You are on page 1of 24

REVATHY PARAMESWARAN

ON UNDERSTANDING THE NOTION OF LIMITS


AND INFINITESIMAL QUANTITIES

Manuscript submitted 15 February 2006; Final version received 24 May 2006

ABSTRACT. In this paper we explore the influence of students’ personalized notion of


Fsmall_ numbers based on real life experiences on their understanding of limits. Tests
were conducted on two samples of students. The first sample, consisting of students in
the XII grade, had been taught limits using an informal approach (i.e., without recourse
to the    definition) and the second sample, consisting of first year undergraduates,
had been taught the formal    definition of limits. Our research points out that most
students in both samples round off to zero such Fsmall numbers_ while evaluating limits
wherever such numbers might occur because they perceive limit as a process of
approximation.

KEY WORDS: approximations, formal and informal definitions, graphical


representations, limits

INTRODUCTION

Learning calculus has been the subject of extensive research for a long
time. One of the significant conclusions arising out of this research is
that students in general develop routine techniques and manipulative
skills rather than conceptual understanding of the theoretical concepts
(Berry & Nyman, 2003; Ervynck, 1981; Sierpinska, 1987; Robert, 1982;
Davis & Vinner, 1986). Asiala, Dubinsky & Schwingendorff (1997)
report that many researchers have found that students have a weak
concept image of function and that they tend to rely on the need for
algebraic formulas when developing the function concept.
The subject of calculus is rich in abstraction and calls for a high
level of conceptual understanding which many students find hard to
cope with. Ferrini-Mundy & Graham (1991) argue that the students’
understanding of central concepts of calculus is exceptionally
Fprimitive_: Bstudents demonstrate virtually no intuition about the
concepts and processes of calculus; they diligently mimic examples
and their attempts to adapt prior knowledge to a new situation usually

International Journal of Science and Mathematics Education (2007) 5: 193Y216


# National Science Council, Taiwan 2006
194 REVATHY PARAMESWARAN

result in very persistent and often inadequate conceptions whose change


the students firmly resist.^
While attempting to reach a wide range of students in teaching
calculus, it is perhaps more practicable to appeal to one’s intuition to
convey mathematical concepts and ideas, building on what they have
already learned, without making heavy demands on their aptitude for
abstract and rigorous mathematical understanding. The author is broadly
in agreement with the following opinion expressed by Koirala (1997):
BAn introductory calculus course should be informal, intuitive and
conceptually based mainly on graphs and functions... Formulas and rules
should not be given as granted but they should be carefully developed
intuitively on the basis of students’ previous work in mathematics and
sciences^ (see also Heid, 1988; Orton, 1983).
The concept of limit is the cornerstone of several related concepts
such as continuity, differentiability, integration, convergence of se-
quences and series, etc. The precise, formal definition of the concept of
limit is so complex and counterintuitive that it fails to bring out readily
the simple and intuitively obvious ideas which led to it in the first place.
The    definition of limit involving universal and existential
quantifiers is designed to solve mathematical difficulties and not
psychological ones (Williams, 2001).
The purpose of this paper is to report on an experiment the author
carried out to study students’ understanding of the limit concept.
Researchers have described students’ beliefs about the limit process in
various ways such as a dynamic process, limit being unreachable, limit
as approximation, limit as a bound, etc. (Williams, 1991; Nardi, 1996). It
is common practice in real life as well as in classroom that one often
Fignores_ negligible quantities andpffiffiffi rounds off numbers to convenient
significant digits. For example, 2 is usually taken as 1:414, and the
value of  is often taken to be 22=7 or 3:14. While this practice may be
of practical value, can it be a source of conflict in understanding the limit
concept? In this paper the author attempts to understand students’
conception of limit by introducing questions about limit involving
extremely small numbers or extremely large numbers in the expression
for the function involved. The purpose of introducing such quantities in
the functions is to challenge the students’ commonly held beliefs such
as: limit is a process of approximation, and limits can be evaluated by
computing the function value at a few sample points close to the point at
which limit is to be evaluated. We explore how real life experiences and
the classroom practices of rounding off of real numbers to a convenient
rational number affect students’ understanding of limits. The issues that
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 195

initially arose concerning students’ comprehension of the concept of


limits and which led to the present study were the following:
(i) What are the errors, if any, that seep in and affect understanding the
concept of limits, where Fsmall_ numbers are used as parameters in
the definition of the function, when the students have not been
taught the    definition?
(ii) How are such erroneous practices overcome when the students are
exposed to the    definition?
The author has attempted to understand mental constructions made
by the students regarding the notions of infinitesimals and limits. This
was based on classroom experiments which consisted of tests and
interviews.

Research Framework
Concept Image and Concept Definition. We recall here a brief account of
the learning theory developed by Tall & Vinner (1981) which is
particularly relevant to students’ understanding of the notion of limit.
They introduced the notions of concept definition and concept image. A
concept definition specifies, in symbols and words, precisely what is
needed for a notion to be an instance of a concept. A concept image is a
learner’s mental picture together with all the associated links to other
cognitive structures. The concept image may not be a faithful represen-
tation of the concept definition. Dreyfus and Vinner (1989) suggest that
concept images are not formed merely by the definition but by
experience. The concept images are therefore personal and experiential.
Barnard & Tall (1997) define cognitive units as parts of concept
images that one focuses on at a given time. They may be any aspect of a
concept image. A rich concept image would include not only the
definition but also many links among the cognitive units. Not all its parts
are evoked simultaneously and it is quite possible that some parts of it
are incompatible with others. These give rise to potential conflict factors.
The potential conflict becomes a cognitive conflict when the contradic-
tory cognitive units are evoked simultaneously.
The student may be unaware of the conflict if the implications of the
concept definition are not a part of his concept image. The lack of
coordination between the concept image that one develops and the
implications of the concept definition can lead to obstacles to learning.
This has led to the notion of cognitive obstacles (Brousseau, 1983). Also,
there can be obstacles due to the complexity inherent in the notion that is
196 REVATHY PARAMESWARAN

being learnt; such obstacles are referred to as epistemological obstacles


by Bachelard (1938).
Fischbein (1993) has proposed the notion of figural concept to refer to
concepts which are not merely abstract notions but have an image
represented by figures as well. These geometrical figures represent
mental constructs that simultaneously possess conceptual and figural
properties. The two aspects, conceptual and figural, should be harmo-
nized to achieve understanding of the concept.

