Reed 2021

Download as pdf or txt
Download as pdf or txt
You are on page 1of 43

Cognition and Instruction

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/hcgi20

Leveraging a Categorization Activity to


Facilitate Productive Generalizing Activity and
Combinatorial Thinking

Zackery Reed & Elise Lockwood

To cite this article: Zackery Reed & Elise Lockwood (2021): Leveraging a Categorization Activity
to Facilitate Productive Generalizing Activity and Combinatorial Thinking, Cognition and Instruction,
DOI: 10.1080/07370008.2021.1887192

To link to this article: https://doi.org/10.1080/07370008.2021.1887192

Published online: 02 Mar 2021.

Submit your article to this journal

Article views: 62

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=hcgi20
COGNITION AND INSTRUCTION
https://doi.org/10.1080/07370008.2021.1887192

Leveraging a Categorization Activity to Facilitate Productive


Generalizing Activity and Combinatorial Thinking
Zackery Reeda and Elise Lockwoodb,c
a
Department of Mathematics, Oklahoma State University, Stillwater, OK, USA; bDepartment of Mathematics,
Oregon State University, Corvallis, OR, USA; cCenter for Computing in Science Education, University of Oslo,
Oslo, Norway

ABSTRACT
In this paper, we present data from two iterative teaching experiments involving students’ con-
structions of four basic counting problems. The teaching experiments were designed to leverage
the generalizing activities of relating and extending to provide students with opportunities to
reflect on initial combinatorial activity when constructing these formulas. We discuss three com-
binatorial ways of thinking that emerged from their work on this task and provide commentary on
the interplay between the students’ generalizing activities and engagement in the emergent ways
of thinking. We address the practical goals of identifying ways that students productively reason in
combinatorics, and we provide a theoretical commentary on students’ reflection on activity as it
occurs in the context of particular generalizing actions.

Introduction
Enumerative combinatorics involves the solving of counting problems, which ask for the number
of outcomes that satisfy particular constraints. Being able to solve such problems has applications
in fields including probability, statistics, and computer science, and the easy–to–state nature of
the problems make them accessible to students and provide opportunities for rich mathematical
thinking (e.g., Kapur, 1970; Tucker, 2002). There is evidence that combinatorics is a difficult topic
for students to learn (e.g., Hadar & Hadass, 1981; Martin, 2001; Tucker, 2002), and students at
all ages tend to struggle with correctly solving counting problems (e.g., Annin & Lai, 2010;
Batanero, Navarro–Pelayo, & Godino,1997; Eizenberg & Zaslavsky, 2004; Kavousian, 2008). Prior
research has shown that students find it challenging to distinguish between counting problems
(e.g., Batanero, et al., 1997), and often face difficulties with justifying counting formulas (e.g.,
Batanero, et al., 1997; Lockwood, 2014). We see a need to identify aspects of students’ rich com-
binatorial understandings of basic counting problems and their formulas so that students can be
equipped to solve and understand advanced counting problems.
We are also interested in studying the practice of generalization more broadly, especially in
non–algebraic contexts and among undergraduate students. We seek to contribute to the field’s
understanding of the nature of generalization by documenting the use of specific generalizing
actions to foster students’ engagement with desirable ways of thinking as a means of refining and
regulating their reflection on activity. We demonstrate this within the domain of combinatorics,
but we suggest in the discussion section that we are presenting a case of a broader phenomenon.
In this paper, we discuss the results of two iterative teaching experiments in which under-
graduate students engaged in a set of activities that was designed to elicit reflection on, and gen-
eralization of, initial combinatorial activity. This activity, called the Categorization Activity,

CONTACT Zackery Reed zackkr@okstate.edu Oklahoma State University, 401 Mathematical Sciences Building, Stillwater,
OK, 74078, USA.
ß 2021 Taylor & Francis Group, LLC
2 Z. REED AND E. LOCKWOOD

consisted of three phases: initial solving of counting problems, relating through categorization,
and extending through characterization.
We seek to address the following research goals in this paper. First, we have the pragmatic
goal of gaining insight into ways that students might reason about and productively understand
distinctions among formulas for fundamental counting problems. A second pragmatic goal is to
highlight particular combinatorial ways of thinking that emerged during students’ work on the
Categorization Activity which characterize salient aspects of the students’ reasoning that we found
to be productive. That is, we are not only interested in demonstrating that students could come
up with and justify key formulas–we also want to describe the combinatorial ways of thinking on
which they drew to accomplish this task. We believe that this attention to ways of thinking pro-
vides us insight into the nature of productive combinatorial thinking and activity.
We also seek to offer a more substantial theoretical contribution to the field’s understanding
of the relationship between generalization, ways of thinking, and reflection on activity. We lever-
age generalizing actions as explicit actions through which students can build on initial activity
and generate productive understandings of basic formulas. While reflection on initial activity is a
well–documented aspect of learning and generalization (e.g., Ellis, Lockwood, Tillema, & Moore,
2017; Gravemeijer, Cobb, Bowers, & Whitenack, 2000; Harel & Tall, 1991; Piaget, 2001), we pro-
vide detailed accounts of the students’ reflections through the relating and extending generalizing
actions (Ellis et al., 2017), and demonstrate a relationship between these generalizing actions and
students’ ways of thinking (Harel, 2008a, 2008b). We thus make two contributions to the field’s
understanding of generalization. First, we make use of the Relating–Forming–Extending frame-
work (Ellis et al., 2017) to offer a detailed analysis of the ways students might reflect on their ini-
tial activity to construct general understandings. Second, we highlight specific aspects of the
relationship between students’ generalizations and their mathematical ways of thinking. We do
this by demonstrating that the students’ generalizing activities both drew from and reinforced
these ways of thinking in the different phases of the Categorization Activity, and we argue that
this ultimately helped to refine students’ final understandings of the counting formulas.
In this paper, we attempt to answer the following research question to address our research
goals: What generalizing activities and combinatorial ways of thinking can help students articulate
and distinguish between problem types and understand key formulas in combinatorics?

Literature review and theoretical perspective


We now frame our work within the existing literature and describe our relevant theoretical per-
spectives. First, we characterize how we take generalization, situating our research within the cur-
rent literature on generalization and describing the means by which generalization facilitated the
students’ constructions of the general counting formulas. We relate this to Harel’s (2008a, 2008b)
duality between students’ ways of understanding and ways of thinking to frame the students’ cog-
nition in the study. We then present prior work on students’ combinatorial thinking. Finally, we
conclude with a mathematical discussion that frames the counting problems that the students in
our study solved.

Generalization
Our study explores ways that generalizing actions can be explicitly leveraged to engage students
in reflections on initial activity so as to promote desirable combinatorial ways of thinking and
ways of understanding fundamental counting formulas.
COGNITION AND INSTRUCTION 3

Relevant literature on generalization


Generalization has been recognized as a key aspect of mathematical activity by both researchers
(e.g., Amit & Klass-Tsirulnikov, 2005; Ellis, 2007b) and policymakers (Council of Chief State
School Officers, 2010). Much of the literature on students’ generalizing activity focuses on alge-
braic contexts, particularly among young students (e.g., Amit & Neria, 2008; Carpenter, Franke &
Levi, 2003; Ellis, 2007a, 2007b; Radford, 2008; Rivera, 2010; Rivera & Becker, 2007; 2008). There
has also been some work that investigated generalization at the postsecondary level in a number
of areas including calculus (e.g., Jones and Dorko, 2015; Kabael, 2011), linear algebra (e.g.,
Dubinsky, 1991; Harel & Tall, 1991), combinatorics (Lockwood, 2011; Lockwood & Reed, 2016,
2018), and real analysis (Reed 2017; 2018a; 2018b).
Aside from content–specific investigations, much of the literature regarding generalization is
either descriptive or categorical (i.e., characterizing types or hierarchies of generalizations, or
reporting strategies employed when generalizing). Such categorical studies (e.g., Amit & Neria,
2008; Bills & Rowland, 1999; Carraher, Martinez, & Schliemann, 2008; Doerfler, 2008; Ellis,
2007a; Ellis et al., 2017; El Mouhayar, 2018; Font & Contreras, 2008; Harel, 2001; Harel & Tall,
1991; Oflaz & Demircioglu, 2018 ; Rivera & Becker, 2008; Radford, 2006, 2008; Radford, 2008)
offer empirically–grounded distinctions between students’ generalizations that allow researchers
to infer aspects of students’ cognitive reorganizations that contribute to their generalization of an
idea. We employ the R–F–E framework (Ellis et al., 2017) which unifies many of these disparate
categories of generalizing into concise sub–categories of relating, forming, and extending actions.
These three categories are also designed to apply to generalization across mathematical disciplines
beyond algebra. We make use of the R–F–E framework to demonstrate the ways that empirically
grounded categories of generalization identified in the literature can not only serve as explanatory
constructs, but also can be leveraged to elicit particular reflections on initial activity that are
grounded in ways of thinking.
We also contribute to the literature on generalization by providing insight into students’ reflec-
tions on initial activity in greater detail than has been previously reported. Some characterizations
of generalization suggest that more mathematically mature generalizations entail some kind of
reflection on initial activity,1 as opposed to generalizations rooted in noticing figural and sym-
bolic similarity or numerical regularity (e.g., Amit & Neria 2008; Bills & Rowland, 1999; Doerfler,
2008; Ellis et al. 2017; Harel, 2001; Radford 2006; Rivera & Becker, 2008). While reflection on
activity has been cited as an important component of productive generalization, we see a need
for more research that details the means by which this reflection can support students’ construc-
tions of knowledge. We offer more details in Section 2.2.1.2.

Defining generalization and its role in generating understanding


There are many ways in which researchers have characterized generalization. We follow Ellis
(2007a) (who drew on Kaput, 1999) in defining “generalization” as “engaging in at least one of
three activities: a) identifying commonality across cases, b) extending one’s reasoning beyond the
range in which it originated, or c) deriving broader results about new relationships from particu-
lar cases” (p. 444). Using this definition allows us to view generalization from students’ perspec-
tives and attend to students’ generalizations even if they do not align with normatively correct
generalizations. By taking a student–centered view of generalization, we can characterize the kinds
of generalizing activities that may facilitate construction of productive content–specific
understandings.

1
More broadly, reflection on and coordination of specific mathematical activity is often considered an important component of
mathematical learning (e.g., Dubinsky, 1991; Gravemeijer, Cobb, Bowers, & Whitenack, 2000; Piaget, 2001; Sfard, 1991).
4 Z. REED AND E. LOCKWOOD

For purposes of describing students’ activity as they generalize, we adopt Ellis et al.’s (2017)
Relating–Forming–Extending (R–F–E) framework of generalizing activity. Each category in the
framework represents a different manifestation of generalizing activity in student work. We will
only focus on relating and extending for the purposes of this paper, as these were the only two
constructs that we incorporated into our experimental design. In particular, the hypotheses that
we initially set out to test in our teaching experiments incorporated only relating and extending
as earlier constructs from Ellis (2007a), from which the R–F–E framework was refined.
Relating occurs when students establish “relations of similarity across problems or contexts”
(Ellis et al., 2017, p. 680). The key feature of relating is that students are establishing such rela-
tionships of similarity, and previous work has documented students relating many different kinds
of mathematical objects, phenomena, or processes. For instance, students might articulate rela-
tions of similarity among the wording of problems, symbolic features, formulas or expressions,
numerical regularity, figural similarity within various representations, or solution methods (Ellis,
2007a; 2007c; Ellis et al., 2017; Lockwood, 2011). We will focus on students’ relating of initial
activity, which is but one kind of relationship students can establish when relating.
Extending involves the application of established patterns and regularities to new cases (p.
680). Much like with relating, the nature of the patterns and regularities that students extend can
vary – students may extend solution processes, techniques, numerical patterns, and figurate or
operative relationships. Given the possible variety of ways in which students might engage in
these generalizing actions, we are interested in the kinds of relationships that students establish
between particular counting problems that can contribute to extending that leverages productive
combinatorial understandings. This is in line with our practical goal of generating models of pro-
ductive combinatorial reasoning from empirical data.
Moreover, we consider these generalizing actions to be the observable mathematical actions
from which we can infer students’ cognitive organizations. Another primary goal of our study is
to better understand the roles that the generalizing actions of relating and extending can play in
facilitating students’ reflection on initial activity. In pursuit of this goal, we make use of the
R–F–E framework to explore ways in which generalizing actions can support students to refine
particular understandings and elicit particular ways of thinking. To make claims about the ways
in which our students understood the general counting formulas through engagement in these
generalizing actions, we require a means of elaborating students’ actions and utterances in regard
to their cognition. For this, we turn to Harel’s (2008a, 2008b) constructs of ways of understanding
and ways of thinking in the next section.

Ways of understanding, ways of thinking, and generalization. According to Harel’s (2008a)


Duality Principle, carrying out a mental act might result in a particular product, where “such a
product is the person’s way of understanding associated with that mental act” (Harel, ibid, p.
490). We pay particular attention to the ways that the students understood each counting formula
(and problem type) while working through the Categorization Activity. Similarly, given a collec-
tion of a student’s ways of understanding, it might be true that “ … they share a common cogni-
tive characteristic. Such a characteristic is referred to as a way of thinking associated with that
mental act” (Harel, ibid, p. 490). Thus, ways of thinking are constructs by which researchers con-
vey commonality across a student’s engagements in particular mental acts, attributing characteris-
tics to their mathematical reasoning within or across disciplines. The Duality Principle suggests
that students’ ways of thinking develop through repeated production of ways of understanding,
and that their ways of understanding are influenced by their ways of thinking (Harel, 2008b,
pp. 899).
Harel (2008a) illustrates this duality by examining a student’s multiple hypothetical ways of
understanding the string y ¼ 3x þ 5 including “an equation–a constraint on the quantities, x
and y” and “a thing where what you do on the left you do on the right” (p. 490). This final way
COGNITION AND INSTRUCTION 5

of understanding suggested a “non–referential symbolic way of thinking–a way of thinking where


mathematical symbols are free of coherent quantitative or relational meaning” (ibid, p. 490), sug-
gesting the student’s broad way of thinking about mathematical equations was brought to bear
on their way of understanding a particular equation. In this way, we can see the potential impact
of a way of thinking on one’s understanding of a mathematical object or concept. This example
also highlights that some ways of thinking may be more or less desirable or effective for students
to develop robust understandings.
In line with Harel’s observation that there may be multiple hypothetical interpretations of the
string y ¼ 3x þ 5, we anticipate that students might construct various understandings of count-
5!
ing problems and their solutions. For instance, consider an expression like 3!2! : This may be inter-
preted combinatorially as a process for determining the number of 2–element subsets from a
5–element set, involving arranging objects and dividing to account for duplicates. Or, it may be
interpreted algebraically or numerically as simply an expression involving numerical relationships
with no explicit connection to a combinatorial context. We seek for students to have combinator-
ial, not only algebraic or numerical ways of thinking about such an expression. In line with the
literature below on students’ thinking in combinatorics, students might understand counting
problems and their solutions by appeals to perceived semantic features, or numerical, symbolic,
or formulaic regularity. Some of these ways of understanding might not be productive for gener-
ally solving future counting problems, and we are interested in exploring ways that students con-
struct and leverage productive understandings rooted in mental activity that we consider to be
combinatorial, meaning that the students’ ways of understanding are products of mental acts on
counting processes or sets of outcomes (in the sense of Lockwood, 2013, 2014). Students can of
course solve counting problems and not engage in combinatorial thinking, as we discuss in the
literature. We are therefore interested in ways that students reflect on initial activity that is com-
binatorial in the sense that it is rooted in reasoning about outcomes tied to a particular combina-
torial context.
To this end, our experimental design involved engaging students in novel counting activity,
and then examining ways that they organized their understandings in relating and extending.
Because of this, we will refer to initial activity in which they engage and from which they general-
ize. By this, we mean that a key feature of our analysis will entail making inferences about the
qualities of students’ understandings that are constructed when working on combinatorics prob-
lems for the first time, particularly inferred from their observable activity, and then what aspects
of their understandings they carry forward in generalizing. Part of what allows us to engage in
such work is that we have an assumption that students’ observable mathematical activity is a
manifestation of underlying ways of thinking and ways of understanding.
We will similarly leverage the ways of thinking construct to identify students’ reasoning about
solving combinatorics problems and then demonstrate the influence of these combinatorial ways
of thinking on their understandings2 of specific counting formulas and problem types. We will
show student utterances and activity that suggest common characteristics across their engagement
with multiple counting problems. From this activity, we can infer their repeated enactment of
certain combinatorial ways of thinking,3 which in turn influence the ways they understood gen-
eral problem types and counting formulas being constructed through engagement in generaliz-
ing actions.

