-column proofs, 139, 143, 149, visual representations, 139
167, 191, 216, 217, 222
W
U warrant of an argument, 37, 162–164,
universal statements, 64, 68, 70 179–181, 184, 185, 187–191
universal validity, 39, 41, 166 Whitehead, Alfred North, 7, 8
251

via free access

58-Reid &amp; Knipping (2010)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

58-Reid &amp; Knipping (2010)

Uploaded by

Copyright:

Available Formats

Proof in Mathematics Education

David A. Reid and Christine Knipping - 978-94-6091-246-7

David A. Reid and Christine Knipping - 978-94-6091-246-7

ISBN: 978-94-6091-244-3 (paperback)

Published by: Sense Publishers,

Printed on acid-free paper

All Rights Reserved © 2010 Sense Publishers

David A. Reid and Christine Knipping - 978-94-6091-246-7

Part I: What is Proof?

1. History of Proof ................................................................................................... 3

2. Usages of “Proof ” and “Proving” ..................................................................... 25

Part 2: Important Research Foci, Past and Present

5. The Role of Proof ............................................................................................. 73

David A. Reid and Christine Knipping - 978-94-6091-246-7

What is Proving in School Mathematics? ...................................................... 79

6. Types of Reasoning ......................................................................................... 83

7. Classifying Proofs and Arguments ................................................................ 129

9. Teaching Experiments ................................................................................... 165

Part 3: Processes of Reasoning and Argumentation

10. Argumentation Structures.............................................................................. 179

David A. Reid and Christine Knipping - 978-94-6091-246-7

11. Patterns of Reasoning .................................................................................... 193

12. Implications for Teaching...............................................................................211

13. Directions for Research ................................................................................. 221

Author Index ........................................................................................................ 241

Subject Index........................................................................................................ 245

David A. Reid and Christine Knipping - 978-94-6091-246-7

Figure 1: A misleading diagram showing that 8×8 = 5×13 ................................. 14

David A. Reid and Christine Knipping - 978-94-6091-246-7

David A. Reid and Christine Knipping - 978-94-6091-246-7

Proof 1: Elements Book I Proposition 1............................................................... 5

David A. Reid and Christine Knipping - 978-94-6091-246-7

Table 1: References in Euclid’s proof of Prop. I.1 ............................................... 6

David A. Reid and Christine Knipping - 978-94-6091-246-7

David A. Reid and Christine Knipping - 978-94-6091-246-7

reveals differences in the role of argumentation in different contexts. The second is

David A. Reid and Christine Knipping - 978-94-6091-246-7

David A. Reid and Christine Knipping - 978-94-6091-246-7

Before embarking on a discussion of proof and proving in mathematics education,

THE STANDARD VIEW

The First Proofs

David A. Reid and Christine Knipping - 978-94-6091-246-7

The Legacy of Logic

David A. Reid and Christine Knipping - 978-94-6091-246-7

Proof 1: Elements Book I Proposition 1

5 With centre A and distance AB let the

Now, since the point A is the centre of

Heath, 1956, Vol. 1, pp. 241–242, line numbers adjusted

David A. Reid and Christine Knipping - 978-94-6091-246-7

Table 1. References in Euclid’s proof of Prop. I.1

Euclid’s axiomatic approach is not only of historical importance. It has become

David A. Reid and Christine Knipping - 978-94-6091-246-7

The Restoration of Rigour

David A. Reid and Christine Knipping - 978-94-6091-246-7

symbols is a permissible operation. Thus formulas are derived by applying

Table 2. Chronology of people and publications mentioned in the text

600 BCE Thales, first historically recorded proofs

David A. Reid and Christine Knipping - 978-94-6091-246-7

David A. Reid and Christine Knipping - 978-94-6091-246-7

OTHER VIEWS OF THE HISTORY OF PROOF

David A. Reid and Christine Knipping - 978-94-6091-246-7

David A. Reid and Christine Knipping - 978-94-6091-246-7

Compute the Xian from the Gou

[Find c from a and b (using the

58-Reid & Knipping (2010)

58-Reid & Knipping (2010)