By Maria de Hoyos Guajardo, Ph.D. Candidate, M.Sc., B.Eng.


The theory that is presented below aims to conceptualise how a group of
undergraduate students tackle non-routine mathematical problems during a
problem-solving course. The aim of the course is to allow students to experience
mathematics as a creative process and to reflect on their own experience.
During the course, students are required to produce a written ‘rubric’ of their
work, i.e., to document their thoughts as they occur as well as their emotions
during the process. These ‘rubrics’ were used as the main source of data.

Students’ problem-solving processes can be explained as a three-stage process
that has been called ‘solutioning’. This process is presented in the six sections
below. The first three refer to a common area of concern that can be called
‘generating knowledge’. In this way, generating knowledge also includes issues
related to ‘key ideas’ and ‘gaining understanding’. The third and the fourth
sections refer to ‘generating’ and ‘validating a solution’, respectively. Finally, once
solutions are generated and validated, students usually try to improve them
further before presenting them as final results. Thus, the last section deals with
‘improving a solution’. Although not all students go through all of the stages, it
may be said that ‘solutioning’ considers students’ main concerns as they tackle
non-routine mathematical problems.


An important activity in students’ problem-solving process is to generate
knowledge about the situation; i.e., to generate relevant data and information
and to gain understanding. This is usually conducted at the start of the process,
particularly if students know little or nothing about the situation. For this reason,
generating knowledge and understanding seems a good place to start the
discussion on students’ problem solving processes. However, it must be made
clear that the need to generate knowledge will continue to emerge throughout
the process and that students respond to this need in ways that will be
discussed in this section.

A common strategy that students use as they try to generate information and
understanding is to reduce the complexity of the situation that they are dealing
with. By reducing complexity, students “start at the beginning” and focus on
intentionally simplified or even trivial versions of the situation. Students’ aim
behind reducing complexity is to start gathering the information and
understanding that will allow them to eventually move on to more sophisticated
cases. Reducing complexity may help students gain access to complex
situations by reducing them to simpler, more manageable ones.

Numbers which can be expressed as a single prime to a power may be
a good place to start…(Oscar, Liouville, p. 2)

Right, let’s think about this. Start simple and work my way up…(Hillary,
Steps, p. 1)

Students generate information and gain understanding about the situation in
many ways. Thus, it is hypothesised that the only limit for students as they try to
generate useful information and understanding might be the one imposed by
their own creativity and mathematical abilities. The following is a brief list of the
types of activities that students conduct for this purpose. The list is not extensive
and other activities may be included from further research:

• A common way in which students generate information and understanding is
by ‘specialising’, i.e., by looking at particular aspects of the situation. When
students specialise, they focus on isolated aspects of the situation and thus on
simplified versions of the problem. For this reason, it may be said that
specialising is intrinsically about reducing complexity. Most students specialise
at one point or another in their processes and the choice seems to be made in
a ‘natural’ way (“My instinct to this problem is to start from the easiest case.”)
However, during the course, students were specifically introduced to Mason’s
(1982) idea of specialising. This fact may account for the students’ tendency to
specialise and to label their activity in that way.

I will start by specialising and using squares, since they seem more
straightforward, and then progress to rectangles. (Hannah, Cartesian
Chase, p. 2)

• In order to start making sense of the situation, students sometimes ‘import’
ideas or information from sources other than the problem and the situation that
it presents. These ideas may be relevant to the problem and in the sense that
they may help students to better understand the situation and deal with it.
Recalling past knowledge or experience are common ways of importing

I know a similar problem. Diagonals of a Rectangle, which seems to be
related and I think I can use my solution. (Emilio, Visible Points, p. 1)

Fault line – brings to mind brick walls. In a brick wall you couldn’t have
such a line because the wall would be too weak. Conjecture that brick
laying pattern may prove the answer. I will carry on specialising and will
come back to this conjecture later. (Kirk, Faulty Rectangles, pp. 1–2)

Students may also import information from other sources such as their notes (or
any bibliographical reference, from that matter). Sharing ideas with fellow
classmates may also be a way of gaining information and/or understanding.
Importing requires borrowed ideas to be evaluated in terms of their relevance
and applicability to the present situation. Importing can provide useful
information but also presents the risk of considering irrelevant ideas that may
have to be abandoned at a later time.

• Another way of generating knowledge is by taking a ‘hands-on’ approach and
carrying out the basic operations that are relevant to the situation. For
instance, in ‘Faulty Rectangles’ students physically constructed rectangles with
pieces of domino and observed what combinations could lead to fault-free
rectangles. Another example can be given in relation to ‘Ins and Outs’, where
students conducted hands-on investigations by folding pieces of paper and
observed the sequences of folds that were generated. Hands-on investigations
provide students with first hand experience of the situation and may lead to
gaining important knowledge and understanding.

Shall I try playing it? Use a chessboard and a pawn. (Jules, Cartesian
Chase, p. 1)

• A way of generating useful information and possibly understanding is by
organising the data that is available. This may involve arranging available
information in a convenient way so that further information becomes more
evident and easier to spot. Tables of values are a common example of
organising the data, but any other method for visualising the situation can also
be of help.

