MODELING THE DEVELOPMENT OF THE CONCEPT OF FUNCTION[1]

 

Mindy Kalchman

OISE/University of Toronto

mkalchman@oise.utoronto.ca

Robbie Case[2]

OISE/University of Toronto

 

As children move from a process-based to an object-based understanding of mathematical functions, we suggest they pass through three phases of processing: (1) procedural; (2) interval; and (3) object-based. Analysis of a microgenetic instructional study showed that each of these processes is developed through a particular stage in an experimental curriculum. We present here the (i) psychological processes that characterize each phase of processing, (ii) the component of the curriculum that supports each phase, and (iii) examples of students’ reasoning within each phase of processing. We suggest that standard curricular approaches to the teaching of functions do not adequately address the many psychological connections children must make in understanding functions.

 

            Much has been written about the importance of the process to object conceptual shift for developing a deep understanding of mathematical functions (e.g. Sfard, 1992). However, analyses that characterize students’ psychological processes when going from a process- to object-based understanding are lacking. In this paper, we present the results of such an analysis.

            We consider a process understanding to be one where students use a procedural or computational (Sfard, 1992) approach for deriving pairs of values, which are then used to represent a function in a tabular and graphical way; and we regard an object-based understanding as one where a function is conceptualized as something to which processes may be applied (Harel & Dubinsky, 1992). Three phases of psychological processing have emerged from our analyses of children going from a process to object understanding of function: (1) procedural; (2) interval; and (3) object-based. Because we see a circular relationship between structural modeling and curricular design (Kalchman, Moss & Case, in press), we derive our cognitive models from the analysis of instructional studies (Kalchman & Case, 1998, 1999).

The Study

Three sixth-grade female students attending a university’s laboratory school participated in the study. They spent nine 45-minute sessions engaged in an experimental instructional unit on functions. The first five days were spent in a classroom setting; the next four were spent using spreadsheet technology; and the final day was spent meeting on a one-to-one basis. On days 1, 2, 5, and 9, students were asked the same set of items either at the beginning or the end of the session. This repeated questioning was to identify at which points in the curriculum students made conceptual gains. Students worked at their own pace on the computer for days 6, 7, and 8.

Phases of Processing: Curriculum, Phase Attributes, and Students’ Thinking

Table 1 summarizes the central curricular features that support each phase of processing, the cognitive attributes that characterize children’s thinking in each phase, and examples of how children conceptualized a particular item while working within each level of understanding. The item analyzed was one where students were shown a graph of the function y = x + 7, seen only in the upper right quadrant of the Cartesian grid. Both axes were labeled from 0 to 10, with the points (0, 7), (1, 8), and (2, 9) plotted within the line. Students were asked to give an equation for a function that would pass through the linear function seen, and then to explain their reasoning.

The Procedural Phase

            In the first phase of processing we identified, children go from no understanding of what a function is to a procedural understanding. They become able to generate a table of numbers and a graph for any function of the form y = f(x) from x = 0 to x = n (where n is a positive integer).         We support the development of this understanding by beginning instruction with the context of a walkathon, where a specified amount of money earned per kilometer walked is symbolized, calculated, and then graphed. For example, if the sponsorship arrangement is earning $4.00 per kilometer walked, students first construct an algebraic symbolic representation for the rule, e.g., $ = km * 4. They then calculate the amount of money earned at each kilometer walked by multiplying each kilometer by 4, e.g., 0 km = $0.00, 1 km = $4.00, 2 km = $8.00, etc. They then represent the results of these calculations in a tabular fashion. When the table has been completed for 10 km walked, students place markers at each coordinate point on a Cartesian


Table 1. Phases of processing, their attributes, and students’ thinking.

Phase of Processing

Central Curricular Features

Psychological Attributes

Example of Students’ Thinking

 

Procedural

 

 

Walkathon -- calculation of dollars earned per kilometer walked; creating a table of values; and plotting discrete points on a grid

 

 

Mastery of calculating numerical values for the dependent variable given a set of values for the independent variable and arranging them in a tabular form

 

Mastery of plotting pairs of x and y coordinates that correspond to the pairs of values found in the above table

 

Recognition that each pair of numerical values is associated with a unique point on the graph and vice-versa

 

Mastery of heuristics for moving from a set of points found in a tabular form or graph to the rule that generated them

 

“Let's say someone paid you $9.00 for every kilometer you walked. So, I was thinking the equation could be x times 9 equals money. At 0 kilometers you have $0 and then [when] x is 1 … 1 times 9 is 9 so y = $9, which has already crossed."