2
Our use of the generic term “understanding” should be interpreted as students’ ways of understanding in terms of
Harel (2008a)
3
We acknowledge that according to Harel (2008a), ways of thinking can take a long time to form and stabilize. We clarify that
the characteristics of our students’ mental activity suggest (rather than unequivocally demonstrate) certain ways of thinking.
Thus, we use ways of thinking to attribute observed cognitive characteristics of the students’ initial activities that, given time,
might produce the described ways of thinking.
6 Z. REED AND E. LOCKWOOD

We thus employ the ways of understanding construct to attribute the students’ understandings
of various problem types to the enactment and envisioning of mental acts carried out in counting
contexts. We then draw from the R–F–E framework to examine the ways that the students organ-
ized and then broadened the range of applicability of their initial understandings when engaging
in the relating and extending, and also to examine the relationship between students’ generalizing
and ways of thinking.

Synthesizing our theoretical constructs and discussing reflection on activity. To synthesize our
theoretical perspectives, then, we draw both from Ellis et al.’s framework (2018) for generalizing
and Harel’s (2008) constructs of ways of understanding and ways of thinking to provide a
detailed, comprehensive account of students’ reflection on initial activity. Reflection on activity
has traditionally been broadly construed in terms of cognitive reorganization. Wheatley (1992) –
drawing from Sigel (1981) – described students “distanc[ing] themselves from the activity and
“hold[ing] the activity in thought” to “make their activity an object which can be examined” (p.
536). Similarly, Piaget’s notion of reflective abstraction (von Glasersfeld, 1995; Piaget, 2001) entails
the projection and subsequent reflection (reorganization) of borrowed mental operations to
higher levels of thought (von Glasersfeld, 1995; Piaget 2001), from which other theories such as
APOS (Dubinsky, 1991) have been developed. While these characterizations have repeatedly been
employed in empirical work, we consider there to be room for studies that specify particular
activities that might elicit the occurrence of such reflection. We present one such detailed account
of reflection on activity.
Specifically, we first draw from Harel’s ways of understanding construct to make inferences
about students’ initial solutions of counting problems, considering their understandings as initial
cognitive material from which to generalize. The students’ ways of understanding are akin to the
mental operations to be projected and reorganized in reflective abstraction. We interpret that we
can make inferences about students’ cognition (which we discuss in terms of ways of thinking
and ways of understanding) via observing their mathematical activity. We then employ the relat-
ing and extending constructs in our experimental design as the particular observable mathemat-
ical generalizing actions through which students have opportunities to reorganize and broaden
the applicability of their ways of understanding. We consider relating and extending to serve as
actions that mediates students’ reflection on initial activity.
We then analyze the cognitive organizations from students’ relating and extending activities,
particularly attending to relationships that students find meaningful.4 For instance, we attend to
whether students organize their initial activity according to features of their mental operations,
numerical relationships, or other features they perceive in the initial problems they solve. In this
way, we account for the nature of the students’ reflection on their initial activity by attending to
the features of their activity that they found meaningful when relating and extending.
We will see that the students consistently engaged in particular ways of thinking while count-
ing and generalizing. This engagement allows us to leverage Harel’s ways of thinking construct as
an explanatory mechanism for the students’ initial activity and organizations when generalizing.
Finally, we draw from the students’ utterances and demonstrated mathematical activities to make
inferences about the ways they understood their eventually constructed counting formulas,
attending to the particular mental acts from their initial activity that the students applied to
broader classes of problems through the process of generalizing.
In sum, we attempt to inform the theory of generalization by providing a finer grained analysis
of ways that specific reflections on initial activity serve to regulate students’ generalizations,
emphasizing the nature of their reflection. Being able to observe this phenomenon also brings
into focus the impact of students’ ways of thinking on their generalizing.

4
This is in accordance with an actor-oriented perspective (Lobato, 2003)
COGNITION AND INSTRUCTION 7

Literature on students’ combinatorial reasoning


Although counting offers opportunities for students to engage in rich mathematical thinking
(Kapur, 1970; Tucker, 2002), there is abundant evidence that students struggle with reasoning
combinatorially and successfully solving counting problems (e.g., Batanero, et. al., 1997; Eizenberg
& Zaslavsky, 2004; Lockwood & Gibson, 2016). Prior research indicates that students struggle
with some common issues like handling overcounting and order (e.g., Annin & Lai, 2010;
Batanero et al., 1997; Lockwood & Caughman, 2016;; Lockwood & Purdy, 2019b), understanding
subtle wordings of counting problems (Hadar & Hadass, 1981; Kavousian, 2008), and verifying
answers (Eizenberg & Zaslavsky, 2004).
One particular area of difficulty for students is to understand basic counting formulas and dis-
tinctions between fundamental problem types. In a study involving 14–15 year–old students,
Batanero et al. (1997) list errors of order and errors of repetition among common errors that stu-
dents face. Batanero et al. define errors of order as consisting of “confusing the criteria of combi-
nations and arrangements, that is, distinguishing the order of the elements when it is irrelevant,
or, on the contrary, not considering the order when it is essential” (p. 191–192). They character-
ize errors of repetition as occurring when “the pupil does not consider the possibility of repeating
the elements when it is possible, or he/she repeats the elements when there is no possibility of
doing so” (p. 192). These findings suggest that students may find it difficult to distinguish
between problem situations and what solutions or formulas may appropriately be applied in given
counting situations. Lockwood (2014) found that similar issues arose for undergraduates, as she
reported that when solving a counting problem, an undergraduate student said, “I’m doing the
combination ones because I’m pretty sure order doesn’t matter with combination” (Lockwood,
2014, p. 33). When asked why, the student said, “I’m not sure about that one (laughs). I just kind
of go off my gut for it, on the ones that don’t specifically say order matters or it doesn’t matter”
(p. 33). This response offers an example of a student who did not demonstrate a combinatorial
understanding of what approach or formula could help them solve the problem. These examples
suggest that there is a need to investigate and develop ways to help students effectively identify
counting situations, to distinguish between fundamental problem types or problem constraints,
and to understand how to solve such problems.
In light of such difficulties, some researchers have previously developed ways to help students
reason combinatorially about formulas and problem types. For example, Lockwood et al. (2015a)
conducted a teaching experiment in which two undergraduate students (who were novice coun-
ters) reinvented four basic counting formulas. In this study, students were given a sequence of
tasks that motivated the development of general formulas. On the one hand, this was a successful
reinvention, as students correctly generated counting formulas and used them to solve a variety
of problems, and the reinvention seemed to effectively give the students some more ownership
over the counting formulas. However, although the students were successful at solving problems
and using their reinvented formulas, Lockwood et al. (2015a) explain that their generalizations
were based on numerical patterns and the serendipitous use of the factorial function, not on a
deeply understood or justified counting process. Thus, although the results in that paper provide
an instance of students coming to know basic counting formulas via reinvention, the students did
not attend to the combinatorial processes at play. We interpret that there remains a need for
more work that emphasizes student reasoning about combinatorial processes and structures
(especially in regard to fundamental problem types) in order to develop and justify key formulas
and ideas.
Our study contributes to the current literature by describing students’ productive understand-
ings of the basic counting formulas built from generalizations rooted in combinatorial activity
rather than attention to numerical regularity. In this way, we hope better to inform the ways that
students’ generalizing activities can interact with their combinatorial reasoning.
8 Z. REED AND E. LOCKWOOD

We also examine students’ potential ways of thinking (Harel, 2008a, 2008b) in combinatorics
and explore how these ways of thinking might influence students’ understandings of basic count-
ing formulas. For instance, Lockwood (2013) has emphasized the importance of having students
draw on the set of outcomes of a counting problem, which are the set of objects being counted.
Lockwood (2014) further emphasized this point by introducing a set–oriented perspective, where
the set of outcomes becomes a cornerstone of reasoning about any particular counting problem.
A set–oriented perspective is an example of a combinatorial way of thinking (in the sense of
Harel, 2008a), as engagement with sets of outcomes involves an organizational or structural char-
acteristic of the mental acts involved in solving these problems. Lockwood (2014) defined this
perspective as “a way of thinking about counting that involves attending to sets of outcomes as
an intrinsic component of solving counting problems” (p. 31). Generally, a variety of activities
could suggest that a student is employing a set–oriented perspective, including explicitly articulat-
ing one or more outcomes that a given counting problem is meant to count, creating a partial or
complete list of outcomes, or making a combinatorial argument or justification based on particu-
lar outcomes. This perspective aligns with our view of combinatorial thinking, and we considered
the importance of such a way of thinking as we designed our study and analyzed data.
We will contribute to the literature on student understanding in combinatorics by elaborating
three combinatorial ways of thinking that emerged during students’ initial activity solving four
different types of basic counting problems. A major goal of our research is to identify productive
ways of thinking in combinatorics, and, in this paper, to demonstrate mechanisms by which gen-
eralizing activities and combinatorial ways of thinking were related to students’ understandings of
general counting formulas. We point to three combinatorial ways of thinking rooted in our stu-
dents’ combinatorial activity, from which they then drew when generalizing to ultimately con-
struct the general counting formulas. This highlights ways in which these students’ thinking
address issues discussed in the literature of identifying and justifying formulas for counting prob-
lems (Batanero et al., 1997; Lockwood, 2014).

Mathematical discussion
Here we discuss the specific combinatorial problems with which the students engaged. This will
situate the mathematical context of the students’ activity. We had students solve four kinds of
problems during the teaching experiments. These are some of the same problem types students in
Lockwood et al. (2015a) solved: r–permutation problems, n–permutation problems, selection
problems, and arrangement with unrestricted repetition problems. These four problem types
make up the majority of the basic counting problems in introductory combinatorics texts (see,
for example, Martin, 2001; Rosen, 2007; Tucker, 2002) and lay the foundation for more compli-
cated counting processes (such as arrangement and selection with repetition). By further examin-
ing students’ constructions of the same problems in Lockwood et al.’s (2015) study, we offer a
contrasting case wherein students understand the basic counting problems combinatorially rather
than strictly numerically.

Arrangement without repetition (r–permutations)


In this problem type, we consider arrangements in which different orderings of elements result in
outcomes that we want to consider distinct, with the specification that elements cannot be
repeated. Such a problem may say, A town has 25 restaurants. How many ways are there to rank
your five favorite restaurants in that town? To answer this, we consider the number of options we
have for each of the five distinct spots that stand for the rankings – with the first spot being the
first choice, the second being second choice, etc. Because we cannot repeat elements, we get an
expression of 25  24  23  22  21, which yields 6, 375, 600 outcomes. The general formula for
COGNITION AND INSTRUCTION 9

arranging r objects from n objects is n  ðn  1Þ  ðn  2Þ  :::  ðn  r þ 1Þ, so there is a total of r


terms. This can also efficiently be expressed as ðnrn!
Þ!
, as a number of terms cancel.5 These are
called r–permutations (Tucker, 2002), and are often denoted as Pðn, rÞ:

Arrangement without repetition (n–permutations)


Observe that Pðn, nÞ ¼ n!, which is the number of arrangements of a set of n distinct objects, is a spe-
cial case of r–permutations. This also involves arrangement without repetition. An example of such a
problem is, How many arrangements are there of the letters of the word FAMILY?, and the answer is 6 
5  4  3  2  1, or 6!: We offer this type of problem as an independent category because students in the
study treated it as a different problem type with a distinct formula from r–permutations.

Selection without repetition (r–combinations)


For combinations, repetition is not allowed and outcomes with different arrangements of ele-
ments do not count as distinct outcomes. Combinations count selections (as opposed to arrange-
ments) of r objects from n distinct objects. We note that for any set of r objects, there are going
to be r! ways to arrange such objects (this is Pðr, rÞ). Thus, for however many selections there are
of r objects from n objects (what we are trying to count), we know that there are r! times as
many arrangements of r objects from n objects. Because we know (from above) that there are
n!
ðnrÞ!
, arrangements of r objects from n objects, we can divide this value by r! to find the total
n!
number of selections (rather than arrangements). Thus, there are ðnr Þ!r!
total selections (or com-
binations) of r objects from a set of n distinct objects. This solution leverages equivalence classes,
as the arrangements of r objects from n objects are partitioned into equivalence classes, where
the equivalence is of outcomes with arrangements of the same r elements. As an example, con-
sider the Subsets problem, How many 4–element subsets are there of the set 0, 1, 2, 3, 4, 5, 6, 7, 8,
9? Here we can count arrangements of 4 numbers from 10 numbers by considering the number
6! ¼ 10  9  8  7, and then divide by the ways to arrange
of options for each position, which is 10!
10!
the 4 elements within each arrangement of four numbers. There are thus 6!4! , or 210, outcomes.

Arrangement with unrestricted repetition


Here, we count outcomes in which elements can be repeated, and outcomes with differently
ordered arrangements of elements count as distinct. The general formula is nr , as we have n
options for each of the r positions. As an example, consider the License Plate problem, which
states, How many ways are there to make a 6–character license plate consisting of the letters A–Z
and the numbers 0–9? To answer this, we can consider the number of options we have for the six
distinct positions of the license plate, and we get an expression of 36  36  36  36  36  36,
which yields 2, 176, 782, 336 outcomes.

Methods
In this paper, we draw on two iterative teaching experiments (Steffe, Thompson, & Glasersfeld,
2000) conducted as part of a larger study that investigated the nature of students’ generalizations
in a variety of mathematical settings. We first conducted a paired teaching experiment (hereafter
referred to as ‘PTE’) consisting of fifteen 60 min sessions with two students. One year later, we
conducted a small group teaching experiment (hereafter referred to as ‘SGTE’) consisting of nine

n!
Þ! , in which we consider arranging all n objects of the set. But,
5
There is another useful interpretation of the quotient ðnr
since we only care about the first r positions, for each arrangement of the first r positions, the arrangements of the remaining
n–r positions actually constitute equivalent outcomes (hence the division by ðn–rÞ!).
10 Z. REED AND E. LOCKWOOD

90 min sessions with four students. In pursuit of our practical research goals, as well as the
broader goal of investigating ways that generalization contributes to students’ construction of
combinatorial understandings, we had the following research hypotheses at the onset of the PTE.
Essentially, we hypothesized that it would be productive to have students engage in some robust
combinatorial activity (facilitating a set–oriented perspective), then be given opportunities to
relate and extend based on that initial rich combinatorial activity.

1. Engaging in a set–oriented perspective can help students correctly solve counting problems
and distinguish between problem types and formulas.
2. Students can develop and reinforce desirable understandings about counting problems by
relating previously–solved problems via coordinating combinatorial activity.
3. Students can then extend such desirable understandings to new contexts by drawing on com-
binatorial patterns and regularities that have been established from organizations of men-
tal activity.

Our use of teaching experiments to test these hypotheses and pursue these goals is in line with
the overall goals of teaching experiments, particularly to “experience, firsthand, students’ mathem-
atical learning and reasoning” (p. 267), to “[make] essential distinctions in students’ ways and
means of operating” (p. 274), and to “[experiment] with the ways and means of influencing stu-
dents’ mathematical knowledge” (p. 273). This involves frequent hypothesis generation and test-
ing and produces models of student thinking generated from observing the students’ independent
mathematical activity (ibid.).