I will use a table to search for some patterns:(Keith, Sums of Diagonals, p. 2)

[please see PDF version for all diagrams]

If I try to draw a diagram of the possible outcomes this may help give me a
better idea of what is happening and may lead to further development. (Lila,
Steps, p. 1)

• An important way of finding more about the situation is by carefully analysing
the information that is available or that has been made available. In some
cases, information and understanding may emerge easily by looking at the
data. In other cases, however, students have to make conscientious efforts in
order to generate knowledge. By insistently considering (or reconsidering)
available information and trying to understand it, it may be possible to derive
further information and understanding from it. This may involve reviewing the
data and making deliberate efforts at drawing out observations and ideas.

Ok, let’s look at our previous example.


Stage 1: 1, 2, 4 [Divisors of N]

Stage 2: (1), (1, 2), (1,2,4) [Divisors of divisors of N]

Is there any significance about the numbers at stage 1? (Jared,
Liouville, p. 6)

Can’t see anything from 3 folds. Only – I guess that the sequence that
happened in the previous fold would happen in the current fold again,
so 4 folds should start with in in out in in out out, and something else. I
want to guess more detail about the 4 folds because I want to prove my
prediction is correct. But this is what I can see now. (Patrick, Ins and
Outs, p. 1)

It is not uncommon for students to combine these activities by either conducting
them at the same time or by sharing information from one activity to another. For
instance, students may take a hands-on approach as they gather information for
a table of values. Another example is when students conduct a close analysis of
information that has been generated after a period of specialising. As said, there
is no imposed limit to what students can do in order to generate information and

The need to generate knowledge will continue to emerge throughout the
process. New information and understanding may be required at any stage, from
situations in which students are looking for new ideas to situations where they
are trying to take an idea further. In other words, students may incur in the
activities discussed above at any time during their process.

Finally, students make reference to the information they observe in the form of
written or verbal observations. Trying to gain knowledge about the situation
leads students not only to noticing but also to ‘making a note’ on those new
pieces of information that may be relevant in terms of generating a solution. The
next subsection looks at the observations that students make as a result of
dealing with the data.

Making Observations

The information and understanding that students generate may become
manifest in the form of observations. Observations are facts or ideas about the
situation that students may find interesting or relevant, and that they choose to
point out in a written or verbal way. In some cases, these observations may lead
directly to an initial solution.

AHA! The pattern behind the centre is just a pattern of the previous one,
while those behind is just the opposite way around […]

Therefore, if we repeat this, we would be able to generate a sequence
after 10 folds. (Karina, Ins and Outs, pp. 1–2)

In other cases, however, observations may involve information that may or may
not be used at a later time.

This is to say that not all observations will be useful in the same way. Some may
inform students about ways to generate a solution (like in the example above)
and some may provide less central (though not necessarily unimportant)
information. In some cases, important observations are easily identified as such.
In other cases, it may take the student time and effort to be able to tell whether
a certain piece of information is relevant or not.


AHA! The gradient of slope 1 is 1. I can use the same method and
apply it to slope 2.


Aha! I got it! (Patrick, Sums of Diagonals, p. 2)

Obviously, I can only pull out the numbers 1 and 2 and the difference
between these is 1.

Hmm… could this always be the case (wild guess)? Or is it too early to
tell. (Aminta, Hat Numbers, p. 1)

When students come across an observation, sometimes they adopt what can be
called a ‘pragmatic’ approach. Adopting a pragmatic approach involves focusing
not only on the observation itself but also on how it can be used for generating a
solution. When students adopt a pragmatic approach towards making
observations they ask themselves questions like “How can this [idea, fact, etc.]
be used?” A pragmatic approach can help students decide more efficiently
whether an idea is useful and how.

The examples below (as well as Patrick’s example above) illustrate cases where
students considered observations in a pragmatic way. As the second example
below suggests, a pragmatic approach may help students discriminate
unimportant ideas and thus may help in making their process more efficient.
Thinking in terms of how ideas can be used seems to lead to starting to
generate a solution sooner than if observations are made without considering
their usefulness or applicability.

The answers for 2 and 5 give the answers for 10. Does this work for
other numbers? (Julia, Liouville, p. 5)

Points (i, j), where i, j are positive.

Defined to be BELOW (m, n) where m, n are positive when ≤m and ≤n.

∴ (i, j) is below itself – not particularly important. (Dylan, Visible
Points, p. 1)

Key Ideas

Having discussed how students generate knowledge about the situation and
how this knowledge becomes manifest, this section will look at ‘key ideas’ as
knowledge that is crucial to solving the problem and that students employ
directly to generate a solution. The first subsection discusses ‘looking for
patterns’ as ways of looking for key ideas by investigating the situation in a
particular way. The second sub-section discusses ‘key searching’ as a way of
looking for key ideas in a more direct way.

As said in the previous section, some of the observations that students make
during problem solving lead directly to generating a solution. Since these
observations usually refer to crucial aspects of the situation they can be called
key ideas. Students usually base their solutions on a key plan or idea that
provide hints as to how a solution can be obtained. In order to deal with
‘Diagonals of a Rectangle’, for instance, students used the fact that there is a
relationship between the highest common factor of the rectangle’s dimensions
and the number of rectangles crossed. This fact was the key idea on which most
(if not all) students who provided a solution for this problem based their

Key ideas sometimes emerge as sudden realisations of important aspects of the
situation. These ideas may appear as important breakthroughs (as the student
below suggests) and give students the feeling of having discovered how to
generate a solution.

