Educ Stud Math

DOI 10.1007/s10649-017-9761-8

A conceptual model of mathematical reasoning

for school mathematics

Doris Jeannotte 1 & Carolyn Kieran

1

# Springer Science+Business Media Dordrecht 2017

Abstract The development of students’ mathematical reasoning (MR) is a goal of several

curricula and an essential element of the culture of the mathematics education research
community. But what mathematical reasoning consists of is not always clear; it is generally
assumed that everyone has a sense of what it is. Wanting to clarify the elements of MR, this
research project aimed to qualify it from a theoretical perspective, with an elaboration that
would not only indicate its ways of being thought about and espoused but also serve as a tool
for reflection and thereby contribute to the further evolution of the cultures of the teaching and
research communities in mathematics education. To achieve such an elaboration, a literature
search based on anasynthesis (Legendre, 2005) was undertaken. From the analysis of the
mathematics education research literature on MR and taking a commognitive perspective
(Sfard, 2008), the synthesis that was carried out led to conceptualizing a model of mathemat-
ical reasoning. This model, which is herein described, is constituted of two main aspects: a
structural aspect and a process aspect, both of which are needed to capture the central
characteristics of MR.

Keywords Mathematical reasoning . Theoretical model . Commognition . School mathematics .

Anasynthesis . Structural aspect of mathematical reasoning . Process aspect of mathematical
reasoning

1 Introduction: the conceptual blur regarding mathematical reasoning

What is mathematical reasoning? How might we characterize it? These questions underpinned
the development of the model of mathematical reasoning (MR) (Jeannotte, 2015) that is the
focus of this article. While curricular documents around the world emphasize the fostering of
students’ MR as an important goal (e.g., NCTM, 2000; OCDE, 2006), the way in which MR is

* Doris Jeannotte
jeannotte.doris@uqam.ca

1
Département de mathématiques, Université du Québec à Montréal, Montréal, QC, Canada
 Jeannotte D., Kieran C.

described within these documents tends to be vague, unsystematic, and even contradictory
from one document to the other. As well, within the mathematics education research commu-
nity, the discourse on MR is not monolithic; it does not consist of a single voice.
Various visions of mathematics, and of teaching and learning, confront each other. An
additional factor adds to the confusion, as pointed out by Yackel and Hanna (2003):
“Writing about reasoning in mathematics is complicated by the fact that the term
reasoning, like understanding, is widely used with the implicit assumption that there
is universal agreement on its meaning” (p. 228). But, as an in-depth search of the
research literature makes amply clear, this assumption does not hold.
At the extreme end of the spectrum is that Bmost mathematicians and mathematics
educators use this term [mathematical reasoning] without any clarification or elaboration^
(Yackel & Hanna, 2003, p. 228). Nevertheless, among those who do define MR, various
aspects are stressed by various authors. Arsac (1996) and Cabassut (2005) emphasize its
double nature (product vs. process); Arsac (1996) and Lithner (2008), its function of producing
new knowledge; and Duval (1995), its function of changing the epistemic1 value of a certain
proposition. Definitions of the classic forms of MR, such as deductive, inductive, and
abductive, are all to be found but with various emphases. Duval (1995) insists that
only deductive reasoning can be considered mathematical, while Reid (2003), Rivera
(2008), and Meyer (2010) point to the importance of abductive reasoning in mathe-
matical discovery. In contrast to the structural aspect related to the form of MR, the
literature also discloses characterizations of MR that stress its underlying processes,
with inferencing being at the heart of these processes: conjecturing (Mason, 1982;
Stylianides, 2008), generalizing (Stylianides, 2008), exemplifying (Mason, 1982),
proving (Duval, 1995; Stylianides, 2008), arguing (Pedemonte, 2002), and convincing
(Cabassut, 2005). While the process aspect is well represented in the literature, it
remains relatively unexplored from an epistemological perspective.
These few highlights from the research literature on MR suggest that this area is one that
could benefit greatly from an attempt at coherent conceptualization. What Steen said in 1999 is
still true today:
[MR sometimes] denotes the distinctively mathematical methodology of axiomatic
reasoning, logical deduction, and formal inference. Other times it signals a much broader
quantitative and geometric craft that blends analysis and intuition with reasoning and
inference, both rigorous and suggestive. This ambiguity confounds any analysis and
leaves room for many questions. (Steen, 1999, p. 270)
The current state of the field renders difficult any comparison of not only the various
approaches to, and characterizations of, MR but also the results of related studies. The
importance of developing MR in the teaching and learning of mathematics at the different
levels of schooling, as well as the need for substantive conceptual resources on MR in the
training and professional development of teachers, also serve to motivate a deeper and more
theoretically-sound study of MR.
Thus, the objective of this research is to elaborate a conceptual model of mathematical
reasoning for the teaching and learning of primary and secondary levels of school mathematics
(see also Jeannotte, 2015). The article Ba^ is here very important since it implies that it is A
model among other possibilities, in line with Sfard’s (2012) principle of multivocality. As will

1
The epistemic value of a proposition refers to the notion that an utterance can be true, probable, likely, or false.
A conceptual model of mathematical reasoning

be stated in the next section, our epistemological stance, which is central to the whole process,
goes hand-in-hand with the adopted theoretical frame, which in turn plays the role of an
interpretative lens for drawing out key features of MR from the research literature. Following
the presentation of the theoretical framework, we describe the anasynthesis methodology
(Legendre, 2005), which provided a rigorous tool for searching the mathematics education
research literature. Then we present an overview of some of these key features of MR that were
extracted from the literature, features that co-constituted the emerging conceptualization of a
model of MR for school mathematics. It must be emphasized, however, that the result is not a
unique model that unifies all of the divergences found in the literature. It is rather a model that
systematizes the various converging features of MR within a theoretically coherent frame. As
argued by Balacheff (2008, p. 501):
The scientific challenge of research in mathematics education is […] to shape a body of
knowledge which should be robust (which means theoretically valid) and relevant
(which means instrumental for practitioners and other stake holders). Convergence
should be the rule.