 

 

 

Interval

 

 

Calculation of numeric intervals between values of the dependent variable in both tabular and graphic representations

 

 

Recognizing particular properties of intervals between successive y values as a constant increasing value for functions such as y = 4x (with 4 being that constant value)

 

Recognizing that constant intervals between y values on a graph always result in straight lines

 

Connecting a concrete mathematical rule, the pattern in the sequence of numbers it generates (e.g.., a constant interval), and the pattern seen in the graph (e.g., a straight-line)

 

Repeating above for properties such as y-intercept, non-linearity, and the negative or positive slope or “direction” of any function

 

“I started off with 3, … and each time I’m going to move up 6.  So, it’s going to be a straight line because it’s going up by the same amount and it’s going to be steep because it has a slope of 6 and it does pass through (the existing line) The equation is y = x times 6 + 3.

 

 

Object-Based

 

 

Computer spreadsheet activities where students operate on y = x or y = x2

 

use of the graphic representation of y = x and the y = x2 as mental referents, or concrete objects, for abstracting properties and features of a function and judging the “steepness” of a line, degree of a curve, “directionality”, and y-intercept

 

 

y = x times 5. It goes up by the same amount and it’s a straight line and it’s steep enough because it goes up by 5 each time. You could also do other steep ones but I just did this one.”

 


grid. The markers are then joined to form a line. Several different rules of sponsorship are explored in this way, including those that produce curved lines (e.g., the amount of money earned is equal to the number of kilometers walked times itself [$ = km2]), and those that involve initial "starting amounts," which correspond to the y - intercept (e.g., one is given a starting bonus of $5.00 and still earns $4.00 per kilometer [$ = 5 + 4 * km]).

 The psychological attributes of this phase are summarized in the middle column of the first row of Table 1. In this phase, students work from a procedural standpoint. That is, they perform calculations on the independent variable in order to obtain values for the dependent variable and then organize their results in first a tabular and then graphic way. General properties of these representations are not abstracted, but the link between the coordinate pairs found on the graph and the pairs of values recorded in the table is recognized and understood to have been generated from a specific rule such as multiplying each x by 4.

When asked to provide an equation for a function that would pass through the one described above, prior to instruction students made a series of dots at the coordinates (0,9), (1,8), (2,7), and (3,6) but did not express the meaning of these dots as any sort of functional relationship. Then, on day 2, CP explained: "Let's say someone paid you $9.00 for every kilometer you walked. So, … the equation could be x times 9 equals money. At 0 kilometers you have $0 and then [when] x is 1…1 times 9 is 9 so y equals $9, which has already crossed." With each calculation for the dependent variable, CP plotted the coordinates on the grid.

The Interval Phase

The second phase of processing is one where students develop an interval understanding of a function by noticing the second-order features found in the tables and graphs. These second order features include the value of the numeric and coordinate jumps found between successive y values. The second row of Table 1 describes the particulars of this phase.


We promote an interval understanding of function by having students investigate what generally determines the linearity, degree of “steepness”, and y – intercept of a function. For

example, in the tabular expression of the function $ = km * 4, students add a column to the right of the $ column and write in the difference between successive values, which in this case is consistently 4. Students also graph functions by sketching the graph in intervals by moving their pencils “up by” for example 4 as they move over one unit, rather than as discrete points. Intervals are also calculated for non-linear functions, and students realize that the degree of the “steepness” of a curve is related to the span of the intervals. Students eventually generalize that straight-line functions always have constant intervals between y values, and curved lines do not.

      By noting second-order properties found in the tables and graphs of particular functions (i.e., numeric or graphic jumps or intervals), students abstract general features of functions such as (a) linearity, (b) specific slopes, (c) the degree of an exponent, and (c) the y-intercept.