Literature on Students’ Understanding of Limits


The concept of limit is a topic which has been extensively researched by
mathematics educators (Cornu, 1983; 1991; Davis & Vinner, 1986; Li &
Tall, 1993; Sierpinska, 1987; Williams, 1991; 2001). There are several
models attempting to make sense of students’ understanding of limits
(Vinner, 1983; Tall & Vinner, 1981; Cornu, 1983; Cottrill, Dubinsky,
Nichols, Schwingendorf, Thomas & Vidakovic, 1996; Williams, 1991;
2001).
Mathematical concepts where a process and the output of that process
are represented by the same symbol are called procepts (Gray & Tall,
1994). The full symbolism of limit which is a procept evokes different
concept images leading to different conceptions or mental models that
students develop about limit. These conceptions include notions that
limit is a boundary, that functions cannot reach their limits and that limit
is a dynamic process of points or numbers Bgetting close^ to a limit point
or number.
Sierpinska (1985) has attributed students’ conceptual difficulties
involving limiting processes to be closely associated to their perceptions
about infinity. She has classified the students who use the word infinity
but in fact refer to very large or very small quantities as unconscious
infinitists. The literature is abundant with descriptions of cognitive
obstacles caused by these beliefs.
Some authors (Cornu, 1980; Sierpinska, 1990; Szydlik, 2000) report
that a high percentage of students who have what they call a static view
of mathematics are able to deal with only specific calculations that are
placed before them. These students can evaluate limit of a function at a
specified point by finding the value of the function at that point or at a
point close to it by using the formula given.
Cottrill et al. (1996) argue that when any calculation involves an
infinite number of steps it can be understood only through a process
conception. The process conception of limit is also sometimes referred to
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 197

as the dynamic model of limit. While this dynamic conception is easily


developed in students (Tall, 1981; 1992), the difficulty lies in the
transition from the dynamic conception to understanding the formal
definition of limit (Ervynck, 1981; Williams, 1991, 2001). Also, many
students having a dynamic concept of limit seem to believe that the limit
is Funreachable_ or is attained only at infinity (Tall, 1992; Williams,
2001; Mamona-Downs, 1990).
Tall (1980a) observes that when students are taught limits in an
informal manner, with formal definition coming much later if at all, Ba
concept image is built up long before any formal concept definition is
given.^ This approach leads to a state where Bcertain implicit properties
which are not part of the concept definition become part of the concept
image.^ This is very relevant to our study of students’ understanding of
the notion of limits. Students who have experiences of quantities
becoming small in limiting processes begin to imagine a variable
quantity that can be Fas small as you like_ (Cornu, 1983; 1991).
According to Tall, Bthe language used in classrooms tends to promote
such thoughts, leading to individuals thinking of variables tending to
zero as being arbitrarily small.^
Cottrill et al. (1996) argue that the limit concept is not a single pro-
cess but a more complicated Fschema._ It involves a pair of coordinated
processes in which the values of the function approaching a limit are to
be observed as the values of the independent variable approaching a
specified point. Constructing this schema is essential for understanding
the limit concept. They conjecture that it is the requirements of
constructing a schema together with the need for a sophisticated use of
existential and universal quantifiers rather than the formal nature of the
definition of limit that makes the limit concept inaccessible to most
students.
The essence of the research that has been discussed so far reveals the
extent of the students’ imagery built on their previous experiences and
their difficulty in coping with the potentially infinite process. What few of
these papers have taken into account is the full nature of the students’
previous learning experiences with arithmetic using decimal representa-
tion and finite approximations. We believe that this experience can give
rise to epistemological obstacles relating to the need to simplify the new
ideas in a way which fits with the students’ experience. What is more
natural than to reduce the limit problems by taking simple approxima-
tions? The problem, of course is in the fact that taking the approximations
before the limit can lead to misconceptions, thus adding to the cognitive
difficulty in making sense of the limit process as a formal limit concept.
198 REVATHY PARAMESWARAN

RESEARCH PROCESS

Aim
As mentioned in the introduction we aim to understand students’
conception of limits by exploring the effect of real life and classroom
practices of rounding off real numbers to convenient decimals. The
questions which initiated our investigation are listed in the introduction.

Concept of Limits in the Indian CBSE Curriculum


In the curriculum of the Central Board of Secondary Education (CBSE)
in India, the students are exposed to the study of the limit of a function at
a point when they are in grade XII. They are taught in detail about
functions and their graphical representations. The notion of limit is
introduced with the help of graphical representation of a function.
Usually the discussion of limit is restricted to class of rational,
trigonometric functions, exponential and logarithmic functions, while
functions involving split domains are also considered. The formal   
definition is not given.
Sample I This sample consisted of two groups of students all of
whom were in grade XII. They were heterogeneous in their
mathematical abilities. The first group of students belonged
to the same school in which the author was a teacher and
the other group was comprised of students from different
schools. These schools come under the CBSE. The students
were taught the notion of limit in an informal manner with
examples of both algebraic and graphical representations of
limits. The basic properties of limits were explained without
proofs. However, proofs of some standard limits such as
x
lim!0 sin  ¼ 1; limx!0 e x1 ¼ 1 were worked out in the
class. The test was conducted for the first group of students
on limits after an intensive interactive learning programme.
The total number of students in this sample was 68. Ten
students whose answers were representative of the approaches
adopted and who had expressed themselves more clearly in
their test were then interviewed individually one day after the
test.
Sample II The second sample consisted of eleven students who are first
year mathematics majors in Chennai Mathematical Institute,
Chennai. These students were admitted after a very difficult
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 199

nation-wide entrance test in mathematics. A few of them were


recipients of medals in the International Mathematics Olym-
piad. The instructors were drawn from premier research
institutions in the country. The students were undergoing a
course in Calculus I when the test was conducted. These
students have been exposed to the formal definition of limits.
Six students were interviewed one day after the test.