2 Theoretical considerations

2.1 Sociocultural assumptions of this research

The sociocultural turn in mathematics education research (e.g., Cobb, 2007) has sensitized the
community not only to the phenomenon of the co-constructive emergence of the classroom
mathematical culture by its participants – both teacher and learners – but also to the role played
by the ways of reasoning and communicating that draw upon historical practices of the
discipline. These ways of reasoning mathematically are correspondingly reflected in the
didactic literature on MR in schools.
Student discursive activity consisting of what students say, the way in which they say it,
what they do, the representations and drawings they make, the ways in which they use these
representations, and their intonations and gestures, all these are key levers in the activation of a
teacher’s input with respect to enculturating students to the ways of participating and the ways
of reasoning that are expected of them. To provide this input, teachers must not only be aware
of the nature of the forms and processes of mathematical reasoning that they wish students to
learn to participate in but also recognize when students are engaging in the desired aspects of
reasoning. This requires a well-elaborated vision of MR where discourse is fundamental and
that not only reflects the didactical discourse of the discipline but also serves as a conceptual
tool for teachers (and researchers) to analyze students’ discursive activity.

2.2 The commognitive frame

From this sociocultural view, MR can be seen as discursive activity. To conceptualize MR as

discursive activity, we turn to the commognitive framework developed by Sfard (2008, 2012).
Sfard (2008) defines commognition as Bthe term that encompasses thinking (individual
cognition) and (interpersonal) communicating. As a combination of the words communication
and cognition, it stresses the fact that these two processes are different (intrapersonal and
interpersonal) manifestations of the same phenomenon^ (p. 296). Within the commognitive
 Jeannotte D., Kieran C.

framework, discourse is central. According to Sfard (2008), discourse is a Bspecial type of

communication made distinct by its repertoire of admissible actions […] and discourses in
language are distinguishable by their vocabularies, visual mediators, routines, and endorsed
narratives^ (p. 297).
For a commognitive researcher, mathematics is a discourse, that is, a particular
type of communication (Sfard, 2008, 2012). Development of mathematical discourse
(i.e., development of mathematics) involves a change in discourse, which occurs
within the mathematical community in which we evolve. In a movement of individ-
ualization and (re)communication, changes in discourse are proposed, rejected, and
negotiated. For example, Sfard (2008) notes that algebra is a meta-discourse that
subsumes arithmetical discourse and becomes a discourse in itself. From an episte-
mological vantage point, this position has implications for the way in which MR will
be conceptualized: It is the given mathematics community in which we evolve that
fixes the rules, the acceptable visual mediators, and word use.
Another point is that within the commognitive approach, there are two main kinds of
discursive development: object-level and meta-level. Object-level discourse development
refers to an extension of an existing discourse on already constructed mathematical objects.
On the other hand, meta-level discourse development refers to the construction of a new
discourse, changing the rules of the game, which goes hand in hand with the building of new
mathematical objects. As will be seen later, the distinction between the type of discourse and
the type of discourse development will prove crucial in discriminating those mathematical
thinking processes that will be considered processes of MR.
The choice of Sfard’s (2008) commognitive frame as the main underpinning of this
research project on the elaboration of a model of MR for school mathematics has
certain implications and limits. The narrative form of the model is open to different
interpretations, especially if approached from an incommensurable discourse.
Furthermore, the aim here is not to construct a model that provides specific practical
advice with respect to classroom task sequences that are designed to encourage the
development of MR. Nor is the aim to construct a model that illustrates the research-
based ways in which students have been found to develop and communicate certain
modes of MR. Rather the aim is to construct, in harmony with the commognitive
frame, a coherent theoretical model that synthesizes and builds upon the convergences
to be found in the main types and characteristics of MR described in the mathematics
education research literature and that can thereby serve as a conceptual tool for both
teachers and researchers, that is, a narrative model that connects semantically different
concepts in the same coherent network – thus, a means to improve communication
with the help of a shared vocabulary (Lee, 1997).
The Oxford Living Dictionary (https://en.oxforddictionaries.com/definition/narrative)
defines narrative both as a spoken or written account of connected events and as a
representation of a particular situation or process in such a way as to reflect or
conform to an overarching set of aims or values. The connected events that are
represented by the model – connected and interpreted by means of the discursive
framework that underpins the model – are those aspects of mathematical reasoning
that have become a part of the current narratives in the mathematics education
community. And because narratives invite interpretation, the model, which is itself a
narrative, will evolve as others from other vantage points read it, and take it on and
adapt it to their own practice.
A conceptual model of mathematical reasoning