Participant description
As Steffe et al. (2000) point out, the “research hypotheses one formulates prior to a teaching
experiment guide the initial selection of the students … ” (p. 275). Because we were interested in
ways that students would construct understandings of the basic counting formulas through gener-
alizing from initial activities, we sought students that were unlikely to draw from prior con-
structed schemes for the general counting formulas when solving counting problems. We use the
term “novice counters” to describe such students. By novice counters, we mean students who
were not familiar with the four basic counting formulas and problem types that are the focus of
this study (discussed in Section 2.4 above).
While we cannot ensure that students have never seen combinatorics problems before (most
see at least some simple combinatorial tasks in middle school), during the selection interviews for
this study we had three ways to determine that these students were novice counters. First, we
asked what prior mathematics classes students had taken in high school or college. Novice coun-
ters had not taken probability or statistics, discrete mathematics, or computer science classes that
might have involved counting (some counting may occur in precalculus classes, but none had
taken an explicit course involving discrete mathematics). Second, we presented students with sev-
eral symbols, including some typically associated with permutations, combinations, and other
n
basic operations of counting. These were: nPr, Pðn, rÞ, nCr, Cðn, rÞ, n!, ð Þ, A  B, and
P r
1k10 k: We asked them if they had seen the symbols before, and, if they had, where they had
seen them and what the symbols meant. The only formula students recognized was n!, but this
was only in a calculus (Taylor series) setting, and they did not demonstrate having a combinator-
ial interpretation of it.6 Third, students solved multiple counting problems, which allowed us to
evaluate whether their mathematical activity was novel or whether they were drawing on extant

6
It is typical for students to be exposed to n! in calculus in a non–combinatorial context; see Lockwood & Erickson (2017) for
additional insight on students’ reasoning about factorials.
COGNITION AND INSTRUCTION 11

combinatorial schemes (e.g., recalling formulas). We considered further evidence that students
were novice counters if they did not try to apply a formula to these problems, but rather
attempted to list outcomes or guessed at an answer. Recruiting novice counters helped to ensure
that our students’ initial activity was novel, and that they were truly engaging in generalization
rather than relying on extant knowledge of the counting formulas to drive their activity.
We also sought students who demonstrated a willingness to engage in problem solving, a pro-
pensity to work well with others, and who were willing to articulate their thinking. In line with
these participant constraints, we recruited undergraduate students enrolled in a vector calculus
class at the time of enrollment in the study.7 We conducted individual, 30–60 min long selection
interviews which were audio and videotaped. A full list of selection interview tasks is in
Appendix A.
For the PTE, we chose two students (Rose and Sanjeev, pseudonyms) out of the total six inter-
viewees. We had the students work together so that it was more likely that they would look to
each other for confirmation and ideas when problem solving rather than to the teacher–re-
searcher. We paired Rose and Sanjeev together because they satisfied our criteria, and we inferred
from their selection interviews that they were similar in terms of mathematical ability, maturity,
background, and willingness and ability to express their thinking. We sought similar students
because we were engaging in initial testing of our hypotheses and wanted to minimize the varia-
bles of the experiment – we did not want one student to dominate based on their counting abil-
ity, for example.
For the SGTE, we chose four students (Carson, Anne–Marie, Aaron, and Josh, pseudonyms).
Here, our selection criteria were slightly different, and we sought a group of four students that
would represent a more balanced collection of students in a classroom small group.8 We still
sought novice counters who were willing to engage with the material and communicate their
ideas, but we also aimed to have a set of students who were more diverse in terms of their initial
proficiency solving counting problems. We measured this diversity in terms of their success in
solving problems in the selection interviews, and we sought both students that were and were not
able to successfully reason about most of the selection tasks. We attained this more diverse com-
position by roughly ranking the initial 10 students who participated in the selection interviews in
terms of their initial solutions, and we then chose a group of four that spread across our rankings
and whose schedules allowed for participation.9

Data collection
A teaching experiment (Steffe et al., 2000) includes some number of teaching episodes that con-
sist of several main elements: “a teaching agent, one or more students, a witness of the teaching
episodes, and a method of recording what transpires during the episode” (p. 273). Steffe et al. use
the term “teacher–researcher” for the researcher who plays the role of instructor in the teaching
experiment. Teaching experiment methodology allows researchers to “[experiment] with the ways

7
We have had positive previous experience working with vector calculus students. We have found that they are an excellent
population for combinatorics studies because they demonstrate some mathematical maturity but tend to have limited
background combinatorial knowledge.
8
When thinking of a whole class situation, we are thinking of a course where mathematics majors might first be exposed to
combinatorics. At many universities this would be in an introduction to proof course that is typically called Discrete
Mathematics, and vector calculus students are a reasonable population for such a course. We acknowledge that novice
counters may also be enrolled in less–advanced mathematics classes, and we discuss this further in the Future
Directions section.
9
Carson answered all but one problem correctly and demonstrated proficiency with organizing outcomes and early
engagement in what we later identified as positional reasoning. Aaron and Anne–Marie had successfully answered some of
the questions; they reasoned positionally in some instances and demonstrated thoughtful engagement with the set of
outcomes. Josh also demonstrated thoughtful engagement with the set of outcomes and could reason additively, but he did
not successfully coordinate the more complicated combinatorial situations.
12 Z. REED AND E. LOCKWOOD

and means of influencing students’ mathematical knowledge” and to “[understand] the progress
students make over extended periods”, making teaching experiments distinct from other experi-
mental methods such as clinical interviews (p. 273).
The 15 h–long sessions of the PTE occurred over the course of six weeks, and the students
were monetarily compensated for their time. During each session, the students worked together
at a chalk board. We gave the students counting problems to solve together and prompted them
to explain their thinking as they worked. While these sessions were generally designed to test the
three major research hypotheses, student activity within each session elicited responsive and ana-
lytical interactions on part of the researcher as hypotheses were further generated and tested
(Steffe et al., 2000, pp. 280). Further details on the tasks in the first three sessions are elaborated
in Section 3.4.
The SGTE data collection occurred one year after the PTE data collection. During the interim
time between experiments, we conducted retrospective analysis (Steffe et al., 2000) on the data
from the PTE (described below). An outcome of such analysis was refinement of the initial global
hypotheses of the experiment to include emergent productive ways of thinking that were identi-
fied in the PTE data analysis (further discussed in the results, Section 4.1). In particular, we
altered hypothesis 1 to include these productive ways of thinking that might potentially emerge
from student activity. As a result, we engaged in conceptual analysis (von Glasersfeld, 1995;
Thompson, 2008) to generate hypothetical models of these ways of thinking and incorporated
these models into our hypothesis generation and testing during the sessions.
This experiment consisted of nine 90 min sessions with the SGTE students.10 Instead of work-
ing on the board, the students sat at a table facing each other and wrote their individual work on
sheets of paper. The overall progression of the Categorization Activity was largely replicated with
the SGTE students; however, our interventions reflected the testing of our refined research
hypotheses. In particular, the “analytical interactions” (Steffe et al., 2000, pp. 281) of the teacher–-
researcher with the students were geared toward determining the extent to which the students
were engaging in the three ways of thinking identified from the PTE, and to test the boundaries
of their reasoning.

Data analysis
The sessions in both experiments were each videotaped, and we performed a retrospective ana-
lysis (Steffe et al., 2000) on the video records. The sessions were transcribed, and we enhanced
the transcripts with notes and figures representing student activity. For the purposes of this
paper, we focused our analysis on the first three sessions of both experiments, as those sessions
specifically pertained to the Categorization Activity and the students’ constructions of the basic
counting formulas. Results of this analysis are “living models of students’ mathematics” (p. 284)
from which theoretical constructs important to our research program “emerge in the form of spe-
cific and concrete explanations of students’ mathematical activity” (p. 293).
As described above, we draw on the specific constructs of Harel’s (2008) ways of understand-
ing and ways of thinking. To build these models of students’ mathematics, we identified the com-
ponents of the mental act–product–characteristic triad of Harel’s (2008a) Duality Framework for
cognitive activity. Specifically, Harel outlines a particular analytical process for identifying stu-
dents’ ways of thinking and ways of understanding that relies on attention to the particular men-
tal acts in which students are engaged:

10
One student could not attend the first session. He watched a video record of the first session, and he was instructed to also
solve the problems the other students solved and then listen to their discussion so he could be prepared for the
next sessions.
COGNITION AND INSTRUCTION 13

Table 1. Overall steps in the categorization activity.


Step 2: Opportunities for relating Step 3: Opportunities for forming and
Step 1: Initial problem solving activity through categorization extending through characterizing
Students are asked to solve specific Students are asked to categorize the Students are asked to characterize each
counting problems initially. problems they solved in any way group of problems they categorized.
they see fit.

1. Specifying the particular mental acts of interest (e.g., interpreting, proving).


2. Identifying products of each mental act in a form of utterances and actions; these are the
ways of understanding associated with the respective act.
3. Inferring common cognitive characteristics among multitude of products of each act; these
are the ways of thinking associated with the respective act. (2008a, p. 497).

In carrying out this analytical process, we relied on the students’ utterances and actions to
infer the nature of the activities they were carrying out that resulted in their particular under-
standings of the counting formulas in each of the three stages of the task. Over time, we also
made inferences about the recurring characteristics of their mental activity and understandings so
that their reasoning could be identified as adhering to certain ways of thinking that we further
identify in the results.
We in tandem identified when the students engaged in the generalizing actions of relating and
extending, paying attention to the particulars of the students’ mental organizations when general-
izing. More specifically, when the students engaged in relating, we paid particular attention to the
relationships being established by the students and to the initial ways of understanding being
coordinated in relating. Similarly, when analyzing the students’ engagement in extending, we
made inferences from their activity and utterances about the particular established relationships
and regularities being applied in a broader context, and the scope to which these relationships
were being broadened. In this way, we did not merely identify when they were engaging in gener-
alizing actions, but engaged in conceptual analysis (Thompson, 2008; Steffe et al., 2000) to infer
the ways the students were organizing their mental structures while generalizing.11 One applica-
tion of this analytical process was the identification of the students’ continual engagement in the
combinatorial ways of thinking both in their initial activity and throughout the generalization
process. Through this analysis, we identified the students’ process of generalization as an instance
of reflection on activity and examined the particular facets of their reflection.

Categorization activity
We now present the Categorization Activity, the broad sequence of tasks in which the students
engaged when constructing understandings of the four basic counting formulas. The
Categorization Activity consists of three phases: Initial problem solving activity, relating through
categorization, and extending through characterization (Table 1). We describe these and connect
them to specific generalizing activities they are designed to elicit in terms of Ellis et al.’s (2017)
R–F–E framework. In particular, these tasks serve as an experimental design from which to test
our research hypotheses. The goal of this task was to engage students explicitly in relating and
extending, and to give students opportunities to construct robust understandings of basic count-
ing principles. The intentional design to engage students in these generalizing actions reflects our
goal of examining the ways that supporting specific reflection on activity might elicit productive
understandings.

11
For the purpose of this paper, we report particularly on those instances of generalizing that were in conjunction with the
major phases of our experimental design according to the testing of our major hypotheses, noting that the students
generalized throughout.
14 Z. REED AND E. LOCKWOOD

Table 2. Arrangement with unlimited repetition problems solved by the PTE and SGTE students during their respective
experiments.
Problem name Problem statement Solution PTE SGTE
Houses You want to paint 6 different houses on the block, and 36 X X
there are 3 acceptable paint colors that you can pick.
How many possible outcomes are there for how to
paint the neighborhood?
License Plates How many ways are there to make a 6–character license 366 X X
plate consisting of the letters A–Z and the
numbers 0–9?
Quiz Questions How many ways are there to complete an 8–question 48 X X
multiple choice quiz if there are four possible answers
to each question?
3–Letter Sequences How many 3–letter sequences are there consisting of the 63 X
(repeat) letters a  f ?
Ternary Sequences How many 3–digit ternary sequences are there? 33 X

Table 3. n–Permutation Problems (arrangement without repetition) solved by the PTE and SGTE students during their respect-
ive experiments.
Problem name Problem statement Solution PTE SGTE
FAMILY In how many ways can you arrange the letters in the word FAMILY? 6! X X
Recess How many ways are there for 27 kids to line up for recess? 27! X X
Projects How many ways are there to assign 8 different projects to 8 different students? 8! X X
MATH In how many ways can you arrange the letters in the word MATH? 4! X

Initial activity
Throughout the first two sessions of each experiment, we gave each respective set of students the
problems in Tables 2–5. The aim here was to give students opportunities to engage in combina-
torial problem solving and initial activity on which they could subsequently reflect via the specific
generalizing actions of relating and extending. Tables 2–5 show the problems we gave each set of
students (the X in the last two columns indicates that we gave that problem to the students in
that teaching experiment).
Notably, this initial activity was an important part of the task design, as the students’ initial
combinatorial understandings would become the underlying cognitive material on which they
reflected during the next two phases. During this phase we probed the students for utterances
and activities from which we could test and generate hypotheses regarding their developing
understandings of the counting problems they were solving. We would frequently prompt for
exposition, provide further prompts to test the boundaries of their reasoning, and attempt to
draw out activity from which we could make inferences regarding their cognitive products and
characteristics. Differences between the PTE and the SGTE are due to conceptual analyses
between sessions.

Opportunities for relating through categorization and extending through characterization


After solving various problems of each type, we prompted the students to categorize the problems
however they saw fit. The goal was to facilitate student engagement in relating to construct rela-
tionships between the four problem types based on what we hoped would be their previous activ-
ity with the problems. Our hypothesis was that students would be able to leverage this
generalizing activity in order to make important connections, or build on those already estab-
lished, that would allow them to see what made problems similar to or different from each other.
Combinatorially, we hoped to address the issues raised in the literature of students’ difficulties
with distinguishing counting problems and formulas (e.g., Lockwood, 2014) by providing them
with opportunities to construct productive distinctions between problem types. With respect to
COGNITION AND INSTRUCTION 15

Table 4. r–Permutation Problems (arrangement without repetition) solved by the PTE and SGTE students during their respect-
ive experiments.
Problem name Problem statement Solution PTE SGTE
3–Letter Sequences How many 3–letter sequences are 6!
3! ¼654 X
(no repeat) there consisting of the letters a  f
, if no repetition of letters
is allowed?
Restaurants A town has 25 restaurants. How many 25!
20! ¼ 25  24  23  22  21 X X
ways are there to rank your five
favorite restaurants?
Horse Race There are 10 horses in a race. In how 10!
7! ¼ 10  9  8 X
many different ways can the horses
finish in first, second, and
third place?
Cats and Collars You have a red, a blue, a yellow, and 7!
3! ¼7654 X X
a purple collar. In how many ways
can you distribute the 4 collars
among 7 cats so that each cat will
wear at most one collar?

Table 5. Combination Problems (selection without repetition) solved by the PTE and SGTE students during their respective
experiments.
Problem name Problem statement Solution PTE SGTE
8!
Lollipops How many ways are there to distribute 3 identical lollipops to 8 children? 5!3! X X
Subsets How many 4–element subsets are there of the set f0, 1, 2, 3, 4, 5, 6, 7, 8, 9g? 10!
6!4! X X

their generalizing, we wanted to see whether they would look back on their initial activity at all,
and, if they did, to what aspects of their initial activity they would attend. Our goal was not sim-
ply to have the students engage in relating activity, but for us to gain insight into what and how
they would relate, hypothesizing that it would be productive to relate certain combinatorial activ-
ities rather than symbolic features or numerical patterns. After constructing the categories, the
students were prompted to explain why they grouped problems the way they did. In this way,
categorizing facilitated relating activity (in both the sorting and discussing) that would allow us
to observe and make inferences into the ways they organized their initial activity.
Once the students agreed on the categories for the problems, we asked them to describe the
specific problem type from each category, and to construct a general formula for the solution to
each problem type. By characterizing their categories, the students had opportunities to extend
the patterns and regularities they had established through relating.12
By observing the students’ generalizing actions and experimentally determining the boundaries
of their reasoning, we constructed models of the students’ reflection on activity according to the
constraints of our own reasoning and interpretations. These models find their viability in the degree
to which they offer means of explaining students’ independent mathematical activity as observed by
teachers and researchers in future studies and instruction (Steffe et al., 2000, pp. 298).