AHA! This is a huge breakthrough! Anything that happens before the
row marked (*) is not important. As long as we can guarantee that our
opponent moves to (*), we have won, since we can then move to a
definite win position. (Leonard, Cartesian Chase, p. 5)

In other cases, key ideas emerge as less of a surprise. In these cases, key
ideas may come gradually as knowledge and understanding increase.

In either case, it seems that being able to arrive at a key idea requires a good
deal of understanding of the situation. When students are able to see a key
idea, they are also able to see its significance, its importance in relation to the
situation and how it can be of use. In relation to this, Raman (2003) observed
that the key ideas that more experienced solvers use to provide a mathematical
proof “give a sense of understanding and conviction” and show “why a particular
claim is true” (p. 5). In more general terms, Barnes (2000) suggested that when
students and more experienced mathematicians are able to see a key idea the
following takes place:

…there is a claim to a sudden realisation of new knowledge or
understanding. Usually this knowledge is ‘seen’ with great clarity, or
experienced with a high degree of confidence or certainty. (Barnes,
2000, p. 34)

Key ideas can be seen as the product of gathering sufficient relevant knowledge
and understanding to be able to start generating a solution. The following subsections
look at ways in which students generate and search for key ideas.

Looking for patterns

Looking for patterns can be considered as a way of learning about the situation
that can lead to finding key ideas. When students look for patterns, they are
usually looking for particular features of the situation can lead them to start
generating a solution. Students look for patterns hoping that, when they find
one, they will be able to transform it into a formula or to make a general
statement about the situation.

I shall look for patterns which might lead me to a formula of some kind.
(Lila, Sums of Diagonals, p. 1)

Looking for patterns can be a useful activity that generates relevant information.
For instance, noticing a pattern in the way the creases were formed in the ‘Ins
and Outs’ problem allowed students to tell how the creases for the 10th fold
would look like. Furthermore, as students look for patterns, they may also gain
understanding and learn about the situation. Thus, in many cases, looking for
patterns can be a fruitful activity.

However, looking for patterns can also become a ‘blinding’ activity that prevents
students from gaining the necessary information and understanding. When
students focus mainly on looking for patterns and neglect trying to see other
aspects of the situation, the possibility of gaining useful information seems to
decrease. In the example mentioned above, most students were able to see
how creases were formed and thus were able to tell how the 10th fold would
look like. However, very few students were able to provide a general (nonrecursive)
formula for this sequence. Students that were able to provide a
general formula did so not by looking for patterns but by gaining a deeper
understanding of how the sequence of ‘ins’ and ‘outs’ was generated. In
contrast, students that focused mainly on looking for patterns (as illustrated
below) were able to provide a recursive formula but failed to provide a general

I can’t see a pattern or anything jumping at me.

But by counting the number of ‘ins’ and ‘outs’ in any number of folds I can see
that each one seems to be an odd number.


Just comparing the difference between the number of ‘ins’ and ‘outs’
seems to show that they are powers of 2. (Rita, Ins and Outs, p. 2)

Thus, it may be said that looking for patterns can provide some very useful
information. In order to provide a more satisfactory solution, however, further
information and understanding need to be generated as well. Focusing on trying
to find particular information about the situation can lead to a dead end as it
prevents students from genuinely learning about the situation. ‘Key searching’,
as will be discussed in the next sub-section, is a way of looking for key ideas
that is related to this aspect of looking for patterns.

Key Searching

As mentioned above, key ideas allow students to start generating a solution.
Finding a key idea is certainly related to successful problem solving, and
students seem to be aware of this. For this reason, students may look for key
ideas by looking for patterns. Another way of looking for key ideas is by ‘key
searching’. Key searching means looking for key ideas in a direct way by trying
discover special features about the problem or by trying to find “what is so
special” about the situation.

I’m looking to see if the number left in the hat has some special

Still stuck! Maybe I should go back and try the odd numbers. After all,
as this may be the missing clue to the solution…(Aminta, Hat Numbers,
pp. 2–4)

As students try to gain knowledge and understanding of the situation, it is very
likely that they will eventually come across key ideas. Paradoxically, however,
key ideas are less likely to emerge if students focus on actively seeking them.
The reason for this may be that searching for key ideas may divert students’
attention from trying to learn about the situation. During key searching, students
seem to be so concerned about trying to find some “special” clue or quality that
they may neglect other important information. In the case of the Liouville
problem, for instance, some students spent most of their process trying to figure
out what was so special about sequences of numbers that if added and then
squared give the same value as when they are cubed and then added. In these
extreme cases, students were unable to make any significant progress and were
not able to identify any of the key ideas that allowed other students to generate
a satisfactory solution.

When students search for key ideas, they may ignore important information that,
if not a solution in itself, can be used towards that end. Furthermore, in some
cases, students that search for key ideas seem to ponder on the problem rather
than on trying to gain a broader understanding of the situation.