Test Items
The test questions were formed by the author in order to explore and
understand students’ conception of limit based on real life and classroom
practices of approximating whenever convenient to do so. The
parameters which occur in the definition were so chosen that they are
extremely tiny but can significantly influence the nature of limit. We
shall give solutions to these questions in the Appendix for the
convenience of the reader.

List of Questions
(
x if x 6¼ 21000
1. Let f ðxÞ ¼
1000 if x ¼ 21000 :
Evaluate (a) limx!0 f ðxÞ: (b) limx!21000 f ðxÞ.
1
2. Find limx!0 xð0:1Þ 10000 if it exists.
3. Evaluate limx!1 f ðxÞ where
8
< 1 if 0  x G 1  ð0:1Þ1000
f ðxÞ ¼ 1x
: 2004
10 ð1 þ xÞ2004 if x  1  ð0:1Þ1000 :

4. Compute limx!0 gðxÞ where


8
>
> x if xG0
>
>
< 1 1
gðxÞ ¼ if 0x
> 21000 21000
>
> 1
>
:x if x> :
21000
In addition to the above, the following question was also included in the
test for the students of Chennai Mathematical Institute.
200 REVATHY PARAMESWARAN

Figure 1. Graph of y ¼ f ðxÞ.

5. Let f : R!R be the function whose graph is given in Figure 1. Set


x0 ¼ 1, x1 ¼ 0:9.
(i) Let ¼0:25. Find a  such that jx  x0 jG  )j f ðxÞf ðx0 ÞjG 0:25.
(ii) Find an  > 0 for which there exists no  such that jx  x0 j G 
implies j f ðxÞ  f ðx0 Þj G e.
(iii) Let x1 ¼ 0:9,  ¼ 0:25. Find the largest possible  such that
jx  x1 jG  implies j f ðxÞ  f ðx0 Þj G .

Results
Sample I

Question no Number of correct responses Percentage of correct responses (%)

1(a) 18 27
1(b) 11 15
2 22 30
3 13 20
4 19 28

Sample II
Question no Number of correct responses Percentage of correct responses (%)

1(a) 5 45
1(b) 4 34
2 6 56
3 4 34
4 4 34
5 9 87
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 201

Data Collection and Analysis


Our data analysis broadly fits into the grounded theory developed by
Glaser & Strauss (1967). We have tried to make sense of the data in an
informal way. We tried to construct Fmeaningful patterns of facts_
(Jorgenson, 1989) by looking for structures in the data. We tried to
achieve this by looking for commonalities and differences in the
answer scripts. Then we focused on those answer scripts which
displayed a discernible pattern for further analysis. We conducted a
fine-grained analysis of these scripts in order to find those aspects
which could possibly serve as criteria for generating our model of
students’ concept of limit. The interview was conducted based on their
written responses, and, as it progressed, further questions were asked
building on their responses, so as to gain better understanding of the
student’s internal representations of the limit conception. The questions
also aimed to provoke considerable amount of reflection on the part of
the students and attempted to evoke conflicts by making them confront
their errors. The interview aimed at clarifying and ascertaining our
understanding of the mental constructs of the students based on their
answers.
The result of this experiment helped us to gain an insight into
student’s mathematics. Based on our experiment, we shall attempt to
explain, in the next section, our model for understanding these students’
mental representations of limit as an approximation paradigm that is
consistent with our data collection.

Sample Responses
We give below the responses of some of the grade XII students to
various questions. A sample of responses of CMI students followed by
their analysis is given in the next subsection. These are typical
representatives of the strategies adopted to find solutions to the
questions.

Question #1

Student #1: limx!21000 f ðxÞ ¼ limx! 10001 f ðxÞ:


1000 2
2 is an extremely large number, i.e., 1. Therefore, 1=21000 ¼ 1=1 ¼ 0.
limx!0 f ðxÞ ¼ limx!0þ f ðxÞ ¼ 1000:

Student #2: limx!0þ f ðxÞ ¼ limx!0þ 1000 ¼ 1000:


limx!21000 f ðxÞ ¼ 1000:
202 REVATHY PARAMESWARAN

Student #3: If x ! 0; f ðxÞ ¼ 1000 (as x ¼ 21000 $ x ! 0).


If x ! 21000 ; f ðxÞ ¼ 1000:

Analysis of Students’ Responses. Students who could not solve question


#1 typically came up with one or more of the following: (i) limx!0
f ðxÞ ¼ 1000, (ii) limx!21000 f ðxÞ ¼ 0 (iii) limx!0 f ðxÞ does not exist, (iv)
limx!21000 f ðxÞ ¼ 1000. The responses reveal that the students were
misled or confused by the jump discontinuity at 21000 . One possible
explanation could be that since the number 21000 is perceived to be
extremely small in relation to size of objects one usually encounters in
real life the students perhaps believe that as x ! 0 it somehow affects the
value of the limit at x ¼ 0. Would their answer to 1(a) be different if the
function had the jump discontinuity at, say, 1=2 instead? Those students
who were interviewed had no difficulty in doing both parts of the
modified question correctly, although a few still mentioned that limit at
1=2 was 1000. Among those who had given correct answers to 1(a), but
not 1(b), most have merely evaluated the function value at 0 and 21000 ,
respectively (note that the given function is continuous at ¼ 0 but not at
x ¼ 21000 ).

Question #2
1 1 10000
Student #4: limx!0 xð0:1Þ10000 ¼ 00 (since ð0:1Þ ! 0 it is almost 0Þ =
1
0 ¼ 1:
Student #5: It does not exist as ð0:1Þ10000 is so small that it can be
neglected.

Analysis of Students’ Responses. In the case of students #4 and #5 when


the above problem is encountered the concept image of limx!0 1x is
evoked. Observe that student #4 writes Bð0:1Þ10000 ! 0^. One possible
explanation why the student rounds off the number ð0:1Þ10000 to zero is
that he fails to realize that the constants in the problem are not to be
approximated in the process of finding the required limit. The student
approximates the function before applying the limit. This could be
because he believes that it is quite appropriate to make such
approximations when convenient, as is commonly accepted practice in
classrooms and in real life.
Another possible explanation is that it is a result of having been
immersed in social and classroom practices of referring to very small
numbers during a dynamic limiting process, in which values are
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 203

conceived of as getting smaller, as zero in the limit. Perhaps these


students have never seen such small numbers used as parameters in the
definition of the function. Such numbers themselves trigger the idea of
Fcan be ignored in the limit,_ resulting in errors.
The following excerpt from an interview with student #4 shows that
he fails to recognize the same function when presented in a different
form, without involving Fsmall_ numbers. The interview was held one
day after the test.
10000
I: What is limx!0 1010
10000 x1 ?