Fig. 1 The methodological process of anasynthesis

3 Methodology

The commognitive stand taken here creates methodological implications. For a commognitive
researcher, research development is equivalent to development of research discourse. A
researcher has to build on other researchers’ works and attempt to develop a common
discourse. This supports the justification that the data for this study are the mathe-
matics education literature resources dealing with MR. Those data allow for construct-
ing with and grounding upon the discourse already built by members of the
mathematics education community.
The methodological process that framed this research is anasynthesis (Legendre, 2005) –
anasynthesis being a neologism coined from the words analysis and synthesis. We present this
process linearly but readers should consider it a cyclical process (see Fig. 1).
First, we created a corpus from a review of databases and selected texts that had MR as a
keyword or associated keywords such as mathematical thinking, deductive reasoning, induc-
tive reasoning, and so on. Additional cycles provided new keywords that helped to elaborate
the corpus. Four criteria were used to select the corpus: access, completeness, recency, and
authenticity (Van der Maren, 1996). We also added famous texts and those cited by authors
who have studied MR but were not referenced in the database. To assure the quality of the
sources, the Toerner and Arzarello (2012) classification was used to refine the selection of
scientific journals. By the end of the process, 145 English and French texts2 (books, chapters,
articles, and research reports in proceedings) constituted the corpus.
Secondly, at every cycle, we analyzed the resulting corpus for relevant information related
to formal, axiological, and praxis characteristics of MR (Jeannotte, 2015). The formal charac-
teristic refers to accuracy, description, expression, and definition of concepts, terms, notions;
the axiological characteristic refers to aims, goals, principles; and the praxis characteristic
refers to norms, prescriptions, theoretical or experimental practices, habits, and custom.
Content and conceptual analyses supported the model building process. Successive and
repeated readings allowed for locating the MR units that were to be categorized. Every unit
received three codes: i) descriptors linked to the nature of the information (formal, axiological,
or praxis); ii) descriptors linked to its content (e.g., deductive reasoning, abductive reasoning,
inference, conjecturing, proving, …) – descriptors that helped to constitute the corpus; and iii)
descriptors linked to the emergent characteristics of MR (e.g., structure, process) – this third
level of codification going beyond the content and helping in the building of the model. For
example, the unit BTo reason is to infer a proposition, called conclusion, from certain

2
In French, the word raisonnement is usually translated as reasoning (see, e.g., Duval, 1991).
 Jeannotte D., Kieran C.

premises^ (Cabassut, 2005, p. 24, our translation) provided not only formal information, but
was also linked to the keyword inference, and highlighted both the structural aspect (i.e.,
premises, proposition, and conclusion) and the process aspect (in defining reasoning by an
action verb, i.e., inferring). The analysis was guided by commognitive principles (Sfard, 2012)
that led to searching for discursive elements within the texts of the corpus, but that also took
into account the context and epistemological positions behind the text.
Thirdly, this information was then synthesized so as to highlight convergences, diver-
gences, and to point out areas where there were theoretical gaps that would need to be filled
in by any model proposing to represent the central aspects of MR for school mathematics. Via
the commognitive framework that underpinned this research, a theoretical prototype was then
developed, which became a self-standing, theoretically-coherent model of MR after multiple
anasynthetic cycles (see Fig. 1). As the goal was to provide a portrait of the concept of MR, we
stopped the process when no new information emerged and when the model respected internal
coherence, relevance to the issue raised, and heuristic value (local validation). We emphasize
that the coherence of the model is constituted by its commognitive, discursive framing that, in
conjunction with the methodological procedures of anasynthesis, provided the theoretical tools
to discern both the gaps and overlaps within the mathematics education research literature
related to MR and thereupon to move forward in the creative act of conceptualizing the model.
Each element that emerged was either explicitly or implicitly linked to MR by multiple
texts. With the help of the commognitive frame, we reformulated them with a sound vocab-
ulary reflecting a discursive approach to MR. From this perspective, the resulting theoretical
model of MR is designed to enable a better and fuller understanding of MR in the context of
school mathematics, as well as a tool for improved communication by providing a shared
vocabulary. In addition, the model aims to nourish reflection on MR among researchers and
teachers – those who can, by their action, influence directly or indirectly student learning.
Discourse is what teachers rely on to judge whether MR is occurring in the class. Furthermore,
as Lee (1997) has pointed out, Bwithout a theoretical model, research remains piecemeal or
eclectic and without empirical research the diverse models or theories stagnate^ (p. 42). With
this model, we respond to Reid’s (2002) appeal to the mathematics education research
community: BThe aim of developing mathematical reasoning in classrooms calls on the
research community to clarify what is mathematical reasoning and what it looks like in school
contexts^ (p. 7).

4 Mathematical reasoning: what are its central aspects3

Four major elements emerged from the analysis of the mathematics education literature,
elements that helped to clarify the conceptual blur found therein: the activity/product dichot-
omy, the inferential nature of MR, the goal and functions of MR, and what we came to refer to
as the structural and process aspects.
The activity/product dichotomy relates to reasoning activity that is considered inaccessible
and for which the product is but an imperfect hint (e.g., Balacheff, 1988). The inferential
nature of MR is emphasized by many authors who point to the novel ideas that result from
such inferencing; however, the precise nature of this novelty has yet to be clarified. The

3
Please note that, in the interests of space, only a small number of the references that were actually consulted and
that served to help in the development of the model are included herein.
A conceptual model of mathematical reasoning

element of the goal and functions of MR also leads to questions, such as whether the goal of
MR is restricted to proving (e.g., deVilliers, 1999) or whether the function of all MR processes
is to change the epistemic value of a narrative (e.g., Duval, 1995). Finally, MR is traditionally
defined in terms of structure, that is, the form in which the reasoning is expressed, be it
deductive, inductive, or abductive. On the other hand, the process perspective, which is
espoused by others, tends not to be defined or explored epistemologically.
The articulation of these four elements, combined within the commognitive perspective,
leads to defining MR as a process of communication with others or with oneself that allows for
inferring mathematical utterances from other mathematical utterances. So the second major
element is captured in this definition, that is, the inferential nature of MR – the element that
will be seen to play a key role in both the structural and process aspects of MR. Also, we can
say from the analysis of the literature that MR develops the discourse by extension (endog-
enous discursive expansion, in the terminology of Sfard, 2008), that is, there is no change in
meta-discursive rules; there is no new mathematical object in a commognitive sense. The
novelty is situated within the objet-level utterances themselves. This definition also allows us
to avoid the first major element, that is, the activity/product dichotomy, in that discourse within
our framework is seen as both activity and product. Every communicational act presents both
the activity and product aspects, which are captured respectively by the process and structural
aspects of MR.
As will be seen, the dual aspects related to the structural and process element allow for
refining our above definition and at the same time integrate the different functions (the third
element) of MR. Thus, the structural and process element, in conjunction with key discursive
features of the underlying commognitive framework, will be seen to capture all four elements
of MR that emerged from the literature. These two aspects are elaborated in the next sections,
with less space devoted to the structural because it has already been relatively well explored
from an epistemological perspective in the mathematics education literature. It is important to
emphasize, before continuing, that the structural and process aspects of MR represent two
different ways of looking at a given discourse. Both aspects are present and are related
dialectically: structures are part of the process aspect of MR and processes contribute to the
construction of those structures.