Results from the paired teaching experiment


In this section, we present illustrative examples of the PTE students’ work, moving chronologic-
ally through each of the three phases of the Categorization Task. We aim to present the students’

12
Forming (Ellis et al. (2017) implicitly took place at the onset of their engagement in characterization, as the students
attended to regularity in the structures within each group. This subsequently provided the means by which they could extend
their combinatorial understandings to make formal statements about the structure of each group and give general formulas
for solving the counting problems that each group represented. The students’ observable actions were manifestations of
extending, however, and so our analysis attends to the students’ acts of extending.
16 Z. REED AND E. LOCKWOOD

activities from which we infer salient aspects of their cognition, particularly highlighting productive
ways of thinking in combinatorics that emerged from our analysis the data. We also aim to high-
light the interactions between the students’ activities associated with three ways of thinking and
their engagement in relating and extending. We do not have space to elaborate the students’ work
on every task, but we provide examples of their work to give a sense of what the sessions entailed
and to showcase their reasoning and interactions. In describing the students’ initial activity in the
PTE, we highlight three ways of thinking that we associate with the students’ mathematics. We
identify and discuss these ways of thinking before presenting the students’ categorization and char-
acterization stages, so that we may then describe the influence of those ways of thinking both on
their generalizing activity and on their final understandings of the counting formulas. We then
explore the final two phases of the Categorization Activity and discuss the ways in which the stu-
dents leveraged these ways of thinking to construct the final counting formulas as a result of their
generalizing activity. In doing so, we provide a detailed account of their reflection on initial activity
and the leveraging of the students’ ways of thinking in this reflection.

Initial activity and emergent combinatorial ways of thinking


We began by having Rose and Sanjeev solve ten counting problems (seen in Tables 2–5). We
describe the three combinatorial ways of thinking that we contend Rose and Sanjeev were engag-
ing in throughout their initial activity.13 As we describe these ways of thinking, we offer brief
examples of their initial activity to demonstrate this sense–making and highlight important fea-
tures of their initial work for their future generalizing.14 In doing so, we hope to paint a picture
of the students’ activity and interactions and make a case for the ways in which these desirable
ways of thinking emerged during their initial activity. These combinatorial ways of thinking
(Harel, 2008a, 2008b) that influenced the students’ cognition as they generalized are: a) adopting
a set–oriented perspective (Lockwood, 2014), b) reasoning about positions, c) and thinking about
and leveraging equivalence. In subsequent sections, we will draw attention to the influence of this
initial combinatorial activity on their generalizations to highlight the interaction between their
specific generalizing actions and these combinatorial ways of thinking.

Sets of outcomes and a set–oriented perspective


First, we observed that the students’ activity reflected a set–oriented perspective (Lockwood,
2014), in that they were attuned to the set of outcomes they were trying to count. To illustrate
this, we begin by presenting their work on the very first problem we gave them, the 3–letter
sequences problem: How many 3–letter sequences are there using the letters A, B, C, D, E, F if
no repetition of letters is allowed? Rose had encountered a similar problem in her selection inter-
view,15 and she almost immediately solved this and suggested an answer of 6! 3! , which she also
wrote as 6  5  4.
Rose: [ … ] I think it’s just 6 factorial over 3 factorial, which would just be 6 times 5 times 4. Because you
have 6 letters to choose from. And then only 5 letters to choose from, like 4 letters to choose from and
then the last 3 letters could be in any order, it doesn’t matter, because you only want 3 letters.

13
We remind the reader that we do not conjecture the causality of the students’ developing these ways of thinking, nor do
we claim that they had fully adopted them as ways of thinking. Rather, we identify repeated engagement in mental acts that
we see as associated with these ways of thinking.
14
Due to spatial concerns, we can only present a representative sample of the students’ actions and utterances. Rose was
generally more succinct with her statements, and thus we present her responses more of the time than we do Sanjeev. In
accordance with Steffe et al.’s (2000) methodology, we probed and identified the boundaries of both students’ reasonings and
are confident that both students exemplified the kinds of reasoning we describe.
15
Rose was the first selection interviewee, and we refined the selection interviews after her interview, so Sanjeev saw a
slightly different set of problems.
COGNITION AND INSTRUCTION 17

Figure 1. Rose’s identification of duplicate outcomes.

In her response, Rose noted that the last three letters could be arranged in any order. We
asked the students to justify why that quotient made sense combinatorially (and not just algebra-
ically). Rose gave the following response.
Rose: Well, for the last 3 digits you could have … let’s call this just a,b, and c. You could have d,e,f. You
could have e,d,f or f,e,d [lists these sequences as in Figure 1]. I mean, these would all be considered
different arrangements, but since we’re only looking at the first 3 we don’t care if these are different
[crosses out d, e, f; e, d, f; f, e, d]. These [the final three letters] can be as different as they want, but as long
as these [the first three letters] are the same, then we can only show it once.

Notably, in this latter explanation we feel that Rose emphasized sets of outcomes as a way in
which to explain the division by 3!. She seemed to think of the arrangements of the letters of d,
e, and f as creating duplicates that she did not want to count. This discussion of duplicate out-
3! : This suggests that her division
comes was in the context of her justifying her initial answer as 6!
by 3! operationalized her intention to account for identical arrangements of the final three letters
in her desired outcome.
While solving this problem, Sanjeev also drew an array of possible outcomes, given in
Figure 2, and reasoned according to the arrangement of the outcome collection to justify the
solution 6  5  4. About this diagram, Sanjeev said
Sanjeev: “ … this spot right here would be the 4 options. And then within each row [meaning column] this
is – since all the As are used up, there’s 5 different options here, so that would be another 5 options. And
then because there are six letters, and each letter has its own – you can start with each letter, then you end
up with 6 up here.

This is an instance of Sanjeev reasoning about what we would consider to be the set of out-
comes. In particular, Sanjeev envisioned the collection of outcomes could be constructed in a
manner consistent with his partial listing of the outcomes, which is consistent with the mental
actions associated with a set–oriented perspective (Lockwood, 2014). The students then demon-
strated engagement in similar mental actions while solving each novel problem in the initial activ-
ity phase. The students would discuss the organization of outcomes either using a representation
18 Z. REED AND E. LOCKWOOD

Figure 2. A diagram of the outcome set.

of the “branching” that occurs through outcomes construction,16 or discussing lists of particular
outcomes as a means of discerning what Rose called a “pattern”. In sum, the students engaged in
mental activities that we associate with a set–oriented perspective while solving each problem in
the first phase of the experiment.

Positional reasoning
The students also discussed 6  5  4 as a meaningful representation of the solution by conceiving
of the problem in terms of considering options for each of three positions (i.e., imagining con-
structing an outcome by filling in “positions” or “spots”). Rose described what she was thinking,
and she explained the 6  5  4 by saying:
Rose: Well, for the first spot you have 6 different letters to choose from,
a, b, c, d, e, f . And then since you’re only allowed to keep the letters for the second spot you only have 5
letters to choose from, a, b, c, d, e, f minus whichever one you just used for the first one. And then for the
3rd one you only have 4 letters to choose from, because you’ve already used 2.

Sanjeev reasoned in a similar way. He started a process in which he was listing pairs of out-
comes, and first described making pairs starting with each letter. He continued by saying, “Then
I can go and say, okay, I’d like to add in a 3rd letter … So you only have 4 letters to choose
from so in this case if a and b are used up, then I get to choose between c, d, e, or f.” Sanjeev’s
explanation suggests he was reasoning about having certain numbers of options as he constructed
the sequences. In listing a partial set of outcomes, Sanjeev attended to sets of outcomes and set a
foundation for a relationship between his employed counting process and the generated out-
come sets.
This example demonstrates another combinatorial way of thinking upon which the students
drew, which we call positional reasoning. When solving counting problems, it can be useful to
think about generating a representative outcome by considering a counting process that involves
enumerating the choices for some given number of stages in the process, which can support the

16
We call this a tree diagram.
COGNITION AND INSTRUCTION 19

Figure 3. Students organizing a partial outcome (top numbers) and enumerating the options for each potential “spot” (bot-
tom numbers).

use of the multiplication principle.17 Generating the representative outcome might involve writing
representative elements for each stage in the process along with the number of choices available,
or it might involve simply listing the number of options at each stage. The particular cognitive
characteristics associated with positional reasoning is that the carrying out of the counting pro-
cess involves construction of a representative outcome, and the piece–wise construction of the
outcome can occur in singular stages (or “positions”, which Rose and Sanjeev called “spots”). Not
all problems are solved by reasoning about positions in this way, but for many problems it is use-
ful to reason positionally. This example also highlights the interaction of multiple ways of think-
ing, as we interpret that the students are both reasoning positionally and adopting a set–oriented
perspective.
For each problem in the initial activity phase, Rose and Sanjeev described constructing the
outcomes of the counting process through progressive “spot”–filling with elements of outcomes.
They would then negotiate the enumeration of the options for each “spot” according to the con-
straints of the problem. They would then either use this enumeration as counting the outcome
set, or an outcome set from which to build the desired outcome set for the problem.18 In each
problem beyond this example, the students explicitly drew out the “spots” they were describing
(see Figure 3), working with this representation as a means of organizing the progressive steps in
the counting processes they envisioned.

Thinking about equivalence


We also saw that leveraging equivalence was an important aspect of Rose and Sanjeev’s work in
the PTE. By equivalence, we mean the students understood that, depending on the constraints of
the problem, certain outcomes could be considered as equivalent (or duplicate or identical), and
that they could leverage division to account for equivalent outcomes. Elaborated further elsewhere
(Lockwood & Reed, 2020), this way of thinking involves reasoning about the sets of outcomes, as
the outcomes are the objects of “equivalence,” but it entails specifically thinking of certain out-
comes as being identical in some way. Technically, such sets of outcomes should be considered to
form equivalence classes, and division allows us to count the number of equivalence classes, each
of which has a constant size. Although we do not expect students to use the language of equiva-
lence classes explicitly,19 we consider the structure of an equivalence class to be a useful model to
describe such student thinking. In Rose and Sanjeev’s work, this came up most prominently in
their reasoning about combinations and permutations. For example, in the exchange below the

17
The idea behind the multiplication principle is that for successive stages in a counting process, one can multiply the number
of options for each stage as long as the number of options is independent of each previous stage and the composite
outcomes are all distinct. We prefer Tucker’s (2002) statement of the multiplication principle; see Lockwood, Reed, &
Caughman (2017) and Lockwood and Purdy (2019a, 2019b) for additional information about the multiplication principle.
18
For combination problems, this was the first step of their counting process, from which they would further partition
according to “duplicate” outcomes, see below.
19
Here they used the language of “duplicates”.
20 Z. REED AND E. LOCKWOOD

students were solving the Subsets problem (which asks for the number of 4–element subsets of
the set of numbers from 0 to 9) we see that they were reasoning about dividing out by
“duplicates” as a way to count the desired outcomes.
Sanjeev: How many 4–element subsets are contained in this set [f0, 1, 2, 3g]? So, 0,1,2,3 … would be the
same as 3,2,1,0 –
Rose: – 0, yes. … So, we just need to figure out how many different variations you can make from one
group [of four elements], and divide by that.

They had found that there were 24 possible arrangements (or “different variations”) of four
numbers, and they clarified that they should divide.
Sanjeev: … sorry, of those 24 possible arrangements –
Rose: So, it would just –
Sanjeev: – we only want 1.
Rose: We only want 1 –
Sanjeev: Right.
Rose: – because they’re all the same. So, we would just take that whole thing right there [their current
6!  and just … so it’s just 10987 divided by 24.
calculation 10P4 ¼ 10!
Sanjeev: Divided by 24. … This, then, accounts for duplicates.
Rose: Yeah, that [dividing 10! by 6!] accounts for the spots that we don’t care about, and that [subsequently
dividing by 24] accounts for duplicates.

We suggest that Rose and Sanjeev were engaging in mental acts that involve coordinating a
collection of equivalent outcomes so that a single representative generates their desired collection
of representative outcomes. In this example, Rose and Sanjeev leveraged equivalence in two dis-
tinct ways. First, 10! represented the number of ways to arrange all elements in a 10–element set,
and so the division by 6! accounted for the equivalence of those arrangements that reorder the
last 6 of 10 elements, as only four elements were desired (this is similar to Rose’s argument on
the 3–letter sequences problem discussed previously). Second, as more explicitly discussed in the
excerpt, they acknowledged that a set of the form f0, 1, 2, 3g is “the same” as a set where the ele-
ments are rearranged. Since every 4 elements can be rearranged 24 times, they only want a single
outcome for every 24 arrangements, and so the division by 24 accounts for the “duplicate” out-
comes of rearranged 4–element subsets.
The students first independently attended to outcome equivalence in the Lollipops problem
(the third problem they solved). For each subsequent problem, the students made sure to verify
whether there were particular outcomes that they could conceive of producing via a counting
process but would be considered “duplicates” under the constraints of the problem. When appro-
priate, the students would then employ division to re–scale their outcome set enumeration to
account for such duplicates. They also discussed duplicates generated from including “unwanted
spots” specifically for the r–permutation problems.20 In sum, the students incorporated attention
to duplication of desired outcomes into their strategy for solving counting problems, and consist-
ently operationalized outcome–duplication when necessary.
To summarize their initial activity, we contend that Rose and Sanjeev demonstrated engage-
ment in activities that we associate with three ways of thinking: a) thinking about sets of out-
comes, b) reasoning about positions, c) and thinking about and leveraging equivalence. We now
continue with a description of their categorization and characterization stages in the PTE, in
which they constructed the general counting formulas. We will demonstrate that the students

20
These discussions were akin to Rose’s justification that 6!
3! and 6  5  4 both counted the 3–letter sequences of the letters a –
f, presented in Section 4.1.1.
COGNITION AND INSTRUCTION 21

Table 6. Rose and Sanjeev’s Categorization of problems in the Categorization Task (parenthesis indicate researcher labels).
Group Problem name Problem statement Solution
Group 1 (n- FAMILY In how many ways can you arrange the 6!
permutations) letters in the word FAMILY?
Recess How many ways are there for 27 kids to 27!
line up for recess?
Projects How many ways are there to assign 8 8!
different projects to 8
different students?
Group 2 3–Letter Sequences How many 3–letter sequences are there 654
(r–permutations) consisting of the letters a – f , if no
repetition of letters is allowed?
Restaurants A town has 25 restaurants. How many 25!
20! ¼ 25  24  23  22  21
ways are there to rank your five
favorite restaurants?
8!
Group 3 Lollipops How many ways are there to distribute 3 5!3!
(r–combinations) identical lollipops to 8 children?
10!
Subsets How many 4–element subsets are there 6!4!
of the set f0,1,2,3,4,5,6,7,8,9g?
Group 4 (arrangements Houses You want to paint 6 different houses on 36
with the block, and there are 3 acceptable
unrestricted repetition) paint colors that you can pick. How
many possible outcomes are there for
how to paint the neighborhood?
License Plates How many ways are there to make a 366
6–character license plate consisting of
the letters A–Z and the numbers 0–9?
Quiz Questions How many ways are there to complete 48
an 8–question multiple choice quiz if
there are four possible answers to
each question?

continued to organize and engage in the same mental acts while generalizing, and that their con-
tinued engagement in the mental acts associated with these ways of thinking served to regulate
their generalizing.