Student #4: I think the limit is 1010000 :


I: Is this question not the same as question #2 given in the test?
Student #4: This is different. The numbers are different.

We infer that this time, the evoked concept image is not that of
evaluating the limit of 1=x at 0 as in the test question.
One possible explanation for this could be that the number 1010000
is the coefficient of the variable x and not a Fconstant_ in the function. As
the student fails to realize that this function is the same as the one given in
the test, although presented differently, he stays clear of the pitfall and
evaluates the limit correctly. There is no cognitive conflict evoked in his
thought process as, in his mind, Fthe two functions are not the same._
Another possible explanation is also that the student merely
substituted x ¼ 0 in the function to evaluate limit, without approximating
the function as the number 1 in the denominator being not a Fvery small_
quantity to be ignored.
During the interview student #5 was asked to work out the same
question with ð0:1Þ10000 replaced by ð0:1Þ. This time he worked it out
correctly. When asked about his solution to the original question, he said
again that the number ð0:1Þ10000 Ftends to zero in the limit_ and so the
limit does not exist. The student did not seem to have realised the
inconsistency in his argument.

Question #3

Student #6: ð0:1Þ10000 . It is very small. It is equal to zero.


limx!1 f ðxÞ ¼ 1
limx!1þ f ðxÞ ¼ 22004 ð10Þ2004 :
1
Student #7: limx!1 11 ¼ 10 (where 0  x G1) limx!1þ f ðxÞ ¼ 22004
2004
ð10Þ (where x > 1  0 i:e:; x > 1).
204 REVATHY PARAMESWARAN

1
Student #8: limx!1j f ðxÞ could either take values as per 1x in which
1000
case the limit does not exist as 1  ð0:1Þ tends to one or if x is in the
other interval limit exists and is equal to ð2Þ2004 ð10Þ2004 :

Analysis of the General Response to this Question. In general, the


response of the students seems to be in writing the left hand limit as
1
lim f ðxÞ ¼ lim
x!1 x!1 1x
suggests the following: That, since 1  ð0:1Þ1000 G 1, it is possible to find
a neighbourhood to the left of 1 is the required mental construction not
performed by most of the students. This probably indicates difficulties
posed by presence of Bsmall^ numbers involved in the problem. The
evoked concept image is that of the function
8
< 1
if 0  x G 1
FðxÞ ¼ 1x
: 2004
10 ð1 þ xÞ2004 if x  1:
This is another instance of students using approximations before
taking the limit as approximations were common practice in his previous
experience.

Question #4

Student #9: limx!0 gðxÞ ¼ 0, limx!0þ gðxÞ ¼ 0,


1 1
when 0  x  21000 ; f ðxÞ ¼ 21000 ¼0
1000
2 is a very large quantity and hence tends to infinity. Therefore,
1 1
21000 ¼ 1 ¼ 0:

The above answer also leads to the conclusion that the student
approximates Fsmall_ quantities to zero. The presence of small numbers
in the function causes confusion, conflict, or obstruction to computation
of the limit. The immediate response (perhaps triggered on by many
years of practice in dealing with approximate arithmetic) is to overcome
the obstruction by approximating the function to a Fsimpler_ one before
evaluating the limit. In the process, however, the essential features of the
function are altered resulting in error when limit is evaluated.
1
Student #10 (see Figure 2): We see that 21000 tends to zero. Left hand
1
limit = limx!0 gðxÞ ¼ 0 Right hand limit = limx!0þ gðxÞ ¼ 21000 if
1 1
x G 21000 . If x > 21000 and yet tends to zero, limx!0þ gðxÞ ¼ 0: LHL 6¼
1 1
RHL. The given limit does not exist for x G 21000 but exists for x > 21000 :
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 205

Figure 2. Response of student #10 to question #4.

Sample Responses of the Students of the Chennai Mathematical


Institute Question #1
Student #11: limx!0 f ðxÞ ¼ 1000:
As, 21000 may be taken to tend to 0. Therefore x may be taken ¼ 21000 .
This student’s assertion B21000 may be taken to tend to 0^ calls for
some explanation. Many others have made identical or similar statements
(see responses of students #3, #4, #8). One possible explanation is that,
deeply influenced by practical arithmetic of approximating very tiny
quantities to zero, these students view such quantities as Ftending to zero
in the limit_. This seems to strongly suggest that the students view limit
as an approximation. This would be consistent with remarks such as
B21000 may be taken to tend to 0.^ An alternative explanation is that a
very small number seems to evoke the concept image, not that of a single
point distinct from 0 on the real line, and therefore at a positive distance
from the origin, but of a variable point which is arbitrarily close to zero
which in the limit is equal to zero. Also yet another explanation is that
influenced by years of real-life and classroom practices where often
0:0000001 is approximated to 0, perhaps, the number 0:0000001 triggers
in their minds the model not of an abstract point on the real line but a dot
having positive radius on a line segment (representing the reals) which is
indistinguishable from the point representing 0. This is consistent with
206 REVATHY PARAMESWARAN

the practice of rounding off 0:0000001 to 0. The result is that they


unconsciously have a model of a discrete set of points representing the
real number system: a given real number  represented by a point P is
approximated to a Fconvenient_ one  represented by a point Q which,
when plotted on a line segment (representing the real number system), is
visibly indistinguishable from P (following Tall, 1980b, 1980c) F is d-
indistinguishable_ from  where d is the diameter of the point
representing a number). We are led to conclude that models of the real
number system held by the students can be a source of this error. It is
beyond the scope of this paper to discuss the various models of the real
number system held by students and factors influencing their formation.
Student #12: (He has done 1(a) correctly.) limx!0 ¼ 0, given any ,
choose  such that  G minf; 21000 g. It is obviously the required . (b)
limx!21000 f ðxÞ does not exist. Because if the limit was p, take  ¼ 1. But
given any , if q G minf; 1g, j f ðpÞ  f ðqÞj > 1000  1 ¼ 999:
Most of the students who did 1(a) correctly but not 1(b) have given
similar explanations, confusing limit and continuity. Those interviewed
after the test corrected themselves quickly.
The following is a sample response of a student who fails to solve
question #1 using the formal definition, while it could have been easily
solved using the informal notion of limits.