4.1 Structural aspect of mathematical reasoning

The structural aspect of MR refers in general to a more static aspect that is related to the form
of a given piece of MR. More specifically, the structural aspect refers to the way in which the
discursive elements combine in an ordered system that describes both the elements and their
relation with each other. The more cited forms are deduction, induction, and abduction.
Likewise, Toulmin (2007) and Peirce (n.d.) are the more widely used references for
discussing the structural aspect within the literature. The Toulmin model schematizes
the basic elements (data, claim, warrant) along with the qualifier (linked to the
epistemic value), the backing (to further support the warrant), and the rebuttal (to
pre-empt possible counter arguments to the claim). All those elements are narrative in
nature and serve to structure the mathematical discourse. The Peirce model involves
three basic, one-step, modes of inference: the deductive, the inductive, and the
abductive. Every step is composed minimally of data, claim, and warrant (to use
the same terminology across the two models). The deductive, the inductive, and the
abductive each infer a different conclusion.
 Jeannotte D., Kieran C.

4.1.1 Deductive step

Deductive reasoning is, for some authors, synonymous with MR. Duval (1995), for example,
describes deductive reasoning as the only form of reasoning that can change the epistemic
value of mathematical knowledge from likely to true. As a structural aspect, the deductive step
infers a claim from data and warrant. The nature of the qualifier attached to the claim (which is
the conclusion for the deductive step) depends on the epistemic value of the data and the
warrant. The deductive form of reasoning plays an important role within the processes of
proving and formal proving, both of which require deductive restructuring (see Section 4.2.2).

4.1.2 Inductive step

Inductive reasoning is the second most common step in the literature linked to MR. It is
defined inconsistently, partly because it refers to every reasoning that is not deductive (Reid,
2010). In our model, the inductive step infers a warrant from the data and the claim about the
data. The epistemic value (i.e., the qualifier) that is allowed with respect to the conclusion of
the inductive step is that of likely (or probable). Inductive reasoning is linked to the soon-to-be-
described process of generalizing (Pedemonte, 2002; Rivera, 2008) in that this process can, at
one moment or another, be structured inductively.

4.1.3 Abductive step

Researchers interested in the study of exploration activity, such as Reid (2003) and Pedemonte
(2002), introduce the abductive step. It is a less discussed structure that is sometimes mingled
with the inductive step (Rivera, 2008). According to Eco (1983, in Pedemonte & Reid, 2011),
the abductive step can take two forms. The first infers data from the claim and the warrant. The
second infers data and warrant from the claim. For Peirce (n.d.), the abductive step infers
elements that can explain the claim. The abductive reasoning structure can be an element of
every MR process by generating data and warrant in the search for similarities and differences
as in, for example, generalizing (Rivera, 2008), conjecturing (Pedemonte, 2002), and also
validating (Pedemonte & Reid, 2011).

4.1.4 Concluding remarks on the structural aspect of mathematical reasoning

From a commognitive standpoint, the structural aspect highlights the construction rules of
mathematical discourse as well as its diverse components. It foregrounds the nature of the
conclusion and its epistemic value. According to Peirce (n.d.), the validity of the reasoning is
not judged only from its structure, but from the epistemic value attached to the conclusion, thus
giving a special status to the deductive step. (This position stands in contrast to standard
definitions of logical validity where an argument is valid if and only if it takes a form that
makes it impossible for the premises to be true and the conclusion nevertheless to be false; it is
not required that a valid argument have premises that are actually true.) The meta-rules that
control mathematical discourse (within a mathematical community) demand that the discourse
of reasoning be eventually structured into the deductive step. It is the unique structure that
allows for theorization of mathematical discourse.
Even if necessary to a model of MR for the teaching and learning of school mathematics,
the structural aspect is not sufficient for fully understanding the nature of MR in school. While
A conceptual model of mathematical reasoning

it puts forward, in a static way, the narrative elements, relations, and epistemic values that
constitute MR, it neglects the temporality and axiological nature of MR that are central to
reasoning activity. The process aspect, less developed in the literature, will fill that gap.

4.2 Process aspect of mathematical reasoning

Several action verbs are linked to MR in the literature, verbs that put forward the
temporal nature of MR. However, few texts conceptualize MR as a process. From the
commognitive perspective adopted here, the following definition of the process aspect
of MR emerges:
MR processes are commognitive processes that are meta-discursive, that is, that derive
narratives about objects or relations by exploring the relations between objects.
Of the several overlapping MR processes found within the literature, nine distinct
processes emerged. Eight of these were classified into one of two categories: the
processes related to the search for similarities and differences, or the processes related
to validating. These categories, which are similar to those described by Stylianides
(2008), materialized after several feedback cycles. The ninth process, that of exem-
plifying, was classified as a support for both of the other two categories. The
description of each of these nine processes follows. It includes: i) some background
literature that gave rise to the inclusion of the given process; ii) the creation of a
definition of the given process that is consistent with the commognitive framework;
and iii) a brief discussion of the process from a commognitive point of view.

4.2.1 Processes related to the search for similarities and differences

The following five processes relate to the search for similarities and differences: generalizing,
conjecturing, identifying a pattern, comparing, and classifying.