In general, not all students incur in key searching and those who do may
eventually abandon this activity and try to generate information and
understanding. However, the implications of key searching make this activity an
important one to consider. There is no evidence to suggest that key-searching is
related to mathematical background. What can be suggested is that keysearching
may be related to the features of the problems involved. This
hypothesis is supported by the fact that more students key-searched in the
‘Liouville’ problem than in any other. There is not sufficient evidence to state take
this hypothesis further. This issue can only be suggested for further research.


The above sections deal with the way students generate knowledge during their
problem-solving processes. This knowledge constitutes the information and
understanding that will allow them to deal with the problem and eventually to
achieve a solution. This section deals more closely with the issue of gaining
understanding. This issue plays an important role in being able to generate a
solution and most students will seek to gain understanding about the situation.
However, as it is discussed below, students may also ignore or avoid trying to
gain understanding and concentrate on manipulating data.

A good place to start a discussion on the characteristics of gaining
understanding during problem solving is by considering the following quote from

On a more everyday level, it is common for people first starting to
grapple with computers to make large-scale computations of things they
might have done on a smaller scale by hand. They might print out a
table of the first 10,000 primes, only to find that their printout isn’t
something they really wanted after all. They discover by this kind of
experience that what they really want is usually not some collection of
answers – what they want is understanding. (Thurston, 1995, p. 29;
emphasis in the original)

Although Thurston’s assertion was made in reference to professional
mathematicians, it may be said that it applies to many students as well.

Gaining understanding is an important aspect of the problem solving process.
Most students try to gain understanding of the situation to be able to start
generating a solution. As a student put it, it is easier to generate a solution by
“understanding the underlying principles” of the situation. In general, it seems
that having a better understanding of the situation empowers students and
allows them to generate a solution and take it further.

I can’t believe how I missed how every entry in the grid is the product of
its coordinates…

This means that given any coordinates we can work out what the entry
is. (Nadia, Sums of Diagonals, p. 4b)

An important way of gaining understanding is by reasoning in terms of how the
data is created, or how it stems from the situation. Although not all students try
to gain understanding in this way, and those who do may not do so all the time,
it may be said that thinking in terms of how information is created is a common
practice. Thinking in terms of how the sequences of ‘ins’ and ‘outs’ were created,
for instance, provided students with useful understanding of the situation. In
most cases, this allowed them to generate an initial solution for the ‘Ins and
Outs’ problem. The following quotes illustrate the type of reasoning that was
conducted in an attempt to gain understanding in relation to this problem.

What I’m going to do is take the five folds sequence and identify which
creases come from which fold. (Lydia, Ins and Outs, p. 7)

Maybe I should start to think about things on a more subtle level. What
actually happens every time I add a crease of paper? I’ll try to get this
into a diagram. (Leonard, Ins and Outs, p. 4)

When students try to think in terms of how the data is created, they usually gain
a kind of understanding that allows them to make informed decisions on what to
do next. In other words, they achieve what Skemp (1976) called ‘relational
understanding’. This type of understanding allows students to know “both what
to do and why” (p. 20) and for this reason it is usually an important asset during
problem solving. The understanding achieved by the students in the following
examples is relational in the sense that it provides information that can be useful
for understanding the situation and deciding what to do next. Furthermore, their
understanding seems to have been generated by reasoning in terms of how
what they observe stems from the observed situation:

Let’s try to think logically about specifically when a diagonal would pass
through a corner.

AHA! I think the diagonal will pass through a corner when n and m have
a common factor greater than 1. This makes a lot of sense because it
implies that the rectangle can be split up into smaller rectangles with
the same diagonal, and therefore the diagonal would pass through the
corners. (Hannah, Diagonals of a Rectangle, pp. 3–4)

Finally, considering the benefits of trying to think in terms of how the data is
created may look as if all students worked naturally in this way. However, this is
not the case. Students with stronger mathematical backgrounds are usually
keen on reasoning in terms of how the data stems from the situation. On the
other hand, students for whom mathematics is not a main subject seem more
prone to look for patterns without considering the situation that gives rise to the
data. The reasons for this behaviour are difficult to trace. It can be speculated
that thinking in terms of how data relates to the situation requires students to
combine thinking about the situation while, at the same time, trying to identify
useful patterns. Thus, some students may unconsciously avoid such an
increased complexity and choose to focus on only one task at the time. In such
situation, they may prefer to work on the simpler one which will be, presumably,
trying to spot patterns. This, however, is a tentative explanation; a more
grounded explanation certainly requires further research.


The previous sections looked at how students generate knowledge about the
situation. It was discussed how students make key ideas available and what
courses of action may hinder their emergence. Some ways in which students
gain understanding about the situation were also considered. In spite of its
importance, it may be said that generating knowledge is not the final aim of
problem solving but a means of making necessary resources available. The aim
of problem solving is to generate a solution and students will start attempting to
do this as soon as sufficient knowledge has been gathered. Two ways in which
students may try to generate a solution is by reasoning deductively and
inductively. Reasoning in terms of how data is generated from the situation can
also play an important role in generating a solution.

In order to generate a solution, students may rely on deductive reasoning. In
other words, they may follow logical implications from one idea to another until a
conclusion is reached. Reasoning deductively seems to be held in high regard
by most students since, whenever possible, they will try to arrive at a solution in
this way. In the Liouville problem, for instance, most students’ first attempt at
generating a solution involved providing some version of the following deductive

A prime number n has divisors 1 and n only, by definition.