Student #13: limx!21000 f ðxÞ ¼ 0 as x ! 21000 ) x 6¼ 21000


Therefore f ðxÞ ¼ x and therefore j f ðxÞ  0j G  ( > 0) ) jx  0j G 
 G x G  )   21000 G x  21000 G   21000 G  þ 21000 )
jx  21000 j G  þ 21000 ¼  (say). Therefore f ðxÞ ! 0 as x ! 21000 .
Student #13 ’s solution to question #5:
(i) Here x0 = 1,  ¼ 0:25.
jx  x0 j G  ) j f ðxÞ  f ðx0 Þj G 0:25.
jx  1j G  ) j f ðxÞ  1:2j G 0:25. After some calculation he obtains:
) 0:95 G f ðxÞ G 1:45.
Now x ! 1þ ; f ðxÞ ¼ x þ 0:4. Therefore, 0:95 G x þ 0:4 G 1:45
0:55 G x G 1:15 in case x ! 1þ .
Now, x ! 1 , f ðxÞ ¼ x. ) 0:95 G x G 1:45 in case x ! 1 .    that
is 0:05 G x  1 G 0:45.
The common interval is 0:05 G x  1 G 0:15. Therefore, 0:05 G
x  1 G 0:05. i . e . , jx  1j G 0:05 ) j f ðxÞ f ðx0 Þj G 0:25.   
Therefore,  ¼ 0:05:
(ii) It is clear from the graph that such an  is  ¼ 0:1.
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 207

(iii) Here
8
< x þ 0:4;
> x>1
f ðxÞ ¼ 1:2; x¼1
>
:
x; xG1

   Therefore, j f ðxÞ  0:9j G  ¼ 0:25. j f ðxÞ  0:9j G 0:25 ) jx  0:9j


G 0:25. Therefore, x ! 0:9þ ; f ðxÞ ¼ x and also x ! 0:9 , f ðxÞ ¼ x.
Therefore, the largest  ¼ 0:25.

Analysis of Response to Question #5. The response of student #13 is


representative of the general response to this question. Most of the
students were able to give the correct answer. That student #13 was able
to solve question #5 using the    (as did most others) definition shows
that he understands the formal definition of limits. However, he seems to
mechanically unravel the definition rather than taking clues from the
graphical representation of the function.
One possible reason for not taking hints from the graphical
representation is that some students believe such a proof to be imprecise
or non-rigorous. Such a belief could have been imbued from the
instructors’ emphasis on the    definition as the rigorous, mathemat-
ically accurate alternative for the intuitive, graphical approach to the
notion of limit. The following statement, from Nardi (1996, Ch. 7), lends
support to such an explanation: B[The students’] difficulty with finding
limits either via the definition or via the algebra of limits was largely
attributed to their growing mistrust towards school mathematical
knowledge. Using inequalities in order to manipulate quantities, graph-
ing functions, guessing limits and using the algebra of limits are
mathematical practices that the students questioned as to their rigour and,
hence, as to their acceptability.^
The theory of figural concepts is relevant in the context of limit
problems, particularly those based on graphical representation of func-
tions. Fischbein (1993) introduced the notion of figural concepts to refer
to mental entities which contain both conceptual and figural properties.
Note that in his response for 5(iii), student #13 translated the graphical
representation of the function into an algebraic expression to work with.
We note, however, student #13 did make use of the graph to complete
5(ii).
In the context of problems on limits based on graphs, the geometry
(the shape, the fluctuations, the discontinuities, etc.) of the graph
208 REVATHY PARAMESWARAN

contains vital features of the function that it represents. Of course, the


actual Figure 1 (drawn with pencil on a paper, or with chalk on a
blackboard) is itself a pictorial representation of the mathematical entity,
namely, graph. While the mathematical graph is a faithful representation
of the function, the pictorial representation of the graph itself can only be
a rough approximation (The pencil drawing will have a definite
thickness, and points will have positive diameter, whereas the geomet-
rical point is an abstract entity). Although this could itself cause
problems in understanding, this kind of pictorial representation is
commonly used in classrooms. The CMI students did not have any
difficulty in translating into algebraic expression the pictorial represen-
tation of the graph of the function, but many of them did not go further
on to guess the relevant ’s and ’s from the figure itself in order solve
the given problem.

DISCUSSION AND CONCLUSION

We attempted to study the influence that real life and classroom practices
of approximating small numbers to zero has on students’ learning of the
conception of limit. We had begun with a general premise that any
cognitive obstacles experienced by those who had been taught the concept
through an informal approach (such as using graphical representation of
functions) would be easily overcome once the students are exposed to the
formal    definition of limit.
Taking an intuitive approach to teaching the limit concept, typically
the students are led to the notion of limit (of a sequence) as a process of
successive approximation leading to a possibly unattained value using a
sequence of smaller and smaller numbers, or when limit of a function at
a point is explained in terms of the graphical representation of the
function, as was the case with our students in sample I. The formal   
definition is given in a higher level course, which our students of the
second sample underwent.
The very fact that there are multiple representations of this notion of
limits implies that there is scope for potential conflicts in developing
different concept images.
Limits as Approximation
Students in both samples were aware that the set of positive integers is
unbounded, that the set of rational numbers is dense in the set of real
numbers, and that the Archemedian property holds for the real number
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 209

Figure 3. Model showing the error triggered by presence of very small numbers.