Generalizing For Stylianides (2008) and Artzt (1999), MR is all about generalizing and
arriving at valid conclusions. According to Stylianides (2008), generalizing is Bthe transpor-
tation of mathematical relations from given sets to new sets for which the original sets are
subsets^ (p. 9). For Pedemonte (2002), the processes of argumentation that involve general-
izing allow for passing to a larger set and also provide the reasons for believing in the
narrative: BMathematical argumentation always has the objective of seeking truth^
(p. 30, our translation). However, it is the passage from a given set to a larger one
that is highlighted by our anasynthesis (e.g., Dreyfus, 1991). The inferential and
expansion aspects of generalizing are considered its main features, thereby leading
to the following definition of generalizing:
Generalizing: A process that infers narratives about a set of mathematical objects or a
relation between objects of the set from a subset of this set.
From a commognitive viewpoint, we can link generalizing to MR because the process is
clearly associated with inference and discourse, without necessarily creating a new incom-
mensurable discourse. In contrast, abstracting (Jeannotte, 2015) produces a meta-level
development of discourse and is thereby not considered a process of MR from a
commognitive perspective.
 Jeannotte D., Kieran C.

Conjecturing According to Stylianides (2008), conjecturing has to lead to a reasoned

conjecture. The conjecture, as a narrative, is then always associated with the epistemic value,
probable or likely. Other MR processes are thus needed to determine whether the conjecture is
true or false. For Stylianides, conjecturing also leads to an utterance that generalizes the cases
generated, that is, that expands its domain of application. Mason (1982) describes conjecturing
as a cyclical process involving i) enunciating clearly a conjecture, ii) verifying that the
conjecture covers all known cases and examples, iii) being wary of the conjecture by trying
to refute it, and iv) finding out why it is true or modifying it (which brings us back to i). A
related element pulled from the literature is the link between conjecture and theorem.
Pedemonte (2002), who proposes a parallel between the two, states that argumentation is
linked with conjecture and formal proving is linked with theorem.
Thus, several elements are retained in order to build the definition of conjecturing as a MR
process. First, a central element is the search for regularity, the search for some relation. In fact,
it is the search for similarities and differences that emerges from the analysis. This search
allows for building a relation – around objects or other relations, that is, around some
mathematical phenomenon. Second, conjecturing leads to a narrative with probable or likely
as its epistemic value. There is an uncertainty about the built narrative. Third, conjecturing can
lead to a general discourse when it expands an observed relation to a larger set. What
distinguishes conjecturing from generalizing is that it has an epistemic value attached to it.
Conjecturing can thus be defined as follows:
Conjecturing: A MR process that, by the search for similarities and differences, infers a
narrative about some regularity with a likely or probable epistemic value and that has the
potential for mathematical theorization.
From a commognitive viewpoint, conjecturing leads to an extension of the discourse by the
building of likely narratives, based on the search for similarities and differences.

Identifying a pattern Is identifying a pattern different from the process of conjecturing? For
example, is there a probable epistemic value attached to the identifying of a pattern? According
to Stylianides (2008), identifying a pattern (i.e., a recursive relation) can lead to conjecturing,
but the two cannot be equated. For Cañadas, Deulofeu, Figueiras, Reid, and Yevdokimov
(2007), the third stage of conjecturing (i.e., empirical induction from a finite number of discrete
cases) is the search for and the predicting of a regularity in the pattern, which is clearly linked
to identifying a pattern. Identifying a pattern, according to Stylianides (2008), goes further than
observing a pattern. There is active searching, and then taking some distance from the
phenomenon, which are necessary for MR. Furthermore, as for generalizing, there is no
particular epistemic value associated with the inferred narrative. We thus define identifying a
pattern as follows:
Identifying a pattern: A MR process that, by the search for similarities and differences,
infers a narrative about a recursive relation between mathematical objects or relations.
This process, from a commognitive viewpoint, differs from conjecturing and generalizing
in that it is possible to identify a pattern that is applicable to a certain set without expanding it
to a larger set.

Comparing The term comparing is linked by various researchers to certain elements of MR,
such as inductive reasoning (Simon, 1996) and deductive reasoning (Duval, 1995). Pedemonte
A conceptual model of mathematical reasoning

(2002) links comparing to exemplifying and conjecturing. For her, MR must at some point
involve comparing examples so as to be able to conjecture. The key element here is the
inferential nature of comparing, which we define as follows:
Comparing: A MR process that infers, by the search for similarities and differences, a
narrative about mathematical objects or relations.
Comparing can take place along with a plethora of other MR processes: generalizing,
identifying a pattern, validating. For example, identifying a pattern necessitates comparing
cases or examples so as to highlight the pattern. However, identifying a pattern goes beyond
comparing because comparing only infers a narrative about similarities and differences.

Classifying Classifying is associated by some with MR. For Mason (2001), Bclassification is
not just about making distinctions and describing properties, but about justifying conjectures
that all possible objects with those properties have been described or otherwise captured^ (p.
7). Mason (2001) highlights a meta-discursive rule upon which classifying is contingent:
Mathematical properties and definitions are used to classify objects. The process of classifying
is thus defined as follows:
Classifying: A MR process that infers, by the search for similarities and differences
between mathematical objects, a narrative about a class of objects based on mathemat-
ical properties and definitions.
Classifying is an important process that allows for object-level development by putting
together or pulling apart different discursive objects, thereby structuring a discourse.
Classifying can be associated with comparing, conjecturing, and generalizing.