Relating via categorization


As described in Section 3, the students then sorted the problems from the first phase. Specifically,
the interviewer asked, “ … we noticed that you used some similar approaches to solve some of the
problems, approaches and methods, so I wonder if you can categorize the problems according to
your solution method.”21 The students first physically sorted slips of paper on which the prob-
lems were written, and then they also spent some time discussing and trying to describe how
they had established those categories, which we discuss in terms of relating in the R–F–E frame-
work (Ellis et al., 2017). The actual sorting and categorizing happened quite quickly – in 2 1/
2 min they had established their categories. Table 6 lists the categorization and the groups to
which the students referred.22
After they had completed the categorization, we asked the students how they were thinking
about their activity and to what they were attending as they created the categories. In this section,
we provide several examples of their relating activity. In the initial categorization they briefly con-
structed relations between the different counting problems, and then by explicating their groups,
they established relations of similarity (and differences, as we will see). We will particularly

21
Although we phrased this prompt to suggest that the students attend to their activity when relating, we do not consider
our phrasing alone to affect students’ cognitive activity significantly. Rather, we consider the students’ mental coordinations in
the context of these prompts as the primary object of analysis.
22
The students made one change during the discussion of these groups, but the change was due to a misreading of the
wording on the Subsets problem.
22 Z. REED AND E. LOCKWOOD

highlight the students’ organizations of their mental activity while establishing and explicating
their perceived relations between counting problems. In this way, it is the students’ cognitive
activity elicited through their engagement in relating that is the focus of our analysis. Moreover,
we want to highlight that the students continued to demonstrate cognitive characteristics associ-
ated with our identified ways of thinking while generalizing.
We found it noteworthy that the students highlighted both similarities and differences as they
engaged in relating and categorizing the problems. With regard to our practical goal of identify-
ing ways that students might make meaningful distinctions between counting problems, we find
it useful to consider both comparing and contrasting as engagements in relating. That is, for the
purpose of model building we found it useful to consider relating as involving attention to mean-
ingful differences across situations while attempting to regulate the ways in which these situations
might be similar. We now briefly highlight some of these comparisons, which demonstrates how
the students were reasoning about separating the problems into groups. We do not have space to
exemplify every comparison they made, but we provide examples of and elaborate the students
comparing Group 1 versus Group 2, Groups 1 and 2 versus Group 4, and Groups 2 versus
Group 3 (see Table 3 for which set of problems the group numbers refer). By doing this, we are
not just demonstrating that they related, we want to draw attention to what and how they were
relating as they categorized problems.

Group 1 (n–permutations) versus Group 2 (r–permutations)


Rose engaged in positional reasoning to interpret the combinatorial processes outlined by Groups 1
and 2. For instance, describing the Recess problem (a Group 1 problem) she said: “ … it’s just 27
students into 27 spots” and contrasted that with Group 2: “For these 2 down here (Group 2 prob-
lems), where you have less spots than you do objects to put in them, it was a factorial divided by a
factorial. That was the answer on both of these (Group 2 problems).” She concluded that the second
group of problems involved arranging objects into fewer spots than objects to put in them. This
suggests that positional reasoning contributed to her means of differentiating Group 1 from Group
2 by envisioning differences in the construction of their representative outcomes.
While engaging in relating, the students’ mental activity involved coordinating perceived simi-
larities and differences between combinatorial activities that were necessary for carrying out the
arrangement process required for the respective groups. In this case, their relating elicited mental
acts of differentiation between envisioned combinatorial processes, highlighting a distinction
between arranging an entire set of elements versus some subset of a set of elements. In these
examples and other instances, they appealed to positional reasoning to articulate such distinc-
tions, demonstrating the use of this way of thinking to regulate their generalizing.

Group 1 (n–permutations) and Group 2 (r–permutations) versus Group 4 (arrangements with


unrestricted repetition)
As another example of their relating activity during this categorization phase, when they consid-
ered the group of problems that contained the License Plates problem (Group 4), the students
highlighted a difference with the permutation problems (Groups 1 and 2). Rose explained the dif-
ference between these groups.
Rose: And the difference between these like [the License Plate, a Group 4 problem] and the other ones
[permutations problems, Group 2] is like once you – like for the license plate one, once you use A for the
first character you can still use A again.
Interviewer: Okay. Awesome.
Rose: So, it’s different from like these ones [references Groups 1 and 2], for example, where if you’ve used a
letter for the first spot, well, now, you’re making arrangements, so for the second spot you can’t use that
letter, because you only have one of those letters.
COGNITION AND INSTRUCTION 23

Here, to compare problems, Rose distinguished between whether or not options could be
repeated, and she also considered what was allowed in the particular positions. We find this kind
of distinction productive and in accordance with a set–oriented perspective and positional reason-
ing, and it seemed to be meaningful for the students because they were themselves formulating
and articulating these distinctions. In particular, the students explicitly articulated that Group 4
involved constructing outcomes without restricting elements of the outcome. We infer that,
through their relating, the students’ organized their ways of understanding the various
“arrangement with unrestricted repetition” problems from their initial activity according to the
combinatorial mental act of outcome construction with particular attention to the lack of a need
for outcome restriction. Moreover, the students created mental boundaries between their Group 4
and their other groupings according to this mental act.
Their uses of positional reasoning and a set–oriented perspective further highlight the role that
ways of thinking can play in regulating the students’ constructed understandings while engaging
in the generalizing action of relating. In this case, distinguishing between these groups entailed
the re–envisioning of the activity through which the outcomes were constructed, and we argue
that this enabled their engagement in ways of thinking they had used to solve such problems in
the first phase. From this we infer an interplay between the students’ demonstrated ways of think-
ing and their engagement in the generalizing action of relating.

Group 2 (r–permutations) versus Group 3 (r–combinations)


Finally, attending explicitly to the sets of outcomes and equivalence allowed Rose and Sanjeev to
group the combination problems together meaningfully, and to differentiate them from the per-
mutation problems. The discussions about differentiating combinations from other combinatorial
processes revolved around taking care not to count two identical outcomes as distinct. For
instance, when separating the slips of papers with the permutations problems and the combina-
tions problems, Rose and Sanjeev discussed the following:
Rose: It’s [Group 2] – it’s how many – it’s basically how many ways to put certain amount of items into
fewer spots where 1, 2, 3 and 3, 2, 1 are different. And this [their Group 3 problems] is how many ways
you put a certain amount of things into fewer spots where 1, 2, 3 and 3, 2, 1 are the same.
Sanjeev: On these ones [Group 2] you’ve got combinations [not in the typical sense of the word]. So, 1, 2, 3
– 3, 2, 1 would be different combinations. With this one [The Lollipop Problem, a Group 3 problem], for
example, if you have identical lollipops you can label them 1, 2, 3 or you can just label them 1, 1, 1. So 1,
2, 3 and 3, 2, 1 would be the exact same thing, because 1, 2 and 3 are all the same.

We point out that they appealed to actual outcomes to articulate the difference between
groups. The distinction of making f1,2,3g and f3,2,1g the same explicitly attends to which out-
comes should or should not be counted as similar. We contend that they reasoned with sets of
outcomes to understand the differences between combinations and permutations. Further, their
language suggests the influence of positional reasoning and equivalence. In accordance with the
previous two examples of the mental organization of their past activity while engaging in relating,
here we see the students’ leveraging of equivalence served to establish meaningful relationships
between the problems where outcome orderings were considered the same, and simultaneously
establish a meaningful difference between problems where this was not the case. Specifically, the
students understood the collection of combination problems as groupings of equivalent outcomes
organized according to non–equivalent groupings and understood these groupings as a boundary
feature of these problems. This further illustrates the students’ leveraging their ways of thinking
to regulate the relationships created during their relating activity.
24 Z. REED AND E. LOCKWOOD

Summary of the categorization phase


The students’ engagement in categorizing leveraged the specific generalizing activity of relating
(Ellis et al., 2017) as a means of attending to regularities in their initial activity that then allowed
them to identify similarities and differences between combinatorics problems. In particular,
through relating, the students understood their previously solved problems as being similar (or
different) according to specific combinatorial acts in which they had engaged in their initial
activity. Further, when relating, the students leveraged their ways of thinking to formulate and
then refine the distinctions between groups. Often this comparison involved identifying differ-
ences between groups, which seemed to help refine the boundaries between groups of prob-
lems. We find the ways that the students were identifying boundaries between groups to be
productive in that their distinctions were products of their combinatorial mental actions, and
particularly distinct from ways that students have distinguished and understood counting prob-
lems in the literature. In this way, we find the students’ generalizing to be productive not sim-
ply because they engaged in relating, but rather that in their relating they attended to
coordinations of re–envisioned activity. We view this act of emphasizing differences as contri-
buting to a productive understanding of the counting formulas that they would generalize in
the next characterization phase of the Categorization Activity, as the differentiations were
rooted in carrying out different combinatorial activity. The students seemed to benefit from
this generalizing activity – specifically, they developed means of identifying the salient com-
binatorial mental acts that comprised their initial activity. This relating activity in which they
engaged set the stage for their final generalizations to be formal extensions of their ini-
tial activity.

Extending via characterization


After the students had created categories as described above, we then asked them to try to
describe and characterize what each group of problems was counting. The interviewer asked the
following: “So first of all if you just had to like describe a problem of this type, sort of in general
terms, what – how would you describe it? Like I want to know what kind of problem type is
this? What would you say?” This prompted them to formalize their perceived regularities through
extending (Ellis et al., 2017). 23
The students’ extending entailed articulating the general descriptions of problem types, and
the construction of general counting formulas. We present their final characterizations to offer
evidence of the ways they understood the counting problem types and their formulas. We view
these final counting formulas as products of the same mental acts that the students consistently
applied in their initial activity and regulated while relating, which became applied to a broader
range of combinatorial scenarios. In this way, the students’ extending supported their understand-
ings of the counting formulas as the products of refined mental acts, deriving from an expansion
of their initial activity (Ellis et al., 2017). Moreover, their descriptions of the categories further
suggest the influence of their ways of thinking in the extending process as well, again demonstrat-
ing a connection between student’s engagement in these generalizing actions and their ways of
thinking. We describe the final characterizations and formulas here for the sake of efficiency,
although in the PTE they came up with the formulas for each group after writing the character-
izations for all of the groups. We want to highlight that they drew on these ways of thinking as
they extended.

23
We note that they engaged in forming as well, but their observable utterances and actions were manifestations of
extending. We thus focused on extending in our analysis.
COGNITION AND INSTRUCTION 25

Figure 4. The students’ characterization and formula of Group 1 problems.

Figure 5. The students’ characterization and formula of Group 2 problems.

Arrangements without repetition (Group 1)


To characterize this group of problems, the students wrote Figure 4 on the board, which says,
“How many ways to arrange a given number of elements.” Sanjeev clarified that “you assume
that the number of spots is the same” as the number of elements .
In explaining their formula for arrangements, Sanjeev said,
Sanjeev: So you take the total number of elements you have, and then you look at how many spots there
are for each of those elements. And then you figure out how many possible ones you can place into the
first spot. And then since you’re not reusing, then you figure out how many are left in the second spot,
how many are left in the third all the way until you’re out of spots.

We infer from this discussion that he was leveraging positional reasoning. Rose also clarified
that n represents “the amount of objects that we have” and said the answer for this kind of prob-
lem “would just be the amount of objects you have factorial.”
Importantly, their characterization attended to there being a certain number of positions
for specific outcomes, and the number of options per position decreases with each successive
position choice. This suggests that their understanding of n! was informed by their applica-
tion of positional reasoning to structure the specific combinatorial process they
described above.
In constructing this characterization, they extended the mental act of arranging from specific
instantiations to account for any collection of n objects. Through relating, the students attended
to the regularity in their arrangements via progressively filling slots to construct outcomes in
Group 1, and they delineated arrangements from sub–arrangements in Group 2. The students
then extended this envisioned arranging activity to be applied to any “given number of ele-
ments”, thus expanding the range of applicability of this coordinated mental act. Moreover,
through delineating Group 1 from Group 2, they extended those specific mental acts corre-
sponding to a complete arrangement of objects, carrying out a constructive process regulated by
their positional reasoning. We include this detailed analysis as an example of the specific reflec-
tions the students enacted on their initial activity in response to relating and extending, as well
as the regulatory nature of their engagement in positional reasoning. As an additional note, we
comment on Sanjeev’s use of “spots” language and his implicit reference to building out an out-
come based on the process he described. Generally, we would not consider the use of slots,
spots, or positions alone as being clear indicators of a set–oriented perspective, as, in our expe-
riences, students may use such language to convey multiple steps in a process or stages in an
26 Z. REED AND E. LOCKWOOD

operation, which may or may not be linked to outcomes. However, because we can analyze
other work by Sanjeev, we have prior evidence of his understanding that such a process gener-
ates outcomes. We thus consider Sanjeev to be demonstrating engagement in a set–oriented
perspective here.

Arrangements without repetition (Group 2)


In characterizing permutations, Sanjeev wrote down Figure 5 on the board, which says, “How
many ways to arrang[e] a given number of elements into a given number of spots without reus-
ing any elements.”
Rose agreed and said, “ … that makes sense, so that you don’t end up with 5, 5, 5, 5” (referring
to the Restaurants Problem). The following exchange further demonstrates their thinking.
Interviewer: Sweet. Okay. And what do you mean by “a given number of elements” and “a given number
of spots”?
Rose: Well, you – you’d be given, say 8 elements, and you’ll be – you have 3 spots to put that in.
Rose: So, the “given numbers” can be different.
Interviewer: They can be different? Okay.
Rose: If they’re given the same, then it would just be problem number 1 [referring to Group 1] … . Because
then you’d have as many elements as you did spots, and –
Sanjeev: So, in problem 1 you’re assuming the number of elements equals the number of spots in your
arrangement. Whereas problem number 2 [Group 2] you’re saying this is how many elements you have to
choose from, but the number of spots we’re giving to you is different from that.

The underlined portion of the excerpt shows that the distinction the students established when
relating was being extended to the general setting. The students’ positional reasoning informed
this distinction, and their extending resulted in an understanding rooted in their arranging of a
given number of objects into various slots. We infer that the students were envisioning similar
constructive processes as in Group 1, but with the key distinctions they outlined, because of their
associations of Group 2 with Group 1. In this way, we again see that the boundaries between the
categories that the students had established in the relating phase afforded an opportunity for
them to attend to the specific construction of the outcome of a permutation process.
Sanjeev’s comments below also suggest that they engaged in an equivalence way of thinking to
regulate their generalizing. For the formula of Group 2, the students wrote the formula in Figure
5, which Sanjeev explained as follows.
Sanjeev: So, you have your total number of elements that you can arrange into a certain number of places.
And then you have a certain number of places that you care about. And so, if you have a certain number of
elements, and you have a certain number of places that you care about, then you want to take the number
of elements minus the number of places you care about to get the result of places you don’t care about.
And then you divide out the number of places you don’t care about.

We note that for this formula, they communicated that division eliminated the “places you
don’t care about.” This use of division allowed them to focus on just some (but not all) of the
positions, which they viewed as a key distinguishing feature of this problem type. Their com-
ments suggest that their appeals to equivalence when defining the boundaries of the groups
became an integral component of the way they understood the general permutation formula and
its corresponding problem type. Thus, while extending the students not only leveraged their
equivalence way of thinking, but also formalized the conditions under which they could more
generally employ equivalence in counting. Again, we see that extending further elicited engage-
ment in the students’ ways of thinking to refine their understandings of the problem types and
general formulas.
COGNITION AND INSTRUCTION 27

Figure 6. The students’ characterization and formula of Group 3 problems.