system. However, based on our research, we are led to conclude that the
links among these knowledge components are yet to develop. This is
evidenced by our findings that the students identify what they perceive as
Flarge numbers_ with infinity and Fsmall numbers_ with zero. Also, in our
experiments, they unwittingly rounded off very small parameter values
occurring in the definition of the function in question to a convenient
number close to it. This action was perhaps prompted by the feeling that it
simplified the algebraic expression of the function near the point at which
limit needed to be evaluated. An alternative explanation is that the
language of approximation was applied to parameter values, triggered by
its use as part of a dynamic process of approaching the limit (see Figure 3).
Our analysis of the students’ responses clearly points out that most of
the students in our samples view limiting as a process of approximation
when very minute quantities are involved in the definition of the func-
tionVbe it in the domain, algebraic expression, or the point in the
domain where the limit is to be evaluated. They tend to approximate the
given function by changing or ignoring quantities appearing in its
definition which they perceive as Fsmall_ constants to zero. For example,
1 1
many students treat xð0:1Þ 10000 to be the same as x , although the behaviour
210 REVATHY PARAMESWARAN

of the function near zero of the two functions are significantly different.
We surmise that at the root of such misconceptions is the practice, both
inside and out of the classroom, of approximating numbers to convenient
significant digits. Even while learning calculus, the students in our
samples seem to have continued this practice of approximate arithmetic
almost unconsciously. Here, by Fapproximate arithmetic_ we refer to
arithmetic using approximation to a certain number of significant digits.
Indeed the parameters in the definition of the functions in the test
questions and the point where the limit is to be evaluated were so
designed as to be incomprehensibly small or large (so to speak) with a
view to bringing about a cognitive conflict, should such approximate
arithmetic be indiscriminately applied. We wanted to see how students
negotiate their way through the challenge posed to them. The result of
our experiment highlight the effect that the habit of approximate
arithmetic has on the students’ conception of limit. That the students of
the second sample, who had learned the    definition, also committed
the same errors as the first shows how deeply ingrained this habit is in
the minds of students. There seems to be a pattern in the errors
committed by a majority of our students of both samples attributable to
the influence of approximate arithmetic. One way to explain this pattern
is by appealing to the concept of cognitive dissonance, propounded by
Festinger (1957). It is possible that Fequating small quantities to zero_ is
an almost unconscious reaction to the dissonance in the minds of the
students caused by the presence of such small numbers. We quote (from
Mason & Johnson-Wilder, 2004) the following passage: BWhen
dissonance is present, in addition to trying to reduce it, the person will
actively avoid situation[s] and information which would [be] likely [to]
increase the dissonance.^ (Festinger, 1957, p3.)
That several of the students during the interview failed to realise that
function 1010000 =ð1010000 x  1Þ is the same as the one given in question
2, suggests that the misconception arises because the student could be
working with more than one model of the real line: the model of
practical arithmetic embodied as a discrete set of pointsVsuch as, for
example, fn=108 j n 2 ZgVand the model of real number system as a
complete ordered field, represented geometrically by points of a straight
line. While working out a problem of computing limit, the concept image
of the model of approximate arithmetic is evoked, leading to errors. For
example, in response to question #5, one of the students writes:
1
limx!1 f ðxÞ ¼ 11 ¼ 1 (as 10 ¼ 1:)
limx!1þ f ðxÞ ¼ 1ð1 þ 1Þ2004 ¼ 1:
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 211

Note: As 102004 sums up to a very large quantity, it can be said to be


equal to infinity.
During interviews with students there was an almost uniform
10000
response of viewing the limit limx!0 1010 10000 x1 to be distinct from
question #2. They did not have any difficulty in evaluating this limit.
They probably thought that the 1 in the denominator is not a Fsmall_
quantity to be ignored.
We recall here the epistemological stance of constructivism in
mathematics education (von Glasersfeld 1983; 1987; 1991). According
to the constructivist model, mathematics learning is an inward and
personal construction process in which the learner seeks to assign
meaning to mathematical knowledge. It denies objectivity to mathe-
matical structures external to the learner and asserts that acquisition of
mathematical knowledge is a dynamic process. The acceptance of
such knowledge is based on constant interaction and negotiation
within the community. Thus understanding mathematical concepts and
developing internal representations of mathematical structures are
socio-cultural as much as cognitive activities. Hence, the internal
representations so acquired are open to the influence of the socio-cultural
environment of the learner (see Cobb, Yackel, & Wood, 1992, and the
references therein.) This is consistent with our findings regarding the in-
fluence of approximate arithmetic in students’ understanding of the
notion of limit. The students spend several years in school doing
arithmetic using decimal representations and finite approximations and
the influence of such classroom practices is seen in our study.

Informal Versus Formal Definition


In the understanding of the notion of limits one would expect that the
intuitive understanding of the limit concept the students learn through an
informal approach (such as using the graphical representation of
functions) will help in assimilating the formal definition. But, referring
to the students’ response to question #5, we surmise that the intuition
that was gained through the informal approach seems not to have been
put to good use after they have been exposed to the formal definition.
During the interview some of the students expressed the view that the
informal definition is unreliable and misleading. This could be a result of
the (well-meaning) emphasis on rigour and motivational examples on the
need for the    definition made by the instructor.
Our premise that learning the formal    definition would help
students stay clear of the pitfall of indiscriminate use of approximate
212 REVATHY PARAMESWARAN

arithmetic proved too simplistic. Indeed, we are in agreement with


Williams (2001) that the    definition of limit is meant to overcome
mathematical difficulties with the informal notion limit, and not to help
solve cognitive obstacles faced by students. When the student
encounters a situation involving very tiny numbers such as in our test,
possibly for the first time, he feels a cognitive dissonance. As he is
either unable or uncertain as to how to proceed further, what comes to
his rescue is not the mathematical    definition, but the tried and
tested approximate arithmetic practices which help simplify the
problem situation. So, the student reverts to his old practice of
approximating the given function, enabling him to complete the
calculation of limit, which results in error. This perhaps explains why
so many students from a prestigious undergraduate school as Chennai
Mathematical Institute also committed the same kind of errors as the
grade XII students.