4.2.2 Processes related to validating

For the second set of processes, the change in epistemic value is foregrounded. Drawing on
Duval (1995), Lithner (2008), and Cañadas et al. (2007), the term validating is linked to the
epistemic value that an utterance in a given narrative can take on (e.g., likely, true, probable,
false) and depends on the mathematical discourse community where it emerged. Unlike Duval
(1995), from a discursive stand there is no difference between epistemic value and truth-value.
The epistemic value of an utterance depends not only on the logical validity of its structure but
also on the shared discourse (meta-rules and accepted narratives) of a given community. This
epistemic value, already highlighted by the structural aspect, is very important in mathematics
for systematizing discourses and thus for theorizing. The three processes below, which are
related to validating, are defined inclusively as follows:
Validating: A MR process that aims at changing the epistemic value (i.e., the likelihood
or the truth) of a mathematical narrative.
Contrary to conjecturing that infers a narrative that is likely, the validating processes aim at
changing a narrative’s epistemic value one way or another. This change can be from likely to
true, from likely to false, or even from likely to more likely. The meta-discursive rules of
mathematics constrain the possible changes of epistemic value. It is partly this definition of
validating that will help us to highlight the different particularities of the three processes related
to validation that emerged from the analysis of the corpus: justifying, proving, and formal
 Jeannotte D., Kieran C.

proving. For the three definitions, the word searching has been included to emphasize the
process-nature of the activity. It involves searching for discursive information (data, warrant,
backing), which allows for a change in epistemic value.

Justifying Justifying is viewed by Yackel and Hanna (2003) as a social process, that is, more
than one individual can be involved and the process is founded on public knowledge.
Likewise, for Duval (1995), Stylianides (2008), and Cabassut (2005), this process is linked
to a change in the epistemic value of a narrative by searching for data, warrant, or
backing to support this change. However, in the literature, it is mainly the passage
from likely to true that is addressed. Nevertheless, the process of justifying is
associated with two types of epistemic passage. The first is related to the justification
of a conjecture that arises from the process of conjecturing. This passage allows for
changing the epistemic value from likely to more likely, as stressed by Cabassut
(2005) in his discussions of plausible validation. The second type of epistemic
passage is related to a validation that changes the epistemic value from likely to true
or false, without being considered necessarily as constituting the process of proving.
Thus, justifying is defined as follows:
Justifying: A MR process that, by searching for data, warrant, and backing, allows for
modifying the epistemic value of a narrative.
The change of epistemic value is, as just mentioned, not necessarily from likely to true. The
elements supporting the process are constrained by meta-discursive rules within a certain
community. For example, the change from likely to true has to be based on a deductive
structure. On the other hand, in changing from likely to more likely, some meta-rules constrain
the process, but a deductive structure is not necessary.

Proving The literature on proving can be divided into two groups: the texts that deal with
proving and the texts that deal with what will be named formal proving.4 The next section
deals with formal proving. Proving is, as was the case for justifying, a social process.
According to Balacheff (1988), for example, proving is a type of explanation that is socially
acceptable. In addition, proving is linked to changing the epistemic value of a narrative:
BProving is the process employed by an individual (or a community) to remove doubts about
the truth of an assertion^ (Harel & Sowder, 2007, p. 807). However, the process of
proving is associated more with deductive reasoning than is the process of justifying.
For Maher (2009): BProof making is a special type of mathematical activity in which
children attempt to justify their claims by deductive argumentation^ (p. 121). Thus,
we define the process of proving as follows:
Proving: A MR process that, by searching for data, warrant, and backing, modifies the
epistemic value of a narrative from likely to true. This process is constrained by:

i) the narratives that are accepted by the class community (the set of accepted narratives)
that are true (from the viewpoint of the expert mathematician) and available without
additional justification;
ii) a final restructuring that is deductive in nature;

4
Formal proving is referred to in French as démontrer.
A conceptual model of mathematical reasoning

iii) the realizations (in the sense of Sfard, 2008, p. 301) that are appropriate and known,
or accessible, to the class.

In mathematics, the deductive structure is associated with rigor. While the meta-discursive rules
of mathematical discourse dictate that the validating process has to be restructured in a deductive
way at some point, we emphasize that proving as a process does not have to be deductively
structured at every moment. The notion of theorization underscored by Mariotti (2005) is that
proving relies on a set of narratives that are accepted as true. Moreover, by accepting non-
formalized realizations, the proving process can be developed from the primary school onward.
Proving is differentiated from justifying by its potential for theorization. It is also more
constrained than justifying in that it has to be restructured deductively and bear on a set of
accepted narratives that are coherent with the mathematical discourse of the expert (e.g., the
teacher), even if realized differently (informally).

Formal proving For Hanna and Jahnke (1996), Bformal proof arose as a response to a persistent
concern for justification^ (p. 889) among mathematicians. Formal proving5 is thus strongly
associated with change of epistemic value. For Balacheff (1988), formal proving is constrained
by a strict structure and meta-rules. Arsac (1996) highlights the social nature of those meta-rules
with theory brought to the forefront. Formal proving can be differentiated from proving essentially
by its rigor and formalism. While the process of proving is founded upon narratives that are
mathematically true, the process of formal proving goes farther in that the narratives must be
integrated explicitly into some mathematical theory. Formal proving is thus defined as follows:
Formal proving: A MR process that, by searching for data, warrant, and backing, modifies
the epistemic value of a narrative from likely to true. This process is constrained by:

i) the narratives that are accepted by the class community (the set of accepted narratives)
that are true (from the viewpoint of the expert mathematician) and systematized in a
mathematical theory;.
ii) a final deductive restructuring;
iii) realizations that are formalized and accepted by the class and mathematical
communities.

As opposed to proving, formal proving relies on mathematical theory built a priori and on
formalized realizations (axioms and theorems). As a consequence, the generic example as
elaborated by Balacheff (1988), which is acceptable within the previously described process of
proving, cannot be used within formal proving.