1 has one divisor (1)

n has two divisors (1, n)

The sum of the number of divisors or divisors is therefore 1+2=3 and
squared this is 9.

The sum of cubes of the number of divisors or divisors is 13+23=9.

So the two numbers are equal for prime numbers. (Julia, Liouville, p. 2)

Also, as one student put it:

I generally try to use deduction. Deduction is ‘more valid’ in mathematics
although I often use inductive arguments. (Leonard, informal interview)

When students reason deductively, they sometimes base their arguments on a
relevant piece of mathematical knowledge. This piece of knowledge may consist
of mathematical concept or a procedure. In other words, students may build a
deductive argument by applying a concept or a definition in an ingenious way or
by making use of a familiar mathematical procedure. In the example above, the
student based her deduction on the mathematical definition of ‘prime number’.
The way she made use of this definition allowed her to generate a logical chain
of reasoning and to achieve an initial solution. As for applying a mathematical
procedure, the Arithmagons problem provides a good example. In most
solutions to the ‘Arithmagons’ problem it was common for students to base their
arguments on procedures for solving systems of linear equations. Although
making use of procedures may be more straightforward than deciding how to
apply a concept, in the sense of constructing logical chains of reasoning, the
former can also be considered a deductive argument.

Whenever there is the possibility of generating a deductive argument from the
knowledge and information available, students will usually follow this route.
When this is not the case, one option is to continue trying to generate
information and understanding until it is possible to generate a deductive
argument. Another option is to start trying to generate a solution by induction.

Reasoning inductively involves making tentative conjectures or generalisations
out of the information that is available. Making deductions involves deriving
ideas that are a logical consequence of the information available. In contrast,
when students reason inductively, they not only consider the information that is
available (and the logical implications of this information) but also draw upon
other less factual sources such as previous (possibly informal) knowledge and
experience. This knowledge and experience may arrive in the form of insight or
intuition, or in the form of ‘intuitive guesses’, as Fischbein and Grossman (1997)
put it. It is the combination of empirical data with other sources of knowledge
what usually makes inductive reasoning a fascinating process.

All the results are in a range 48–63…

Notice that the last two results are equal.

Conjecture 1: the percentage of visible points converges to a number.

Conjecture 2: the convergent number x=48.7%. (Aminta, Visible Points,
p. 4)

Generating ideas inductively may lead to inaccuracies or even to incorrect
solutions. This is not to say that deductive reasoning is foolproof. What this
suggests is that, due to the nature of inductive reasoning, students sometimes
have to accept, and deal with, the fact that they are working with imperfect
results. However, this is usually not a serious problem since ideas can be reexamined
and modifications can be made. Moreover, checking whether a
tentative solution is correct and makes sense allows students to improve their
solution and increases their knowledge and understanding of the situation. This,
together with the fact that an initial solution – i.e., a starting point – is already
available, seems to outweigh the possible drawbacks of generating a solution in
an inductive way.

As said, most students will try to work deductively if at all possible and if not
they may choose to work inductively. However, inductive and deductive
reasoning are not mutually exclusive as this generalisation may suggest. In fact,
it may be said that students combine both approaches and that they
complement each other. For instance, after reasoning inductively and generating
some feasible conjectures, students may recur to deductive reasoning to show
that these are always true.

Besides reasoning inductively and deductively, students may generate a solution
as a result of reasoning in terms of how data is created. The previous section
discussed how thinking in terms of how data is created may provide students
with information as to what to do next and why. Since this information is easily
translated into a solution, reasoning in terms of how data is created can be
considered as another way of generating a solution that is different to both
induction and deduction. Simon (1996) observed a similar situation. He
suggested that students may invent or infer situations to explain how data is
created and that this may allow them to generate a solution. The following
example illustrates the case of inventing a situation to explain how data is
created and how the understanding that it provides can be used to generate a

Ms. Goodhue: Mary, could you make an isosceles triangle by specifying
two angles and the included side?

Mary pauses and then punches in equal angles.

Ms. Goodhue: Can you tell me what you did?

Mary: Well, I know that if two people walked from the ends from this
side at equal angles towards each other, when they meet, they would
have walked the same distance.

Author [Martin Simon]: What would happen if the person on the left
walked at a smaller angle to this side?

Mary: (Without hesitation) Then that person would walk further [than the
person on the right] before they meet… (From Simon, 1996, p. 199)

Thinking in terms of how data is created can be seen as a way of gaining deep
understanding of the situation that helps generating a solution. Solutions
achieved in this way tend to be more ‘transparent’ than solutions arrived at by
deduction or induction. When students reason in terms of how data is created, it
may become evident how a solution should look like and why.

Guessing and Ungrounded Ideas

It was mentioned before that tentative solutions that are generated inductively or
in any other way are usually a good place to start generating a more
comprehensive solution. However, there does seem to be an exception to this
case. In some cases, students’ apparently inductive reasoning can be better
explained as ‘guessing’. When students guess a solution, their reasoning is
unclear and it is usually difficult to tell where ideas come from.Yet, from the
comments that students make, it usually becomes evident that they may be
testing their luck and proposing ideas without going through conscientious
reasoning about the situation.