Formal and Informal Aspects of Limits in Textbooks


It is typical of calculus books to motivate the notion of limits graphically.
First, examples are worked out by tabulating about a dozen values of a
function (such as a quadratic function, the square root function or a simple
trigonometric function) near the point where the limit needs to be

sinðxÞ
x
x
1:0 0:84147098
0:5 0:95885108
0:4 0:97354586
0:3 0:98506736
0:2 0:99334665
0:1 0:99933417
0:05 0:99958339
0:01 0:99998333
0:005 0:99999583
0:001 0:99999983

evaluated, convincing the students that values can be made as close to


the limit as possible by choosing the independent variable closer and
closer to the point where the limit is to be evaluated. In Example 6 of
Stewart’s (1987) Calculus, the limit limx!0 sinðxÞ
x ¼ 1 is worked out using
the following table and the graph drawn with the aid of the table.
Although it is very clearly mentioned that the table allows one to guess
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 213

the answer but not prove it, the students seem to develop the idea that
limit is no different from a process of approximation. When Fvery small_
numbers like 0:110000 are involved in the problem, this approach could
lead to the unintended impression that such quantities can be ignored.
In order to promote a flexible and robust view of functions in the
students, Carlson and Oehrtman (2005) recommend that early function
curriculum and instruction include more opportunities for students to
experience diverse function types emphasizing multiple representations
of the same functions. We suggest that one possible way to dispel
conceptual misunderstanding in teaching the limit concept is giving
different kinds of examples so chosen as to shed light on the limit
concept. Through this the teacher can hope to emphasise the subtleties
involved in the limit concept which might help to close up the gap
between their concept image and concept definition. Only further
research and experimentation can ascertain the effectiveness of such an
approach to introducing the concept of limits.

ACKNOWLEDGEMENTS

I thank Prof. V. Balaji, Chennai Mathematical Institute, for his valuable


advice and guidance. I am grateful to the referees of an earlier version of
my paper for insightful comments and suggestions. Their comments
resulted in thorough revision of the paper. I am particularly thankful to
Prof. David Tall for his help and encouragement.

APPENDIX

We shall now work out the solutions to questions in the tests.


1. (a) Since f ðxÞ ¼ x in the interval ð21000 ; 21000 Þ, we see that
limx!0 f ðxÞ ¼ limx!0 x ¼ 0:
(b) limx!21000 f ðxÞ ¼ limx!21000 x ¼ 21000 since limit is unaffected
by changing the value of the given function at a finite number of
points.
1 1000
2. limx!0 xð0:1Þ1000 ¼ 1=ð0:1Þ ¼ 101000 since the rational function
1
is discontinuous only when the denominator is zero, i.e.,
xð0:1Þ1000
when x ¼ ð0:1Þ1000 .
3. The relevant interval for computing the limit is ð1  ð0:1Þ1000 ; 1Þ.
On this interval, the function is linear. Therefore, limx!1 f ðxÞ ¼
102004 22004 ¼ 202004 .
214 REVATHY PARAMESWARAN

4. limx!0 gðxÞ ¼ limx!0 x ¼ 0.


limx!0þ gðxÞ ¼ limx!0þ 21000 ¼ 21000 as gðxÞ agrees with the
constant function 21000 in the interval ð0; 21000 Þ. Since limx!0
gðxÞ 6¼ limx!0þ gðxÞ, limx!0 gðxÞ does not exist.
5. The given graph represents the function
8
< x þ 0:4;
> x>1
f ðxÞ ¼ 1:2; x¼1
>
:
x; xG1

(i) We need to find a  such that f ðxÞ is between 1:2  0:25 ¼ 0:95
and 1:2 þ 0:25 ¼ 1:45. So, x must be greater than 0:95 and less
than 1:05. Therefore, take  ¼ 0:05: (Any  > 0 less than 0:05
will also do.)
(ii) From the graph it is clear that f ðxÞ G 1 for any x G 1 whereas
f ð1Þ ¼ 1:2. So if one takes any   0:2, one sees that for any
 > 0, f ð1Þ  f ð1  =2Þ > 1:2  1  .
(iii) Note that f ð0:9Þ ¼ 0:9 and one is required to find a  so that
the function takes values inside the interval ð0:9  0:25;
0:9 þ 0:25Þ over the interval ð0:9  ; 0:9 þ Þ. Note that
although f ð1Þ ¼ 1:2 G 1:15, for any x > 1, we have f ðxÞ >
1:4 > 1:15. Hence we must choose   1  :09 ¼ 0:1, as
otherwise there will always be an x > 1 in the interval
ð0:9  ; 0:9 þ Þ. Evidently  ¼ 0:1 works and so this the
largest possible value.

REFERENCES

Asiala, M., Cottrill, J., Dubinsky, E. & Schwingendorff, K.E. (1997). The development
of students’ graphical understanding of the derivative. Journal of Mathematical
Behavior, 16, 399Y 431.
Bachelard, G. (1938). La formation de l’esprit scientifique. Paris: J. Vrin (Reprinted
1980).
Barnard, A.D. & Tall, D.O. (1997). Cognitive units connections and mathematical proof.
Proceedings of the 21st PME Conference, Lahti, Finland, 2, 41Y 48.
Berry, J.S. & Nyman, M.A. (2003). Promoting students’ graphical understanding of
calculus. Journal of Mathematical Behavior, 22, 481Y 497.
Brousseau, G. (1983). Les obstacles épistémologiques et le problèmes en mathematiques.
Recherches en Didactique des Mathématiques, 4, 165Y198.
Carlson, M. & Oehrtman, M. (2005). Key aspects of knowing and learning the concept of
function, research sampler #9. Mathematics Association of America, http://www.maa.
org/t_and_l/sampler/research_sampler.html. Accessed 04 July 2006.
ON UNDERSTANDING THE NOTION OF LIMITS AND INFINITESIMAL QUANTITIES 215