4.2.3 Exemplifying: a support for the other mathematical reasoning processes

Despite the fact that a smaller number of texts have dealt with exemplifying, the analysis of the
corpus yielded some highly relevant treatments. Mason (1982) defines exemplifying as a

5
Formal proving, an action derivative of formal proof (a term widely used in the mathematics education research
literature), should not be construed as reasoning that operates only syntactically.
 Jeannotte D., Kieran C.

process that allows for exploring a problem with the aim of conjecturing, or verifying the
conjecture and refining it. For Pólya (1968), exemplifying can lead to generalizing. The
generic type of exemplification (Balacheff, 1988) can be associated with the validating
process. Thus, linked to both the search for similarities and differences and validating,
exemplifying can support each of the processes of MR cited previously:
Exemplifying: A MR process that supports other MR processes by inferring examples
that assist in:

i) the search for similarities and differences;

ii) the search for validation.

Exemplifying allows for inferring data about a problem (which can be linked to the
abductive structure). Those data can then be recycled in the search for similarities or differ-
ences in patterns and relations, but also within the processes of validating. Exemplifying thus
generates elements that will serve in generalizing, in conjecturing, and even in validating.

4.2.4 Concluding remarks on the process aspect of mathematical reasoning

Even if they have been treated separately, all the processes of MR are interrelated. They
stimulate and influence each other, allowing for the development of an increasingly more
complex mathematical discourse by the generation of new narratives on already existing
discursive objects. In particular, conjecturing and proving play an essential role in mathemat-
ical theorization. Indeed, conjecturing infers narratives that can potentially enrich mathematical
theories and proving allows for systematizing the discourse, with the idea of theorizing it. Even
if school mathematics is not formalized in the same way as the mathematics of mathemati-
cians, systematization of discourse, even if somewhat local, can be engaged in within the
processes of conjecturing and proving. We have also seen, for instance, how exemplifying is
tightly tied both to the processes related to the search for similarities and differences and to the
processes related to validating.

5 A model of mathematical reasoning for school mathematics

The outcome of this research project is a conceptual model of MR for school mathematics.
Framed by commognitive theory and the methodology of anasynthesis, the model unifies a
previously unstructured domain according to two central aspects: the structural and the process
aspects – two different ways of looking at a given discourse that are related dialectically. More
specifically, the discursive framing of the research project provided a means of analyzing the
various types of MR within the mathematics education literature and reformulating them
according to their various discursive elements and the interrelationships among them. It is noted
that while the model is general enough to be applied to various mathematical content areas, it
remains a model of mathematical reasoning, as opposed to a model of reasoning per se, by the
mere but important fact that it is rooted in a particular body of literature, that of mathematical
reasoning, and is thus grounded within the discourse already built by members of the mathemat-
ics education community. By a similar argument, and also in line with the same commognitive
stance, it is a model of MR for school mathematics – accounted for not only by the educational
A conceptual model of mathematical reasoning

literature source from which the model arises but also by the manner in which its different
elements (e.g., conjecturing and proving) have been characterized with the student of school
mathematics in mind. In sum, rephrasing MR discourse according to commognitive theory can by
its very nature help both researcher and teacher focus on what is visible and audible in the
classroom. Furthermore, adopting a common discourse on MR carries with it the potential for
members of the various mathematics education communities not only to better communicate with
respect to MR but also to develop the learning resources needed for the improvement of MR in
school. Nonetheless, because a narrative model is a living thing, this model of MR will surely
evolve and grow in the hands of its users. Lastly, in the spirit of Balacheff (2008), the explicit
formulation of the epistemological underpinnings of this model of MR can also be seen as a first
step in eventually creating bridges between different epistemologies.

Acknowledgments The authors express their appreciation to the Social Sciences and Humanities Research
Council of Canada for its financial support of this research and to the reviewers for their helpful comments. In
addition, the first author wishes to thank the members of her doctoral research committee for their academic
support and encouragement during the years of this project.

References

Arsac, G. (1996). Un cadre d’étude du raisonnement mathématique [A mathematical reasoning framework]. In

D. Grenier (Ed.), Séminaire didactique et technologies cognitives en mathématiques. Grenoble: IMAG.
Artzt, A. F. (1999). Mathematical reasoning during small-group problem solving. In L. V. Stiff (Ed.), Developing
mathematical reasoning in grades K-12. (1999 yearbook, pp. 115–127). Reston, VA: NCTM.
Balacheff, N. (1988). Une étude des processus de preuve en mathématiques chez des élèves du collège [A study
of students’ proving processes in junior high school]. (Unpublished doctoral dissertation). Université de
Grenoble 1, France.
Balacheff, N. (2008). The role of the researcher’s epistemology in mathematics education: An essay on the case
of proof. ZDM, 40(3), 501–512.
Cabassut, R. (2005). Démonstration, raisonnement et validation dans l’enseignement secondaire des
mathématiques en France et en Allemagne [Proof, reasoning and validation in mathematics secondary
teaching in France and Germany]. (Unpublished doctoral dissertation). Université Paris Diderot, France.
Cañadas, M. C., Deulofeu, J., Figueiras, L., Reid, D. A., & Yevdokimov, O. (2007). The conjecturing process:
Perspectives in theory and implications in practice. Journal of Teaching and Learning, 5(1), 55–72.
Cobb, P. (2007). Putting philosophy to work. In F. K. Lester (Ed.), Second handbook of research on mathematics
teaching and learning (pp. 3–38). Charlotte: Information Age.
deVilliers, M. (1999). The role and function of proof with Sketchpad. In M. deVilliers (Ed.), Rethinking proof
with Sketchpad (pp. 3–10). Emeryville: Key Curriculum.
Dreyfus, T. (1991). Advanced mathematical thinking processes. In D. Tall (Ed.), Advanced mathematical
thinking (pp. 25–41). Dordrecht: Kluwer Academic.
Duval, R. (1991). Structure du raisonnement déductif et apprentissage de la démonstration [Structure of
deductive reasoning and learning of proof]. Educational Studies in Mathematics, 22(3), 233–261.
Duval, R. (1995). Sémiosis et pensée humaine: Registres sémiotiques et apprentissages intellectuels [Semiosis
and human thinking: Semiotic registers and intellectual learning]. Berne, CH: Lang.
Hanna, G., & Jahnke, H. (1996). Proof and proving. In A. Bishop et al. (Eds.), International handbook of
mathematics education (pp. 877–908). Dordrecht: Kluwer Academic.
Harel, G., & Sowder, L. (2007). Toward comprehensive perspectives on the learning and teaching of proof. In F.
K. Lester (Ed.), Second handbook of research on mathematics teaching and learning (Vol. 2, pp. 805–842).
Charlotte: Information Age.
Jeannotte, D. (2015). Raisonnement mathématique: proposition d'un modèle conceptuel pour l'apprentissage et
l'enseignement au primaire et au secondaire [Mathematical reasoning: a conceptual model for learning and
teaching at the primary and secondary levels of schooling]. (Unpublished doctoral dissertation). Université
du Québec à Montréal, Canada.
 Jeannotte D., Kieran C.