Selection (Group 3)
The articulation of combinations seemed to be a bit more difficult for the students to reason
about, but they ultimately used particular outcomes to clarify that in this set of problems they
did not want to count 3, 2, 1 as the same as 1, 2, 3. The students extended the distinctions they
had previously made in the categorization phase. Rose discussed how these problems in Group 3
related to problems in Group 2.
Rose: – I’m saying that this [Group 3] is the same as number [Group] 2 except you’re getting rid of all the
repeated sequences.
Sanjeev: So, we could say how many ways to arrange a given number of elements into a given number
of spots.
Rose: Without reusing any elements, and technically to get rid of the amount of repetition you can just
divide by the factorial of how many spots you have. Because that – because the factorial of 4 spots is 24,
and you’re going to have 24 sequences that are the same.
Sanjeev: Uh–huh (in the affirmative).
Rose: So, for 3 I can just write like dot, dot – [ … ]

To characterize the Group 3 problems, the students attended to a previously constructed rela-
tionship between Groups 2 and 3 that had been identified when relating. In addition to distin-
guishing between permutation and combination problems via equivalence, the students also
viewed the combination formula as an extension of the process they had carried out in solving
permutations. Indeed, notably their description of combinations continued from the Group 2
characterization (as indicated an arrow in Figure 6). After some discussion, they settled on the
final articulation “ … and divide by the factorial of the given spots to delete repeated sequences
because any arrangement of the ne ! same given elements is considered the same combination,” and

they wrote the final formula ðnenns !s Þ! : The students clarified that ne was the number of elements and
ns was the number of spots. We see their final statement of justification as the actual articulation
of the problem type for Group 3.
Their exchange demonstrates how they were reasoning about the formula.
Rose: The third one starts off pretty much the same [as Group 2]. You want to figure out how many objects you
can put into how many spots, when the number of spots is less than the number of objects. So, it starts out with
the same formula, except this time we have to get rid of the duplicates, because 1, 2, 3 is the same as 3, 2, 1.
Sanjeev: So, let’s start with that. So, we have the number of elements that you can completely arrange, and
you divide out the number of elements you’re going to want to not focus on. So, it’s going to be the total
minus the number you do want to focus on. And then you have – you have some result of arrangements.
But now within each of those arrangements that you’re now focusing on, you want to divide out the
number of duplicates.
Rose: Yeah.
28 Z. REED AND E. LOCKWOOD

Sanjeev: So, in – in this case you’re assuming that within each of those arrangements every single one of
those arrangements is the same. So now you have like a subset of your arrangements. And in that subset
every single one of them is the same, so you want to divide it again.
Interviewer: Okay.
Sanjeev: So you take this, and you divide that out by –
Rose: I’d say it’s ns factorial.
Sanjeev: Factorial.
Rose: Because that’s the number of spots you have, and if you have 4 elements going into the same 4 spots
you’re going to get n factorial duplicates out of that.

In this exchange we see that the students conceived of solving these Group 3 combination
problems as first counting arrangements (Group 2), and then dividing out to account for dupli-
cates (see the underlined portion above). They seemed to understand that the need to account for
possible duplicates differentiated these problems from arrangements, and they understood the
role that division played in accounting for duplicates. This further highlights the utility of the
comparing and contrasting they engaged in while relating, in that restating the refined boundaries
between the groups allowed them to coordinate and articulate a productive way of distinguishing
combinations upon construction of the general problem type.
We maintain that the students’ use of their own notation and language suggests that they were
developing personal understandings of the formula and articulating a formula that they could
understand. This is natural given that they had no knowledge of the combination formula prior
to this experiment. Further, we also suggest that attending to the set of outcomes provided a
meaningful distinction between combinations and permutations, and they leveraged equivalence
to achieve division that accounted for the duplicates.
In this instance, the students extended the notion of accounting for “duplicates” when order-
ings of the elements in the final outcome were inconsequential. The characterization the students
put forth was a formal articulation of the means by which they differentiated between Groups 2
and 3 in the categorization phase. This again underscores ways in which the students’ engage-
ment in the generalizing actions of relating and extending were supported by the developing
ways of thinking from their initial activity.

Arrangements with unlimited repetition (Group 4)


Finally, the students considered their fourth group of problems, and they wrote down the charac-
terization, “How many ways to put a certain amount of thing[s] into a certain amount of places
assuming you can put the same thing in more than once place,” and they wrote the formula nc ne
(Figure 7).
The students justified their formula in the following way.
Rose: Well, this one’s like – instead of you put something in a spot, and then you have 1 less options for
the next spot – you just have the same number of options. So, instead of a factorial, we’re going to raise it
to the power of something, because instead of like 25 times 24 times 23, it’s just 25 times 25 times 25.
Interviewer: Okay. Great.
Rose: So, because you have – so we’re going to have how about n elements, and we can have n choices
(writing on blackboard) we would want – it would be the number of choices you have raised to the power
or the number of elements you have, because the first element you have n choices – and then for the
second element you have multiplied by n choices. And then for the third element you have multiplied by n
choices. So, I think that makes sense.

Here Rose demonstrated positional reasoning. Note that in Rose’s first line, she was engaging
in relating by comparing Group 4 with Groups 1–3, and she contrasted the problem with a
COGNITION AND INSTRUCTION 29

Figure 7. The students’ characterization and formula of Group 4 problems.

problem involving factorial, explaining how in this situation the number of options would not
decrease. We additionally point out that both students arrived at the formula and discussed its
form (Rose noted, “instead of factorial, we’re going to raise it to the power of something”) but
they understood exponentiation not just as some figurative formula but as representing the multi-
plicative process supported by positional reasoning.
Thus, we contend that Rose’s articulation of their formula reflected the combinatorial activity
from which it was extended, and it also suggested engagement in positional reasoning. More spe-
cifically, the students’ descriptions of the Group 4 problems are indicative of the same mental
acts used for iterative–outcome construction in which they had engaged during their initial activ-
ity. While engaging in relating and extending, the students isolated the particularly meaningful
characteristics of their arrangement activity and then broadened the applicability of those isolated
mental acts.

Summary of the PTE data from the categorization activity


To summarize the PTE data, the students organized counting problems into four groups, on
which they then built to generate four distinct counting formulas that could be used to solve
problems more generally. We contend that the PTE students’ generalizing activity provides a
detailed account of reflection on activity while highlighting potential relationships between gener-
alizing and ways of thinking. Specifically, the models of student mental activity presented above
are indicative of specific means by which students’ reflection on activity was enacted through the
generalizing actions of relating and extending. Rose and Sanjeev first carried out a collection of
mental acts whose products constituted their ways of understanding the various counting prob-
lems, such as successively constructing outcomes through unrestricted choices from a set of
options into a number of “slots”.
Engaging in relating, the students then re–envisioned these mental acts for the specific purpose
of establishing relations between problems that elicit these mental acts, and the students com-
pared features of their mental acts for the purpose of establishing boundaries between groups of
problems they had solved. An example of this is the construction of Group 4, in which they iso-
lated the process of constructing outcomes with unrestricted repetition as meaningfully distinct
from the process of constructing outcomes without repetition found in the other groups.
Once the students isolated certain processes and features of their initial activity, they then
broadened the applicability of these organized mental acts, identifying the particular features of
counting problems that would occasion the enactment of these mental acts. Again, the students’
construction of the Group 4 category exemplifies this phenomenon, as the students described the
same iterative outcome construction that they used in their initial activity. From this, we infer
that the students envisioned enacting those same mental acts on a broader class of problems.
The students’ generalizing activity constitutes specifications of the reflection process that is
unique to engaging in relating and extending, but it also highlights the dependence of the coord-
ination of the different mental acts on this type of reflection on activity. To this last point, a
30 Z. REED AND E. LOCKWOOD

unique feature of these students’ reflection on activity is the interdependencies of the students’
understandings of each problem type in their constructive process. Specifically, whereas other
reflection on activity might more specifically feature reflection on regularities of singular mental
acts in isolated iterative stages, the students’ reflection on activity specifically featured the estab-
lishing of barriers between different types of mental acts as a core feature of their reflection. We
elaborate this point in the discussion.
Finally, the students engaged in specific combinatorial ways of thinking in each phase, which
highlights a novel synergy between ways of thinking and generalizing actions as together contri-
buting to the students’ refined ways of understanding the final problem types and general count-
ing formulas. By explicitly leveraging generalizing actions in this way, we have detailed the kinds
of ways of thinking that helped students articulate problem types and produce these formulas.
This detailed account of their generalizations allows us to identify the ways the students under-
stood the counting formulas and allows us to attribute their understandings of the general formu-
las to the ways that they engaged in initial activity, the developing ways of thinking implemented
in generalization, and the ways in which their generalizing activity both elicited and drew from
their engagement in those ways of thinking.

Small group teaching experiment


In this section, we provide a brief depiction of the Small Group Teaching Experiment (SGTE),
which we conducted with Carson, Anne–Marie, Aaron, and Josh. While we are limited by space,
our goal is to demonstrate that the models developed from Rose and Sanjeev’s teaching experi-
ment are useful explanatory mechanisms for describing and characterizing student thinking more
broadly. Specifically, we want to demonstrate that the PTE students’ developed understandings
and ways of thinking were not just happenstance or a lucky anomaly, but rather that these phe-
nomena were also inferred from other students’ activity on the same set of tasks, and thus these
constructs potentially have explanatory power for application in future research.
We also see this additional iteration as a way of strengthening what we contend is a theoretical
contribution about the nature of generalization, namely that generalizing actions can be closely
connected to students’ ways of thinking. As with the PTE students, the SGTE students’ final
understandings were refined by the students’ engagement in combinatorial ways of thinking
throughout their generalization.
The SGTE students’ activity closely resembled that of the PTE students. In particular, they
completed the same categorization activity (and also engaged in relating) and the same character-
ization activity (and also engaged in extending) as did the PTE students. They even categorized
the problems in the same way, so Groups 1 through 4 below represent the same four groups dis-
cussed in the previous section. Due to space, we only present the SGTE students’ final character-
izations, share their utterances describing the problem types, and offer a brief history of their
engagements in positional reasoning and equivalence to highlight the influence of these ways of
thinking on their work. We note that throughout the students’ description of their categories,
they referred to outcomes and used construction of outcomes as a means of reasoning about their
formulas, thus adopting a set–oriented perspective (Lockwood, 2014); we do not elaborate each
instance of this for spatial reasons.

Positional reasoning supporting construction of groups 1 and 4


The students wrote “limited slots and limited objects, unique arrangement of unique objects” to
characterize Group 1 and wrote “x objects ! x!” for their general formula.
The students said the following when describing their understandings of Group 1:
COGNITION AND INSTRUCTION 31

Anne–Marie: So, with these ones [refers to Group 1], they’re all arrangement problems. So, how many ways
can you arrange a number of unique characters?
Aaron: One is used up and then it’s kind of a matter of slots just finding how many unique ones can
go inside?
Carson: Right, it’s dependent events. So, what you have for the second depends on what you get for
the first.
Josh: I see it as having a limited amount of slots and a limited amount of objects that you can put into
those slots.
Carson: Yeah, yeah. So, I mean these ones [refers to Group 1] are unique from the other arrangement ones
that we have [refers to Group 2] in that all of the slots are filled.

[On how to describe the problem type]
Anne–Marie: How many unique arrangements?
Aaron: Mm–hmm.
Carson: Yeah, how many arrangements of a given set of unique objects.

Josh: Well, given x number of objects, you can arrange them in x! number of ways.

The students here described a process of filling slots such that the slot choices depend on pre-
vious choices. Their utterances are consistent with an understanding that the process of filling in
slots yields an arrangement of objects, and that x! enumerates the result of the slot–fill-
ing process.
Similar to their characterization of Group 1, the students wrote “Independent events with
given number of outcomes” for their characterization of Group 4, and “ab , a ¼ possibilities of
each choice, b ¼ total # of choices you make”. Their discussion focused on the lack of depend-
ence between the steps of the outcome construction.
Carson: Independent events.
Aaron: So, it doesn’t matter that you’re repeating an element more than once.
Carson: For an outcome.
Aaron: Yeah.
Anne–Marie: Your options are not reduced every time?
Carson: Right.
Anne–Marie: So, you have a consistent number of multiplication.
Carson: Right.

[On how to describe the problem type]
Carson: Trials?
Josh: I would see it as you’re making a certain number of choices. You have a certain number of possibilities for
each choice.
Carson: Right.
Josh: So, it would be the number of possibilities that you have for each choice raised to the power of how
many choices you’re making.
Carson: Right.
32 Z. REED AND E. LOCKWOOD

Anne–Marie: So, possibilities of each choice.


Josh: Yeah, and the choices are independent from one another.

The students’ language while determining and describing their characterizations of Groups 1
and 4 display what we observed to be their consistent engagement in positional reasoning, specif-
ically their continued counting throughout the experiment via outcome–organization into slots.
After the first two problems in the initial activity phase, the students frequently phrased novel
counting problems in terms of a number of available “slots” (or “spots”), how many options were
available to fill those “slots”, and whether or not the sequence of choices affected subsequent
“slot” options. Moreover, the students consistently organized representative outcomes according
to a spot–based representation throughout the initial activity phase. Much like the PTE, we inter-
preted the SGTE students’ consistent use of a partial outcome construction and appeal to sequen-
tial slot–filling enumeration strategies as evidence of their engagement in mental actions
consistent with positional reasoning.
During categorization, Carson described the sequential filling of slots as contrasting
“dependent” versus “independent” events, particularly when sorting the problems and describing
the differences between categories. The other students then later adopted this language in the
characterization phase to convey the distinctions Carson drew out in categorization. We observed
that the students’ utterances during characterization were consistent with the mental activity we
were inferring from their work throughout the experiment. As such, we contend that the stu-
dents’ understandings of these two groups were the culmination of their reflection on the slot–-
filling activity that they had engaged in for the vast majority of the counting problems in the
initial activity, then made distinct via relating due to the “dependent” and “independent” con-
straints on the slot–filling process.

Equivalence supporting construction of groups 2 and 3


The students wrote “selection from arrangement” for their characterization of Group 2, and “ðab
a!
Þ! ,
a ¼ total amount of objects, b ¼ how many objects you choose”. Their discussion highlights the ways
they understood permutations and demonstrates the students’ language as consistent with an equiva-
lence way of thinking.
Carson: Same as above [Group 1], except only asking for unique number of places.
Josh: Well, we have more unique objects than unique places.
Carson: Right … So, we have ten horses in a race. How many ways can the horses finish, but then how
many of those have a unique podium, right? So, how many times are the first, second, and third
place different?
Aaron: So, you’re not really looking at four through ten in that case?
Carson: Right.

Aaron: So, just getting rid of all the arbitrary combinations that you’re not looking for.
Carson: Right.
Anne–Marie: So, you just have a bag of something and how many different ways can you grab a certain
amount from that?

Carson: It’s like selection from an arrangement. … So, I guess the way that I think about this is that A is
your total number of arrangements for the entire thing and then you want to divide by the number of ways
that the places you’re not selecting can be arranged, right?
COGNITION AND INSTRUCTION 33

We interpret that the students understood the combinatorial products of Group 2 as similar to
Group 1, but distinct in that the number of desired objects arranged could be chosen from a total
amount of objects. The students’ first comments highlight attention to a difference between the
number of slots filled and the presented number of objects. The students’ language of “selection
from arrangement” conveys this emphasis on varying the desired number of slots to be filled after
arranging the total number of objects. We infer from their language that the students’ under-
standings were supported by their consistent reasoning with equivalence, of which we observed
first during the latter half of the initial activity phase.
The students then described the differences between Group 3 and the previous groups by say-
ing “arrangement does not matter, divide out duplicates” and “similar to 2”, with a formula
a!
ðabÞ!b! , where a and b are assumed to have the same meaning as their counterparts in Group 2.
Describing this group, the students said the following:
Carson: So, the same as above [Group 2], with another caveat that some of the elements are–
Josh: Arrangement doesn’t matter.
Carson: Right, that arrangement doesn’t matter.
Josh: It’s the same thing [as Group 2] except the order in which you pick them doesn’t matter.
Anne–Marie: Arrangement, yeah, without order.

Josh: Yeah, I think that you’re also going to – it’s basically this [Group 2] except you’re also dividing by the
number of things that you’re picking from because order doesn’t matter.

Aaron: You’re trying to get rid of all the combinations that you’re not looking for that you can make out of
those three slots because they’re all the same. So, that just accounts for that.