Cobb, P., Yackel, E. & Wood, T. (1992). A constructivist alternative to the


representational view of mind in mathematics education. Journal for Research in
Mathematics Education, 23, 2Y33.
Cornu, B. (1980). Interference des modeles spontanes dan l’apprentissage de las notion
de limite. Seminaire de Recherche Pedagogique, 8.
Cornu, B. (1983). Quelques obstacles a l’apprentissage de la notion of de limite.
Recherches en Didactique des Mathématiques, 4, 236 Y268.
Cornu, B. (1991). Limits. In D.O. Tall (Ed.), Advanced mathematical thinking
(pp. 153Y166). Dordrecht: Kluwer.
Cottrill, J., Dubinsky, E., Nichols, D., Schwingendorf, K., Thomas, K. & Vidakovic, D.
(1996). Understanding the limit concept: Beginning with a coordinated process
schema. Journal of Mathematical Behavior, 15, 167Y192.
Davis, R.B. & Vinner, S. (1986). The notion of limit: Some seemingly unavoidable
misconception stages. Journal of Mathematical Behavior, 5, 281Y303.
Dreyfus, T. & Vinner, S. (1989). Images and definitions for the concept of function.
Journal for Research in Mathematics Education, 20, 356 Y366.
Ervynck, G. (1981). Conceptual difficulties for first year university students in aquisition
for the notion of limit of a function. In L.P. Mendoza & E.R. WIllams (Eds.),
Canadian Mathematics Education Study Group. Proceedings of the Annual Meeting,
Kingston, Ontario, Memorial University of Newfoundland, 330 Y333.
Ferrini-Mundy, J. & Graham, K.G. (1991). An overview of the calculus curriculum
reform effort: Issues for learning teaching and curriculum development. American
Mathematical Monthly, 98, 627Y635.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford: Stanford University Press.
Fischbein, E. (1993). The theory of figural concepts. Educational Studies in Mathemat-
ics, 24, 139Y162.
Glaser, B.G. & Strauss, A.L. (1967). The discovery of grounded theory: Strategies for
qualitative research. New York: Aldine de Gruyter.
Gray, E. & Tall, D.O. (1994). Duality, ambiguity, and flexibility: A proceptual view of
simple arithmetic. Journal for Research in Mathematics Education, 25, 116 Y140.
Heid, K.M. (1988). Resequencing skills and concepts in applied calculus using the
computer as a tool. Journal for Research in Mathematics Education, 19, 3Y25.
Jorgenson, D.L. (1989). Participant observation, a methodology for human studies.
London: Sage.
Koirala, H.P. (1997). Teaching of calculus for students’ conceptual understanding. The
Mathematics Educator, 2, 52Y62.
Li, L. & Tall, D.O. (1993). Constructing different concept images of sequences and
limits by programming. Proceedings of PME 17, Japan, 2, 41Y 48.
Mamona-Downs, J. (1990). Pupils’ interpretation of the limit concept: A comparison
between Greeks and English. Proceedings of the Fourteenth Conference for the
Psychology of Mathematics Education, Mexico City, 69Y76.
Mason, J. & Johnson-Wilder, S. (2004). Fundamental constructs in mathematics
education. London: Routledge Falmer.
Nardi, E. (1996). Novice mathematicians’ encounter with mathematical abstractions:
Tensions in concept-image construction and formalisations. PhD Dissertation, Oxford:
Oxford University.
Orton, A. (1983). Students’ understanding of integration. Educational Studies in
Mathematics, 14, 1Y18.
216 REVATHY PARAMESWARAN

Robert, A. (1982). L’Acquisition de la notion de convergence des suites numeriques dans


l’enseignement superieur. Recherches en Didactique des Mathématiques, 3, 307Y341.
Sierpinska, A. (1985). Obstacles epistemologiques relatifs a la notion de limite.
Recherches en Didactique des Mathématiques, 6, 5Y 67.
Sierpinska, A. (1987). Humanities students and epistemological obstacles related to
limits. Educational Studies in Mathematics, 18, 371Y397.
Sierpinska, A. (1990). Some remarks on understanding in mathematics. For the Learning
of Mathematics, 10, 24 Y36.
Stewart, J. (1987). Calculus (p. 63). California: Brooks/Cole Publ. Monterey.
Szydlik, J.E. (2000). Mathematical beliefs and conceptual understanding of the limit of a
function. Journal for Research in Mathematics Education, 31, 258Y276.
Tall, D.O. (1980a). Mathematical intuition with special reference to limiting processes.
Proceedings of the Fourth International Congress on Mathematics Education,
Berkeley, 170Y176.
Tall, D.O. (1980b). Intuitive infinitesimals in the calculus. Poster presented in
Proceedings of the Fourth International Congress on Mathematics Education,
Berkeley, C5.
Tall, D.O. (1980c). The notion of infinite measuring number and its relevance in the
intuition of infinity. Educational Studies in Mathematics, 11, 271Y284.
Tall, D.O. (1981). Intuitions of infinity. Mathematics in School, 10, 30Y33.
Tall, D.O. (1992). The transition to advanced mathematical thinking: Functions, limits,
infinity and proof. In D.A. Grouws (Ed.), Handbook of research on mathematics
teaching and learning (pp. 495Y511). New York: Macmillan.
Tall, D.O. & Vinner, S. (1981). Concept image and concept definition in mathematics,
with particular reference to limits and continuity. Educational Studies in Mathematics,
12, 151Y169.
Vinner, S. (1983). Concept definition, concept image and the notion of function.
International Journal of Mathematics Education in Science and Technology, 14,
293Y305.
von Glasersfeld, E. (1983). Learning as a constructive activity. Proceedings of the fifth
PME-NA, Montreal, vol. 1, 41Y69.
von Glasersfeld, E. (1987). Learning as a constructive activity. In C. Janvier (Ed.),
Problems of representation in the teaching and learning of mathematics (pp. 3Y17).
Hillsdale, NJ: Earlbaum.
von Glasersfeld, E. (1991). Abstraction, representation and reflection. In L. Steffe (Ed.),
Epistemological foundations of mathematical experience (pp. 45Y67). New York:
Springer.
Williams, S. (1991). Models of limit held by college students. Journal for Research in
Mathematics Education, 22, 219Y236.
Williams, S. (2001). Predications of the limit concept: An application of repertory grids.
Journal for Research in Mathematics Education, 32, 341Y367.

Graduate Student
Chennai Mathematical Institute,
SIPCOT IT Park, D-19 Padur P.O., Siruseri, Tamil Nadu 603103, India
E-mail: revathy@cmi.ac.in

You might also like