Lee, L.F. (1997). La compréhension algébrique: La recherche d’un modèle dans la communauté d'éducation
mathématique [Algebraic understanding: The search for a model in the mathematics education community].
(Unpublished doctoral dissertation). Université du Québec à Montréal, Canada.
Legendre, R. (2005). Dictionnaire actuel de l’éducation [Dictionary of education] (3rd ed.). Montréal: Guérin.
Lithner, J. (2008). A research framework for creative and imitative reasoning. Educational Studies in
Mathematics, 67(3), 255–276.
Maher, C. (2009). Children’s reasoning: Discovering the idea of mathematical proof. In D. A. Stylianou, M. L.
Blanton, & E. J. Knuth (Eds.), Teaching and learning proof across the grades (pp. 120–132). New York:
Routledge.
Mariotti, M.A. (2005). Proof and proving in algebra. Retrieved from http://www.lettredelapreuve.
org/OldPreuve/Newsletter/06Hiver/Mariotti-GARME.pdf
Mason, J. (1982). Thinking mathematically. London: Addison-Wesley.
Mason, J. (2001). Questions about mathematical reasoning and proof in schools. Opening address. QCA
Conference, UK. Retrieved from http://xtec.cat/centres/a8005072/articles/proof_and_reasoning.pdf
Meyer, M. (2010). Abduction - a logical view for investigating and initiating processes of discovering mathe-
matical coherences. Educational Studies in Mathematics, 74(2), 185–205.
NCTM. (2000). Principles and standards for school mathematics. Reston: NCTM.
OCDE. (2006). Compétences en sciences, lecture et mathématiques: Le cadre d’évaluation de PISA 2006
[Competencies in sciences, reading and mathematics: PISA 2006 evaluation framework]. Paris: OCDE.
Pedemonte, B. (2002). Étude didactique et cognitive des rapports de l’argumentation et de la démonstration
dans l'apprentissage des mathématiques [Didactical and cognitive study of argumentation and proof in the
learning of mathematics]. (Unpublished doctoral dissertation). Université Joseph Fourier, Grenoble, France.
Pedemonte, B., & Reid, D. A. (2011). The role of abduction in proving processes. Educational Studies in
Mathematics, 76(3), 281–303.
Peirce, C. S. (n.d.). The collected papers of Charles Sanders Peirce (Electronic ed.). Charlottesville: InteLex Past
Master. Retrieved from http://library.nlx.com/xtf/view?docId=peirce/peirce.00.xml;chunk.id=div.peirce.
pmpreface.1;toc.depth=1;toc.id=div.peirce.pmpreface.1;brand=default&fragment_id=.
Pólya, G. (1968). Mathematics and plausible reasoning (2nd ed.). Princeton: Princeton University Press.
Reid, D. A. (2002). Conjectures and refutations in grade 5 mathematics. Journal for Research in Mathematics
Education, 33(1), 5–29.
Reid, D. A. (2003). Forms and uses of abduction. In M. A. Mariotti (Ed.), Proceedings of 3rd Conference of
European Society for Research in Mathematics Education (pp. 1–10). Bellaria: CERME.
Reid, D. A. (2010). Proof in mathematics education. Rotterdam: Sense.
Rivera, F. D. (2008). On the pitfalls of abduction: Complicities and complexities in patterning activity. For the
Learning of Mathematics, 28(1), 17–25.
Sfard, A. (2008). Thinking as communicating: Human development, the growth of discourses, and
mathematizing. New York: Cambridge University Press.
Sfard, A. (2012). Introduction: Developing mathematical discourse - some insights from communicational
research. International Journal of Educational Research, 51-52, 1–9.
Simon, M. A. (1996). Beyond inductive and deductive reasoning: The search for a sense of knowing.
Educational Studies in Mathematics, 30(2), 197–210.
Steen, L. (1999). Twenty questions about mathematical reasoning. In L. V. Stiff (Ed.), Developing mathematical
reasoning in grades K-12. (1999 Yearbook, pp. 270–285). Reston: NCTM.
Stylianides, G. J. (2008). An analytic framework of reasoning-and-proving. For the Learning of Mathematics,
28(1), 9–16.
Toerner, G., & Arzarello, F. (2012). Grading mathematics education research journals. Newsletter of European
Mathematical Society (December, issue #86), 52-54. Retrieved from https://www.ems-ph.
org/journals/newsletter/pdf/2012-12-86.pdf
Toulmin, S. E. (2007). The uses of argument. New York: Cambridge University Press.
Van der Maren, J.-M. (1996). Méthodes de recherche pour l’éducation [Educational research methods] (2nd ed.).
Montréal: Les presses de l’Université de Montréal.
Yackel, E., & Hanna, G. (2003). Reasoning and proof. In J. Kilpatrick, W. G. Martin, & D. Schifter (Eds.), A
Research Companion to Principles and Standards for School Mathematics (pp. 227–236). Reston: NCTM.