As with the students in the PTE, we see these students making explicit connections between
permutations and combinations, not just in the formulas but in the combinatorial operations that
the formulas represent. The students’ repeated attention to outcome ordering conveys that they
found relevant whether or not the result of an enacted counting process was an ordered arrange-
ment or an unordered grouping. Further, the students’ language conveys attention to the division
of the b! as a means of accounting for this ordering by removing the “same” outcomes, from
which we infer influence of an equivalence way of thinking.
The students’ utterances describing Groups 2 and 3 were consistent with the terminology of
“duplicates” and “getting rid of” unwanted slots or outcomes that the students adopted at the end
of the first session, and they continued to employ such language through the rest of the initial
activity phase and throughout the categorization phase. In particular, while solving the Horse
Race Problem, the students first recognized that while all of the horses in the race could be
ranked, different arrangements of the non–medal winners would result in “duplicates” of the
same three winning horses. After a lengthy discussion of whether subtraction or division was
the appropriate way to account for the duplicates, the students agreed that division would create
the single representative outcome from the “duplicates” of what they wanted to count. Following
this first discussion, the students would identify when there was an abundance of “unwanted”
slots, and they would engage in division to “get rid of duplicate” arrangements that they “were
not looking for”.
Anne–Marie then described that they needed to “add another layer” when duplicate outcomes
could be generated when the ordering of outcomes did not matter. The students then intention-
ally described whether the problems that they were solving asked for “unique” outcomes, or
whether orderings would create “duplicate” outcomes, all while attending to how many “slots”
the problem asked for and making sure to remove the unwanted slots. The students’ habitual
34 Z. REED AND E. LOCKWOOD

attention to the duplication of desired outcomes, and subsequent division to eliminate the dupli-
cates is consistent with the mental actions we associate with an equivalence way of thinking.
Moreover, this language continued in their categorization of the problems, wherein multiple times
the students made distinct those problems involving “divid[ing]” to “cancel out the duplicates”
from others lacking this solution feature. Moreover, as seen in their language for the Group 3
characterization, the students also refined the boundary between duplication due to an abundance
of “wanted slots” and duplication because of outcome ordering. As such, engagement in an
equivalence way of thinking supported the students’ solving of novel counting problems, their
establishing of relationships and boundaries between problems, and their determining of the fea-
tures of the Group 2 and 3 problem types.
In sum, we contend that the same emergent theoretical constructs from the PTE can serve as
explanatory mechanisms for the SGTE students’ counting and generalizing activity.

Discussion, implications, and avenues for future research


We recall both our practical and theoretical goals in this paper. We aimed to discover ways that
students might productively understand and differentiate between four basic problem types in
combinatorics, and we also sought to highlight particular combinatorial ways of thinking that
emerged during the Categorization Activity. In addition, we aimed to contribute theoretically to
the field’s understanding of the relationship between generalization, ways of thinking, and reflec-
tion on initial activity. In this section we summarize ways in which we have addressed these
goals, and we offer several points of discussion. We conclude with avenues for future research.
In this study, we leveraged specific generalizing activities defined by Ellis et al. (2017) (relating
and extending) to provide students with opportunities to develop productive understandings of
general formulas for four basic combinatorial problem types – arrangement without repetition,
permutation, selection, and arrangement with unrestricted repetition. We have demonstrated that
both relating and extending each provided the students opportunities to reflect on initial activity
in unique ways, and we have documented ways that each generalizing action contributed to the
students’ work. An important aspect of these generalizing activities was the provision of opportu-
nities to engage repeatedly in three combinatorial ways of thinking: a set–oriented perspective,
positional reasoning, and thinking about equivalence. As we will discuss below, eliciting specific
reflection on activity and establishing connections to ways of thinking both offer novel contribu-
tions to our understanding of the phenomenon of generalization.
Practically, we have identified models of students’ combinatorial understandings of the basic
counting formulas, addressing a need identified in the literature of finding effective ways to intro-
duce counting formulas that facilitate reasoning that is rich and combinatorial in nature rather
than superficial (e.g., Lockwood, 2014; Lockwood et al., 2015b). In particular, our findings offer
evidence of students constructing productive understandings of basic counting formulas through
reflection on combinatorial mental activity rather than from strictly numerically–based regularity
(such as in Lockwood et al., 2015b).
We have thus addressed our research question by reporting on the ways that specific generaliz-
ing activities (relating via categorization and extending via characterization) and combinatorial
ways of thinking (a set–oriented perspective, positional reasoning, and thinking about equiva-
lence) contributed to students articulating and distinguishing between problem types and under-
standing and justifying key combinatorial formulas. We now offer points of discussion and
implications, first regarding generalization, and then regarding combinatorics. In doing so, we
highlight more specifically how both our practical and theoretical goals have been addressed.
COGNITION AND INSTRUCTION 35

Insights about generalization


In our study, we drew on Ellis et al.’s (2017) R–F–E generalization framework to support stu-
dents’ combinatorial understandings through particular means of facilitating reflection on activity.
In doing so, we have contributed to the field’s knowledge of generalization in three ways, each of
which we elaborate below: 1) We have engaged students in specific generalizing actions as explicit
mechanisms of reflection on initial activity, thus exploring potential ways that reflection might
occur when generalizing; 2) we have highlighted a relationship between students’ ways of thinking
and their engagement in generalizing activity, demonstrating the utility of generalizing actions
that enable students to make use of their ways of thinking; and 3) we have elaborated the con-
struct of relating to entail students’ constructing of relational differences as well as relational
similarities.

Specifying reflection on initial activity


In this study, we have operationalized Ellis et al.’s (2017) R–F–E framework to facilitate specific
and targeted reflection on initial activity. By leveraging the generalizing actions first of relating
and then of extending, we built models of students’ general understandings as constructions from
organizations of mental acts first isolated and made meaningful through a refinement and differ-
entiation process and then envisioned as applicable to a broader class of problems. Further, the
making of distinctions in relating situated the students’ reflections on the four groupings of men-
tal acts as each meaningful initially in relation to the other groups. We also saw the same rela-
tionships guide the students’ extending activity, and the students extended by attending to those
same relationships and distinctions that emerged in their relating activity. This process was
largely influenced by their engagement in the combinatorial ways of thinking, and we elaborate
on this aspect of their generalizing below.
We suggest that our findings provide novel insights into ways that generalization can occur
through specific kinds of reflection on initial activity. In our particular experimental design that
focused on relating and extending, we had students solve a number of different problems that
they ultimately categorized. There are other possible means for reflection on initial activity in
generalization. For instance, we hypothesize that the generalizing action of forming can facilitate
reflection by having students solve the same kind of problem repeatedly (in increasingly general
form), which contrasts our experimental design of engaging students in solving different problem
types from which boundaries can be made to make particular mental acts meaningful in isolation.
Making detailed distinctions between mechanisms of reflection offers novel theoretical insights
into the ways that generalization can be supported across mathematical contexts, rather than
attending to students’ reflections in specific contexts. Future research could investigate ways that
different generalizing actions elicit students’ specific reflections on initial activity and explore
novel interventions that could elicit such generalizing actions and ways of thinking.
While our students’ generalizing was in the context of combinatorics, our use of the R–F–E
framework provides insight into the impact of specific mechanisms for reflection on activity that
can be more broadly applied in future research on students’ generalizing. We contend that our
students’ engagement in the three phases taken together presents a comprehensive and detailed
account of their reflection on initial activity.
As discussed above, our results suggest that the students benefited in different ways from
engaging in each of the specific generalizing actions of relating and extending. Because of this, we
contend that studies examining students’ generalizations would benefit from attending to specific
generalizing actions as we have. We demonstrated this within the context of combinatorics, but
we conjecture that our design is more broadly applicable to other mathematical contexts
and domains.
36 Z. REED AND E. LOCKWOOD

Generalization and ways of thinking


Our results also suggest a possible synergistic relationship between students’ ways of thinking and
the kinds of reflections from which they organize more general understandings. In doing so, we
have further developed the existing theoretical contribution of Ellis et al.’s (2017) framework by
making explicit some deeper connections between students’ generalizing and their ways of think-
ing (Harel, 2008a).
We saw that, when relating, the students’ establishing of relationships between problems was
regulated by their repeated engagement in certain ways of thinking that emerged during their ini-
tial activity. Specifically, the students leveraged their ways of thinking as means of identifying and
articulating the relevant combinatorial activities that distinguished the four groups. For instance,
the students made combinations distinct from permutations through discussions of duplicate out-
comes, such as whether or not 1, 2, 3 and 3, 2, 1 were the same.
They then further engaged in these ways of thinking during characterization, which contrib-
uted to their final articulations of general combinatorial processes and formulas. In particular, the
students demonstrated engagement in the three ways of thinking to formalize and articulate the
general characteristics of each group. For instance, the students’ discussions of the formula for
combinations referenced the need to account for equivalence because of their continued attention
to identifying when outcomes should or should not be considered the same.
By demonstrating the ways in which generalizing actions facilitated reflection on initial activity
that allowed the students to simultaneously draw from and also reinforce desirable combinatorial
ways of thinking, we gain further insight into the potential for a synergistic relationship between
generalizing actions and students’ ways of thinking. Both sets of students engaged in such activity,
suggesting that it is a phenomenon that is not idiosyncratic to one particular set of students, but
that such a relationship between generalizing and ways of thinking can be reinforced for students
more broadly. We note that while our findings are in the domain of combinatorics, there is
opportunity to investigate similar phenomena in other mathematical areas.

Relating by establishing relationships from both similarities and differences


Finally, we observed both sets of students comparing and contrasting problems (and groups of
problems) as they engaged in categorization, and we want to emphasize the importance of this
contrasting activity. It seemed that articulating not only similarities but also differences between
problems and categories of problems was especially useful for the students. Focusing on what was
different between problems not only served to delineate categories but was also a tool for students
to refine processes, justify expressions, and specify the nature of what was being counted. In par-
ticular, the students’ attention to differences played a key role in their generalization as the refin-
ing of the boundaries between groups was a means by which the students isolated the particular
mental acts for extension that were salient for each group. Even more, the emphasis on differen-
ces came up even as they engaged in extending via characterization. Not only did the students
restate many of the distinctions that they established in categorizing when engaging in character-
izing, but articulating the differences regulated the students’ engagement in the ways of thinking
they drew from when extending.
Moreover, the students highlighted this through attention to their engagement in the activities
such as “filling in slots” through positional reasoning and dividing to account for “duplicates”
through equivalence. Engagement with the combinatorial ways of thinking influenced their final
understandings and helped address concerns raised in the literature pertaining to students’ diffi-
culties determining when to apply appropriate counting formulas (Batanero et al., 1997;
Lockwood, 2014). This further elaborates the notion of relating from its previous characterization
in the R–F–E framework. Specifically, Ellis et al. characterize relating as establishing “relations of
similarity across problems or contexts” (Ellis et al., 2017, p. 680). We have demonstrated that also
developing relationships between problems rooted in perceived differences enabled the students
COGNITION AND INSTRUCTION 37

to attend to the particular features of their activity that made the counting problems and formu-
las meaningful. Thus, we would contend that the establishing of relationships is a key component
of relating that we drew from heavily, but that in some instances it was productive for the stu-
dents to also establish the relationships by attending to the differences between the problems
and contexts.

Insights about combinatorics


We now briefly summarize some implications for combinatorics education in particular. From
this study, we primarily gain insight into the nature of three combinatorial ways of thinking and
what makes them particularly effective and important for students. It is noteworthy that these
ways of thinking we identified were regularly and consistently applied by students throughout
both sets of interviews.

Insights about combinatorial ways of thinking


First, we posit that engendering strong understandings of equivalence and a set–oriented perspec-
tive in students can help them make sense of the phenomenon of overcounting and, hopefully,
avoid it. This kind of reasoning provides a possible way to alleviate tendencies toward Batenero
et al.’s (1997) errors of order and errors of repetition. Specifically, the identification of
“duplicates” is a productive way to characterize the ordering of outcomes in counting problems
and leverages equivalence to reduce the kinds of errors Batenero et al. identified. We also saw
that the students constructed meaningful differentiations between problem types during the cat-
egorization phase, which were then extended during the characterization phase. These distinctions
complemented and informed their understandings of the formulas as described above, and
address Batenero et al.’s (1997) concerns as well as the concerns raised by Lockwood (2014), who
identified ways that students decided to use the counting formulas in non–rigorous ways based
on “gut” feelings of when order mattered and did not matter. Thus, we contend that the students’
generalizing activity and combinatorial activity both contributed to their productive understand-
ings of the counting formulas that address issues raised by Batenero et al. (1997) and Lockwood
(2014). Future work can further explore ways that identification of “duplicate” outcomes can
address issues of overcounting.
The math education literature has not explicitly examined positional reasoning as a target of
study, but we see some places in the literature in which it has come up, especially related to stud-
ies of the multiplication principle (e.g., Lockwood, Reed, & Caughman, 2017; Lockwood & Purdy,
2019a, 2019b). For example, English (1991, 1993) discusses the odometer strategy, which is a
common and entirely important strategy for enumerating outcomes. While she does not explicitly
use the term positional reasoning, she does discuss the strategy of holding elements of outcomes
constant and cycling through the other elements to produce a structured list of outcomes. As
another example, Maher, Powell & Uptegrove (2011) demonstrate students exhibiting a strategy
they call “controlling for variables,” which they describe as keeping an object in a particular pos-
ition constant (p. 35). Our study demonstrates the ways that positional reasoning can influence
students’ combinatorial activities and contribute to a consistent method of constructing represen-
tative outcomes that reinforce the uses of the multiplication principle in counting. We contend
that engaging in positional reasoning brought structure to the students’ understandings of the
sets of outcomes and the use of multiplication in solving a counting problem. The utility of
engaging in positional reasoning is found in the development of a coherent way of thinking about
structuring counting problems that leverages multiplication as a primary operation. We suggest
that more work needs to be done to understand the nature of positional reasoning in combina-
torics education.
38 Z. REED AND E. LOCKWOOD

We feel that our study also highlights the need for further examination of the combinatorial
ways of thinking we identified in this study. While the set–oriented perspective has been previ-
ously identified in the literature as a productive combinatorial way of thinking (Lockwood, 2014),
we see a need for additional investigation of the equivalence and positional ways of thinking.
Lockwood and Reed (2020) elaborated the subtleties of an equivalence way of thinking, but we
posit that there is more work to be done exploring ways that equivalence can be leveraged as a
productive way of thinking across mathematical domains. Additionally, further exploration into
the broader impacts of all of these ways of thinking in students’ combinatorial reasoning is war-
ranted, specifically outside of the context of the Categorization Task and these basic count-
ing formulas.

Future directions
We conclude with some general comments on directions for further research. We note that both
groups categorized and characterized problems similarly to each other, and thus we get little
insight into what might occur if students were to start with a different way of categorizing prob-
lems. We think that instances of student generalization that differ from the described mental acts
of the students in this study would be interesting and insightful, both from a combinatorial and a
generalization perspective. We frame this as an avenue for future research and suggest that it
would be helpful to investigate what other kinds of categorizations students would identify among
the problem set we gave our students.
Further, we primarily worked with relatively strong students – vector calculus assumes a certain
level of mathematical maturity experience. It also suggests (as noted previously) that we were con-
ceiving of working with students who, as vector calculus students, could eventually see combinatorics
in an introduction to proof class. There are also opportunities for students with less mathematical
experience (who may not be math majors) to engage with combinatorics – often there are introduc-
tory discrete mathematics classes for computer science students, for example. Working with students
who have less mathematical experience would be interesting, potentially as a way to observe alterna-
tive ways of categorizing. If other student groups recreate the results of this study, then that would
be informative, but it would also be useful to see other kinds of aspects of the problems they might
attend to. Thus, working with students such as college algebra, precalculus, or introductory computer
science students could be a worthwhile avenue to pursue.
A natural next step for future research will be to explore the effectiveness of the Categorization
Activity as an instructional intervention. Researchers could investigate the possibility of adapting the
Categorization Activity in an instructional setting, either a classroom or a small group setting. For
instance, it might be useful to understand potential issues that could arise due to the constraints of
implementing this activity with multiple groups. We would want to better understand how some-
thing like the Categorization Activity could scale and what instructors would need to do to imple-
ment it in whole–class environments. This highlights that an eventual extension of this research
would be to develop materials that extend the Categorization Activity for use in classrooms that
teach combinatorics – this could be in introductory discrete mathematics, discrete mathematics,
introduction to proof, combinatorics, or even probability courses.
Finally, as discussed above, other potential future projects could entail examining other ways
that the generalizing activities in the R–F–E framework can be leveraged in instruction. By speci-
fying the students’ mechanisms of reflection through relating and then extending, we were able to
document the factors influencing the students’ reflections when generalizing in greater detail than
has been done before. Future studies can examine similar ways to structure students’ learning by
employing constructs from the generalization literature.
COGNITION AND INSTRUCTION 39

Funding
This material is based in part upon work supported by National Science Foundation under Grant No. 1419973.