In response to the item, on day 2, HJ explained that “if your starter offer was 1 and your next point was 6, so you go up by 5…then your next one is 10 so each time you go up by 5.” Her interval understanding appears incomplete at this point, however, because she did not include the $1.00 starting value in the equation or mental computation. This omission may have been due to a still emerging understanding of how to connect the mathematical rule, the pattern in the sequence of numbers it generates, and the pattern seen in the graph. On day 5, however, HJ seemed to make these connections: “I started off with 3…and each time I’m going to move up 6.  So, it’s going to be a straight line because it’s going up by the same amount and it’s going to be steep because it has a slope of 6 and it does pass through. The equation is y = x times 6 + 3.

            Object-Based Phase

In this phase, students construct and abstract third-order properties such as slope and y-intercept from the second-order properties already noted. The curricular feature that supports this phase is students’ use of spreadsheet technology. Students work with a pre-configured computer screen, which displays the graph of y = x (or y = x2) on the left-hand side, and a spreadsheet with the corresponding table of values (found in columns X and Y) on the right. A console displays the equation of the function. On the spreadsheet, students are asked to change individual parameters of the function (slope or y-intercept) in order to move y = x through randomly placed pre-plotted points. All actions carried out are reflected instantly and automatically in the graph and in the numeric pattern found in the Y column. Then, using columns to the right of Y, students program their own functions with given specifications such as a slope greater than that of the original line (y = x) and a y-intercept < 0. Students also carry out these sorts of activities for curved lines.

Using y = x and y = x2 as concrete mental objects, students are able to operate on the functions as entities in order to produce new functions that have, for example, constant intervals greater than 1, and thus, are steep relative to y = x. Others with constant fractional intervals are relatively flat. Students may operate on y = x2 to produce functions whose curves are steeper than y = x2 by making the exponent larger than 2, the coefficient greater than 1, or some combination of the two. Likewise, for functions that curve down, the degree of “steepness” is still described by the exponent and the coefficient, and the “down by” is defined by the xn being multiplied by a negative coefficient. With respect to the y – intercept, students see the addition or subtraction of a constant value as a means to qualitatively and quantitatively translate the base function.

On the final day, students seemed to be using the given line as a mental referent for determining a suitable equation for a function that would pass through the one given. HJ drew a straight line starting from the origin and passing through (1,5) and (2, 10). She explained: “y = x times 5. It goes up by the same amount and it’s a straight line and it’s steep enough because it goes up by 5 each time. You could also do other steep ones but I just did this one.”

Conclusions

We attribute students’ success with this difficult topic to their engagement with the innovative curriculum, which allowed them to work within the walkathon context to master the necessary sorts of procedures. First, they calculated values for the dependent variable from values for the independent variable. Then they calculated the intervals between successive values for the dependent variable – a calculation that essentially defines the general shape (i.e., linear or curved) of a function. As students moved among tables, graphs, and symbolic and verbal rules, they recognized and generalized certain important features of a function such as y-intercept, degree of “steepness” and linearity. We think that experts in mathematics underestimate the time it takes to make the many psychological connections proposed here and neglect to provide an adequate body of examples necessary for abstracting the entailments of particular functions.

References

 

Harel, G. & Dubinsky, E. (Eds.) (1992). The concept of function: Aspects of epistemology and       pedagogy. West laFayette, IN: Mathematical Association of America.

Kalchman, M. & Case, R. (1998).Teaching mathematical functions in primary and middle school: An approach based on Neo-Piagetian theory. Scientia Pedagogica Experimentalis, 35(1), 7-53.

Kalchman, M. & Case, R. (1999). Diversifying the curriculum in a mathematics classroom streamed for high-ability learners: A necessity unassumed. School Science and Mathematics, 99(6), 320-329.

Kalchman, M., Moss, J. & Case, R. (in press). Psychological Models for the Development of Mathematical Understanding: Rational Numbers and Functions. To appear in S. Carver & D. Klahr (Eds.), Cognition and instruction: 25 years of progress. Mahweh, NJ: LEA.

Sfard, A. (1992). Operational origins of mathematical origins and the quandary of reification: The case of function. In G. Harel & E. Dubinsky (Eds.), The concept of function: Aspects of epistemology and pedagogy. West laFayette, IN: Mathematical Association of America, 59-84.



[1] This work was made possible through the generous support of the James. S. McDonnell Foundation and the Social Sciences and Humanities Research Council of Canada. Many thanks to Karen Fuson for her support and comments.

[2] Robbie Case passed away while this work was in progress.