Try completely new approach. Convert sequence into a straight number
using binary representation (might get lucky). (Sebastian, Ins and
Outs, p. 5)

We can see by looking at the diagram that there are three points that
would not be visible. Could I work this out algebraically so that it applies
to any size grid square?

Maybe it could be (i–j)/j, that would be (9–3)/3=6/3=2. That doesn’t

Maybe (i–j)/i would be better: (9-3)/2=3.Would this work for other (i, j)?

There only seems to be two points which means that my formula is not
correct. (Gina, Visible Points, pp. 3–4)

Ideas that are arrived at by guessing are usually ungrounded, i.e., they are more
the product of inventiveness than of carefully analysing the data. Although the
relation between guessing and ungrounded ideas is somewhat evident,
guessing a solution is not the only way in which students may generate this type
of ideas. Trying to invent a situation to explain how data is created may also lead
to generating ungrounded ideas, particularly when used without considering
sufficient empirical data. In other words, in an attempt to provide an account of
how data is generated or of how the “system in question works”, students may
fall into ‘making up’ an explanation that is more the product of their ingenuity
than of what they know about the situation.

Ungrounded ideas tend to be inconsistent and thus can lead to problems and
frustration. This was the case of a student that provided an interesting
explanation as to why it is not possible to build a fault-free rectangle (see the
‘Faulty Bricks’ problem). Since fault-free rectangles can be built, and since the
explanation was the result of the student’s creativity, she found it hard to
elaborate the argument further. In general, although ungrounded ideas can be
problematic, a positive aspect is that the frustration that they cause may
become, in some cases, a good place for starting to learn about the situation.

Summarising, students may generate an initial solution by reasoning
deductively, inductively or in terms of how data is generated. Although students
may have a ‘predilection’ for deductive reasoning, it seems that this predilection
is based more on their beliefs about mathematics (deductive reasoning being
‘more valid’) than on the results that they obtain from reasoning in this way.
Inductive reasoning may allow students to generate initial solutions that can
later be improved. Thinking in terms of how data is generated is a good way of
generating ‘transparent’ solutions. Although the last two types of reasoning may
not be the students’ first choices, they can be efficient ways of generating

Once a solution is generated, it may be validated and/or improved. The next two
sections look at ‘validating’ and ‘improving’ results, respectively.


During their problem solving processes, students look for ways of validating the
ideas that they are generating. To do this, they may try to validate their results in
terms of whether they are correct and make sense. In other words, students try
to verify that their results are correct and seek to explain why this is the case.
When students validate their results in this way their main concern is being on
the ‘right track’ and having a clear understanding of the situation. Thus, the
arguments that they produce can be considered as personal ‘proofs’ aimed at
convincing themselves, their peers and possibly even a sceptical reader trying
to follow their process (i.e., convincing oneself, a friend and an enemy, in
Mason’s (1982) terms).

Once students have achieved a satisfactory solution, they sometimes seek to
provide a formal mathematical proof of their work. However as the quote below
suggests, providing a formal argument seems to have a different purpose than
making sure that a solution is correct and makes sense.

This certainly seems to hold for all m, n [where m and n are natural
numbers], but whether or not I can prove it is a different matter.
(Leonard, Diagonals of a Rectangle, p. 19)

It seems that trying to provide a formal mathematical argument that proves that
a solution is true is more a way of improving a solution than of making it
convincing for themselves and for others. For this reason, providing a formal
proof will be discussed in the next section below (‘Improving Results’).

Making Sure Results are Correct and Make Sense

Students may validate their results by verifying that their ideas are correct and
make sense. In order to verify that results are correct, students may review their
reasoning and look for any errors or inconsistencies. For instance, they may
check that suitable procedures were chosen and that they were properly
conducted. Besides verifying their procedures, students may check to see
whether their generalisations work in particular cases. If the results obtained
from particular cases are as expected or match with previous data, then they
can be accepted. Verifying that results are correct allows students to move on,
whereas noticing any inconsistencies will require them to go back and try to
correct them.

Now I want to check it again that my result is right before I go any
further from here. Therefore I count the number of grid squared that are
touched by the diagonal again from the grid squares that I have already
drawn. And it’s correct. (Anibal, Sums of Diagonals, p. 4)

I will now see if it works for the numbers I have so far. (Jasmine, Sums
of Diagonals, p.6)

Check: Does this match the examples I have tried so far? (Julia,
Liouville, ca. p. 10)

Students may verify that the ideas being generated make sense by looking for
explanations as to why they must be true. Explaining why an idea is true
reassures students that the solution that they are generating is congruent with
their knowledge and with what they know so far about the situation.
Furthermore, when students try to make sure that their generated ideas make
sense they may resort to thinking in terms of how data is created.
Understanding of how the situation ‘works’ and how the data is derived from the
situation provides students with ideas that can be used to explain why a solution
must be true.

Why does this work? Aha! Looking at any diagonal, moving down one
adds 1 to the first element, 2 to the second, etc. And then finally one
more element equal to the new ‘x’. (Marcus, Sums of Diagonals, p. 3)

Trying to verify that results are correct and making sure that they make sense
are related activities that are usually combined. In many cases, after checking
that their results are correct, students may proceed to explain why this is the
case. The following quote illustrates this situation.