ORCID
Zackery Reed http://orcid.org/0000-0001-6956-8903

References
Amit, M., & Klass-Tsirulnikov, B. (2005). Paving a way to algebraic word problems using a nonalgebraic route.
Mathematics Teaching in the Middle School, 10(6), 271–276. doi:10.5951/MTMS.10.6.0271
Amit, M., & Neria, D. (2008). Rising to the challenge: Using generalization in pattern problems to unearth the
algebraic skills of talented pre–algebra students. ZDM, 40(1), 111–129. doi:10.1007/s11858-007-0069-5
Annin, S. A., & Lai, K. S. (2010). Common errors in counting problems. The Mathematics Teacher, 103(6),
402–409. doi:10.5951/MT.103.6.0402
Batanero, C., Navarro–Pelayo, V., & Godino, J. (1997). Effect of the implicit combinatorial model on combinatorial
reasoning in secondary school pupils. Educational Studies in Mathematics, 32(2), 181–199. doi:10.1023/
A:1002954428327
Bills, L., & Rowland, T. (1999). Examples, generalisation and proof. Advances in Mathematics Education, 1(1),
103–116. doi:10.1080/14794809909461549
Carpenter, T., Franke, M., & Levi, L. (2003). Thinking mathematically: Integrating arithmetic and algebra in elemen-
tary school. Portsmouth, NH: Heinemann.
Carraher, D. W., Martinez, M. V., & Schliemann, A. D. (2008). Early algebra and mathematical generalization.
ZDM, 40(1), 3–22. doi:10.1007/s11858-007-0067-7
Doerfler, W. (2008). En route from patterns to algebra: Comments and reflections. ZDM Mathematics Education,
40, 143–160.
Dubinsky, E. (1991). Reflective abstraction in advanced mathematical thinking. In D. Tall (Ed.), Advanced mathem-
atical thinking (pp. 95–123). Dordrecht, the Netherlands: Kluwer.
Eizenberg, M. M., & Zaslavsky, O. (2004). Students’ verification strategies for combinatorial problems.
Mathematical Thinking and Learning, 6(1), 15–36. doi:10.1207/s15327833mtl0601_2
El Mouhayar, R. (2018). Trends of progression of student level of reasoning and generalization in numerical and
figural reasoning approaches in pattern generalization. Educational Studies in Mathematics, 99(1), 89–107. doi:
10.1007/s10649-018-9821-8
Ellis, A. B. (2007a). A taxonomy for categorizing generalizations: Generalizing actions and reflection generaliza-
tions. Journal of the Learning Sciences, 16(2), 221–262. doi:10.1080/10508400701193705
Ellis, A. B. (2007b). Connections between generalizing and justifying: Students’ reasoning with linear relationships.
Journal for Research in Mathematics Education, 38(3), 194–229.
Ellis, A. B. (2007c). The influence of reasoning with emergent quantities on students’ generalizations. Cognition
and Instruction, 25(4), 439–478. doi:10.1080/07370000701632397
Ellis, A. B., Lockwood, E., Tillema, E., & Moore, K. (2017). Generalization across domains: The relating–formin-
g–extending generalization framework. In Galindo, E., & Newton, J., (Eds.). Proceedings of the 39th annual meet-
ing of the North American Chapter of the International Group for the Psychology of Mathematics Education.
Indianapolis, IN: Hoosier Association of Mathematics Teacher Educators.
English, L. D. (1991). Young children’s combinatorics strategies. Educational Studies in Mathematics, 22(5),
451–474. doi:10.1007/BF00367908
English, L. D. (1993). Children’s strategies for solving two– and three–dimensional combinatorial problems.
Journal of Mathematical Behavior, 24(3), 255–273.
Font, V., & Contreras, A. (2008). The problem of the particular and its relation to the general in mathematics edu-
cation. Educational Studies in Mathematics, 69(1), 33–52. doi:10.1007/s10649-008-9123-7
Hadar, N., & Hadass, R. (1981). The road to solve combinatorial problems is strewn with pitfalls. Educational
Studies in Mathematics, 12(4), 435–443. doi:10.1007/BF00308141
Harel, G. (2008a). A DNR perspective on mathematics curriculum and instruction, Part I: Focus on proving. ZDM,
40(3), 487–500. doi:10.1007/s11858-008-0104-1
Harel, G. (2008b). A DNR perspective on mathematics curriculum and instruction. Part II: With reference to
teacher’s knowledge base. ZDM, 40(5), 893–907. doi:10.1007/s11858-008-0146-4
Harel, G. (2001). The development of mathematical induction as a proof scheme: A Model for DNR–based instruc-
tion. In S. Campbell & R. Zaskis (Eds.), Learning and teaching number theory (pp. 185–212). Norwood, NJ: Ablex.
40 Z. REED AND E. LOCKWOOD

Harel, G., & Tall, D. (1991). The general, the abstract, and the generic. For the Learning of Mathematics, 11, 38–42.
Gravemeijer, K., Cobb, P., Bowers, J., & Whitenack, J. (2000). Symbolizing, modeling, and instructional design. In
P. Cobb, E. Yackel, & K. McClain (Eds.), Symbolizing and communicating in mathematics classrooms:
Perspectives on discourse, tools, and instructional design (pp. 225–273). Mahwah, NJ: Erlbaum and Associates.
Jones, S.R., & Dorko, A. (2015). Students” understanding of multivariate integrals and how they may be general-
ized from single integral conceptions. The Journal of Mathematical Behavior, 40, 154–170. doi:10.1016/j.jmathb.
2015.09.001
Kabael, T.U. (2011). Generalizing single variable functions to two variable functions, function machine and APOS.
Educational Sciences: Theory and Practice, 11(1), 484–499.
Kapur, J. N. (1970). Combinatorial analysis and school mathematics. Educational Studies in Mathematics, 3(1),
111–127.
Kaput, J. (1999). Teaching and learning a new algebra with understanding. In E. Fennema, & T. Romberg (Eds.),
Mathematics classrooms that promote understanding (pp. 133–155). Mahwah, NJ: Erlbaum.
Kavousian, S. (2008). Enquiries into undergraduate students’ understanding of combinatorial structures
(Unpublished doctoral dissertation). Simon Fraser University, Vancouver, BC
Lobato, J. (2003). How design experiments can inform a rethinking of transfer and vice versa. Educational
Researcher, 32(1), 17–20. doi:10.3102/0013189X032001017
Lockwood, E. (2014). A set–oriented perspective on solving counting problems. For the Learning of Mathematics,
34(2), 31–37.
Lockwood, E. (2013). A model of students’ combinatorial thinking. The Journal of Mathematical Behavior, 32(2),
251–265. doi:10.1016/j.jmathb.2013.02.008
Lockwood, E. (2011). Student connections among counting problems: An exploration using actor–oriented transfer.
Educational Studies in Mathematics, 78(3), 307–322. doi:10.1007/s10649-011-9320-7
Lockwood, E., & Caughman, J. (2016). Set partitions and the multiplication principle. PRIMUS, 26(2), 143–157.
doi:10.1080/10511970.2015.1072118
Lockwood, E., & Erickson, S. (2017). Undergraduate students’ initial conceptions of factorials. International Journal
of Mathematical Education in Science and Technology, 48(4), 499–519. doi:10.1080/0020739X.2016.1259517
Lockwood, E., & Gibson, B. (2016). Combinatorial tasks and outcome listing: Examining productive listing among
undergraduate students. Educational Studies in Mathematics, 91(2), 247–270. doi:10.1007/s10649-015-9664-5
Lockwood, E., & Purdy, B. (2019a). Two undergraduate students’ reinvention of the multiplication principle.
Journal for Research in Mathematics Education, (50)(3), 225–267. doi:10.5951/jresematheduc.50.3.0225
Lockwood, E., & Purdy, B. (2019b). An unexpected outcome: Students’ focus on order in the multiplication prin-
ciple. International Journal of Research in Undergraduate Mathematics Education, 6(2), 213–244. doi:10.1007/
s40753-019-00107-3
Lockwood, E., & Reed, Z. (2018). Reinforcing mathematical concepts and developing mathematical practices
through combinatorial activity. In E. W. Hart, & E J. Sandefur (Eds.), Teaching and Learning of Discrete
Mathematics Worldwide: Curriculum and Research (pp. 93–110). Cham, Switzerland: Springer.
Lockwood, E., & Reed, Z. (2016). Students’ meanings of a (potentially) powerful tool for generalizing in combina-
torics. In T. Fukawa–Connelly, N. Infante, M. Wawro, & S. Brown (Eds.), Proceedings of the 19th Annual
Conference on Research on Undergraduate Mathematics Education (pp. 1–16). Pittsburgh, PA: West Virginia
University.
Lockwood, E., & Reed, Z. (2020). Defining and demonstrating an equivalence way of thinking in enumerative com-
binatorics. The Journal of Mathematical Behavior, 58, 100780. doi:10.1016/j.jmathb.2020.100780
Lockwood, E., Reed, Z., & Caughman, J.S. (2017). An analysis of statements of the multiplication principle in com-
binatorics, discrete, and finite mathematics textbooks. International Journal of Research in Undergraduate
Mathematics Education, 3(3), 381–416. doi:10.1007/s40753-016-0045-y
Lockwood, E., Swinyard, C. A., & Caughman, J. S. (2015a). Patterns, sets of outcomes, and combinatorial justifica-
tion: Two students’ reinvention of counting formulas. International Journal of Research in Undergraduate
Mathematics Education, 1(1), 27–62. doi:10.1007/s40753-015-0001-2
Lockwood, E., Swinyard, C. A., & Caughman, J. S. (2015b). Modeling outcomes in combinatorial problem solving:
The case of combinations. In T. Fukawa–Connelly, N. Infante, K. Keene, & M. Zandieh (Eds.), Proceedings of
the 18th Annual Conference on Research on Undergraduate Mathematics Education (pp. 601–696). Pittsburgh,
PA: West Virginia University.
Martin, G. E. (2001). The art of enumerative combinatorics. New York, NY: Springer.
Maher, C. A., Powell, A. B., & Uptegrove, E. B. (Eds.). (2011). Combinatorics and reasoning: Representing, justify-
ing, and building isomorphisms. New York, NY: Springer.
National Governors Association Center/Council of Chief State School Officers. (2010). Common core state stand-
ards for mathematics. Washington, DC: Council of Chief State School Officers.
Oflaz, G., & Demircioglu, H. (2018). Determining ways of thinking and ways of understanding related to general-
ization of eighth graders. International Electronic Journal of Elementary Education, 11(2), 99–112.
COGNITION AND INSTRUCTION 41

Piaget, J. (2001). Studies in reflecting abstraction. Hove, UK: Psychology Press Ltd.
Radford, L. (2008). Iconicity and contraction: A semiotic investigation of forms of algebraic generalization of pat-
terns in different contexts. ZDM, 40(1), 83–96. doi:10.1007/s11858-007-0061-0
Radford, L. (2006). Algebraic thinking and the generalizations of patterns: A semiotic perspective. In S. Alatorre,
J. L. Cortina, M. Saiz, & A. Mendez (Eds.), Proceedings of the 28th annual meeting of International Group for
the Psychology of Mathematics Education, North American Chapter (Vol. 1, pp. 2–21). Merida: Universidad
Pedag ogica Nacional.
Reed, Z. (2018a). Generalizations of convergence from R to R2 : In A. Weinberg, C. Rasmussen, J. Rabin, M.
Wawro, and S. Brown (Eds.), Proceedings of the 21st annual conference on research in undergraduate mathemat-
ics education, San Diego, California.
Reed, Z. (2018b). Undergraduate students’ generalizing activity in real analysis: Constructing a general metric
(Unpublished doctoral dissertation). Oregon State University.
Reed, Z. (2017). The effects of the   N relationship on convergence of functions. In A. Weinberg, C. Rasmussen,
J. Rabin, M. Wawro, & S. Brown (Eds.), Proceedings of the 20th annual conference on research in undergraduate
mathematics education, San Diego, California.
Rivera, F. D. (2010). Visual templates in pattern generalization activity. Educational Studies in Mathematics, 73(3),
297–328. doi:10.1007/s10649-009-9222-0
Rivera, F. D., & Becker, J. R. (2007). Abduction–induction (generalization) processes of elementary majors on fig-
ural patterns in algebra. The Journal of Mathematical Behavior, 26(2), 140–155. doi:10.1016/j.jmathb.2007.05.001
Rivera, F. D., & Becker, J. R. (2008). Middle school children’s cognitive perceptions of constructive and decon-
structive generalizations involving linear figural patterns. ZDM Mathematics Education, 40(1), 65–82. doi:10.
1007/s11858-007-0062-z
Rosen, K. H. (2007). Discrete mathematics and its applications (6th ed.). New York, NY: McGraw Hill.
Sfard, A. (1991). On the dual nature of mathematical conceptions: Reflections on processes and objects as different
sides of the same coin. Education Studies in Mathematics, 22(1), 1–36. doi:10.1007/BF00302715
Steffe, L. P., Thompson, P. W., & Glasersfeld, E. V. (2000). Teaching experiment methodology: Underlying princi-
ples and essential elements. In R. Lesh & A. E. Kelly (Eds.), Research design in mathematics and science educa-
tion. Mahwah, NJ: Lawrence Erlbaum Associates.
Tucker, A. (2002). Applied combinatorics (4th ed.). New York, NY: John Wiley & Sons.
Thompson, P. W. (2008). Conceptual analysis of mathematical ideas: Some spadework at the foundations of math-
ematics education. In O. Figueras, J. L. Cortina, S. Alatorre, T. Rojano & A. Sepulveda (Eds.), Plenary paper pre-
sented at the annual meeting of the international group for the psychology of mathematics education (Vol. 1, pp.
45–64). Morelia, Mexico: PME.
von Glasersfeld, E. (1995). Radical constructivism: A way of knowing and learning. Bristol, PA: The Falmer Press.

Appendix
Selection Interview Tasks

1. Demographic questions

a. Which math classes did you take in high school? Which math classes have you taken at the college
level? Did you take any statistics, probability, or computer science classes?
b. What is your major?
c. What year are you in school?

2. Have you seen any of the following symbols? If so, can you describe what you think it means, and where
you have seen it before?

a. nCr or Cðn, rÞ
b.  A B
n
3. (c)
r
a. P nPr or Pðn, rÞ
b. 1k10 k
4. A Domino is a small, thin rectangular tile that has dots on one of its broad faces. That face is split into two
halves, and there can be 0 through 6 dots on each of those halves. Suppose you want to make a set of domi-
nos (i.e., include every possible domino). How many distinguishable dominos would you make for a com-
plete set?
5. How many ways are there to flip a coin, roll a dice, and pick a card from a standard deck?
42 Z. REED AND E. LOCKWOOD

6. There are 4 different Russian books, 3 different Spanish books, and 5 different French books. You want to
pick a pair of books that are not of the same language. How many ways are there to do this?
7. How many ways are there to arrange 5 different pictures in a row on a wall?
8. Suppose there are 6 athletes competing in an event at the Olympics. How many different possible results are
there for who wins the gold, silver, and bronze medals?
9. You have 8 different books on your shelf, and you want to take 3 with you on vacation. In how many ways
can you choose 3 books to take with you?
10. There is a 5–question multiple choice quiz with possible answers A, B, and C. How many possibilities are
there for how to complete this quiz?
11. How many ways are there to arrange the letters in the word CATTLE so that the two Ts are together either
at the beginning or the end of the word?
12. A password consists of 8 upper–case letters. How many passwords contain at least 3 Es?

You might also like