This looks like the number of creases is 2a-1.

Check for a=6.
From previous formula creases = 31+32=63=26-1.

I can see this would be true because each time I am doing n+(n+1) to
get the next term which is equal to 2n+1, so each time I am doubling
the previous number (which is less than 2n as 1 is one less than21=2)
which would give me 2n=2 and then adding one so I get 2n-1.
(Jasmine, Ins and Outs, p. 4)

This is not to say, however, that verifying that generated ideas are correct
implies that students will proceed making sure that they make sense. After all,
not all students are able to conclude their process by saying:

My calculations do work and make sense, and I think the answer is
reasonable. (Hannah, Faulty Rectangles, p. 11b)

In some cases, students may not be interested in explaining why ideas are true
so long as they seem correct. In other cases, students may be able to verify that
their results are correct but may find it difficult to provide an explanation as to
why this is the case.

It does seem to be the case that the Liouville results are always
identical, regardless of the chosen starting number. Sadly, I have no
theories as to why this occurs. (Conrad, Liouville, p. 5; emphasis

Continuously trying to verify that ideas are correct and make sense ensures that
inconsistencies are brought to the fore and provides an opportunity to amend
them. In fact, it seems that verifying that ideas are correct and make sense, and
making the necessary modifications, plays an important role in successful
solutioning. Inglis and Simpson (in press) suggest that it is error-correcting –
rather than error-free processes – that may account for the fact that
mathematicians perform better than non-mathematicians in logic tasks.
Furthermore, in a study of collaborative problem solving in combinatorics,
Eizenberg (2003) found that it was not peer collaboration that was directly
related to successful problem solving but that successful problem solving is
closely related to ‘control behaviours’, i.e., to constantly monitoring whether
ideas are correct and making the necessary modifications. In the author’s

Our study provides evidence that success in problem solving in
combinatorics is not a direct outcome of collaborative problem solving. It
is mostly a result of enhanced control behavior. (Eizenberg, 2003, p.

In spite of the benefits of validating results, students do not always stop to verify
that their results are correct and make sense. As said, validating solutions in the
ways discussed here may help to reassure students that they are on the ‘right
track’ in terms of the ideas that they are generating. This, in turn, will allow them
to continue with their solving process or, in other words, to ‘move on’. In some
cases, however, being able to move on can be more important than whether
results are correct and make sense. In such cases, students may simply avoid
trying to validate their results or will do it in superficial ways. For instance, they
may check that results are true in one or two known cases. In this way, even if
results are inaccurate, this will not necessarily prevent them from continuing to
work towards a concluding solution.

Number of rectangles formed is 3n+(n–4). E.g., when 5 dominoes are
used 15+1=16.

That seems to work! I will test the formula out when more dominoes are
used.[Continues to work with 3×1 rectangles] (Gina, Faulty Rectangles,
p. 3)

Being able to validate a result may provide students with an acceptable solution.
However, unless the student had already been working on improving this
solution, it is very likely that it will not be final but one that needs to be
improved. The next section looks at ways in which students may seek to improve
a solution once it has been achieved.


This section looks at what can be considered as the last stage of the solutioning
process. Once a solution is achieved, students usually acknowledge the need to
improve their results. This is particularly true when students feel that their
answer is correct but not ready to be presented as it is. If time and mathematical
knowledge allow, they may try to improve their results by providing a formal
mathematical proof or by extending their results to other domains. Alternatively,
they may try to express their solution in more concise ways.

OK – I’m happy that’s worked out in that case. I’m definite there is a
more elegant explanation which might be worth looking for. Argument
sounds a little awkward to me at the moment – could do with being
more persuasive.

Right. Review here – there’s a few different ways to go…
Have shown for odd x even, if I could show for even x even I’d be done!
(Rafael, Faulty Rectangles, p. 12)

I wonder if I could improve this further by rewriting my formula as a
closed expression, i.e., an equation in x and n with no summation signs.
(Hillary, Sums of Diagonals, 15)

Improving a solution can be a straightforward task that involves making simple
modifications or additions. However, this is not always the case and the work
that students need to conduct to improve a solution can vary from being
straightforward to very laborious and time-consuming. In most cases, improving
a solution will involve dealing with situations that are more complex than when
an initial solution was generated. Having to deal with progressively more
complex situations can make it difficult – or even impossible – for some students
to improve their solutions further. The probability of this being the case seems to
be higher when students lack the necessary mathematical background to deal
with more sophisticated mathematical ideas. Lack of time or energy can also
prevent students from improving their solutions. Under these circumstances,
some students will decide to stop their process and will present their solution as
it is.

Reached a dead end at the moment so I am unable to progress any
further. If I had been able to solve this problem properly I could have
also extended it to look at the rest of the items on my brainstorm.
(Lydia, Cartesian Chase, p.13)

Students who are able to improve their solutions recognise that it is almost
always possible to take them even further. However, they only have to continue
improving their solution until a seemingly acceptable solution is found. Such a
solution is one that is clearly (and if possible, formally) stated and that accounts
for a variety of cases.

Trying to Provide Formal Mathematical Proofs

One way in which students may seek to improve their results is by attempting to
produce a formal mathematical proof of their work. Once a satisfactory solution
or initial solution is generated, students may try to improve it by providing a
more rigorous argument. Providing a formal mathematical argument is a way of
putting an already satisfactory solution in such a way that it can be presented as
a final product to others. In other words, providing a formal mathematical proof
involves elaborating a deductive argument that not only satisfies the student’s
understanding but also satisfies certain mathematical requisites.

Producing a formal mathematical proof is something that some students do as
part of their processes. For instance, in ‘Sums of Diagonals’ various students
proved their general formulas by mathematical induction. However, in general, it
may be said that providing a rigorous mathematical proof is usually considered
a secondary aim. For some students, the fact that the results are reliable should
be evident from the way they were generated and validated.

I believe I have the correct answer, although I have no concrete proof. I
believe that, as a possible extension, it would be possible to get an
answer involving trigonometry…This would be a concrete ‘proof’ of the
answer but it isn’t very easy to show. Other extensions [could be]…
(Roberto, Diagonals of a Rectangle, p. 5)

My formulas are very general and because of the way they were
obtained they don’t really need any formal proof or justification, as these
are evident in the method. (Nadia, Sums of Diagonals, p. 7b)

In general, students seem more concerned about producing arguments that are
convincing, both for themselves and for a sceptical reader than of providing a
formal mathematical proof. Moreover, when it comes to improving their solution,
they seem to be more concerned about extending their results, as will be
discussed next.


Once students generate a solution, it is not uncommon for them to try to
improve it by extending it. Students extend their solutions by showing that they
account for all possible cases or by making their results valid for a wider

When students generate a solution, they sometimes notice that the ideas or the
methods that they used can be applied to other situations as well. In other
words, they notice that some of their ideas can be transferred and thus be made
useful for solving, or dealing with, other cases – i.e., for extending.

Aha! If I can do this for a number with two divisors that are prime, I
could probably do it for a number with exactly 3, 4, … or more nontrivial
divisors, all which are prime. (Jason, Liouville, p. 3)

Can I use the same process as earlier to generate more even x even
fault free rectangles? (Camille, Faulty Rectangles, pp. 2b–3)

Although transferring means that previously developed ideas will be used in
other situations, this is not necessarily a simple task. Transferring may require
students to make some changes to the ideas or procedures to be transferred to
make them suitable for the new situation. These changes can be relatively
simple, such as when students decide to introduce a new, more efficient

The largest secret number ‘a’ was found by adding the two largest side
numbers and subtracting the remaining side numbers…I think [this] rule
is most likely to work with arithmagons with >3 sides.

As I am seeing a general rule for arithmagons with n sides, I will need
to alter my notation for improved clarity. Instead of x, y and z for the side
numbers I will use s1, s2, s3, …, sn… (Jules, Arithmagons, p.5)

In other cases, adapting previously used methods or ideas can be complicated
or even impractical.

My proof that there was a path came from visualising, again, what the
path should be, since anything other than the circle seemed unlikely,
and bearing in mind the complete symmetry of the circle. Unfortunately,
this reliance on the symmetry of the circle meant I couldn’t extend the
theory to irregular circles very easily. (Albert, Jogger’s Dog,

In some cases, adapting can be a considerably complicated activity. In situations
like this, students will find that looking for new ways of generating a solution may
be a better option. In a way, finding new ways of solutioning may suggest that
students will need to start the solving process all over again. However, this is
not the case. The knowledge and understanding that students have gained
about the situation are very likely to make this ‘new’ process a more efficient
one. Of course, this will be the case only if students persist in extending their
solutions. They may well decide to stop their process at this stage.

[please see PDF version for appedix problems]


Maria de Hoyos Guajardo, Ph.D. Candidate, M.Sc., B.Eng.
Research Fellow
Institute for Employment Research
Warwick University
Coventry UK


Phone: +44 (0) 779 6614243


Barnes, M. (2000). “Magical” Moments in Mathematics: Insights into the Process of
Coming to Know. For the Learning of Mathematics, 20(1), 33-43.

Eizenberg, M. M., & Zaslavsky, O. (2003). Cooperative Problem Solving in Combinatorics:
The Inter-Relations between Control Processes and Successful Solutions. Journal of
Mathematical Behavior, 22(4), 389-403.

Fischbein, E., & Grossman, A. (1997). Schemata and Intuitions in Combinatorial
Reasoning. Educational Studies in Mathematics, 34(1), 27-47.

Inglis, M., & Simpson, A. (in press). Mathematicians and the Selection Task. Proceedings
of the 28th International Group for the Psychology of Mathematics Education, Bergen.

Mason, J. H., Burton, L., & Stacey, K. (1982). Thinking Mathematically. Avon: Addison-

Raman, M. (2003). Key Ideas: What Are They and How Can They Help Us Understand
How People View Proof? Educational Studies in Mathematics, 52(3), 319-325.

Simon, M. A. (1996). Beyond Inductive and Deductive Reasoning: The Search for a
Sense of Knowing. Educational Studies in Mathematics, 30, 197-210.

Skemp, R. R. (1976). Relational Understanding and Instrumental Understanding.
Mathematics Teaching, 77, 20-26.

Thurston, W. P. (1995). On Proof and Progress in Mathematics. For the Learning of
Mathematics, 15(1), 29-37.