A Generative Model of Mathematics Learning

Author(s): M. C. Wittrock
Source: Journal for Research in Mathematics Education, Vol. 5, No. 4 (Nov., 1974), pp. 181-196
Published by: National Council of Teachers of Mathematics
Stable URL: http://www.jstor.org/stable/748845
Accessed: 13-03-2016 00:56 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/748845?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

National Council of Teachers of Mathematics is collaborating with JSTOR to digitize, preserve and extend access to Journal
for Research in Mathematics Education.

http://www.jstor.org

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 A GENERATIVE MODEL

OF MATHEMATICS LEARNING

M. C. WITTROCK

University of California-Los Angeles

Introduction

An important, welcome evolution from behaviorism to cognitivism is

occurring in the study of human learning and retention. As a result of

recent research, our understanding of the processes involved in cognition

and in the learning of mathematics is increasing.

As a way to organize some of this research and to focus interest on its

meaning for mathematics education, this paper will begin with a hypothesis

about human learning that has been developed from the investigator's re-

search in cognition, discovery learning, and instruction in schools. It will

then present a sample of the empirical studies that led to the generation of

the hypothesis. Finally, some of the meaning of this research for mathe-

matics learning will be discussed.

The first point to emerge will be that we can be proud of the research

in the learning of mathematics, including the development of curricular

materials. The second point will be a recommendation that research in

mathematics learning should increasingly be devoted to studying the step-

by-step specific and higher-order intellectual processes that students engage

in when they learn mathematics; such as when they are adding, subtracting,

differentiating, and integrating.

The hypothesis and empirical studies presented in this paper focus on

the cognitive, generative processes that are involved in the learning of

mathematics. These processes could perhaps be presented in simpler S-R

terminology. The cognitive model emphasizes the learner's active, step-

by-step processing of information, and is more compatible with the

author's point of view.

The data to be discussed will probably arouse recollections of the

This is a slightly edited version of a paper commissioned by the ERIC Information

Analysis Center for Science, Mathematics, and Environmental Education and presented

at a session of the Special Interest Group for Research in Mathematics Education. The

session, held 28 February 1973, was in conjunction with the AERA Annual Meeting.

The paper was prepared pursuant to a contract with the Office of Education, U.S.

Department of Health, Education, and Welfare. Contractors undertaking such projects

under government sponsorship are encouraged to express freely their judgment in pro-

fessional and technical matters. Points of view or opinions do not, therefore, necessarily

represent official Office of Education position or policy.

November 1974 181

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
dilemmas about nativism and the role of abstractions in memory, which

Plato and Aristotle wrote about. In the Meno, Socrates taught the slave boy

to prove that the diagonal of a square is equal to the side of a square twice

the area of the given square. To Plato, it seemed that the abstractions were

inherited and primary.

Aristotle circumvented Plato's stress on nativism and the primacy of

abstractions by reducing the abstractions to the commonalities across par-

ticulars. He made sensory data and the particulars the focus of human

learning. The hypothesis and data presented here may appear to raise

again this ancient dilemma. The reader is left to decide whether or not

that is the case.

Learning as a generative process

Succinctly, but abstractly stated, the hypothesis is that human learning

with understanding is a generative process involving the construction of

(a) organizational structures for storing and retrieving information, and

(b) processes for relating new information to the stored information.

Stated more directly, all learning that involves understanding is discovery

learning. We can determine the effects of instruction in terms of what the

instruction causes the learner to do. Effective instruction causes the learner

to generate a relationship between new information and previous ex-

perience.

Further, one should not construe learning, even so-called reception

learning, as a passive reception of someone else's organizations and abstrac-

tions. It is better to look beyond the nominal stimuli to the functional stimuli

they become for the learner. In the past, the importance of the environ-

ment and instruction on the learner has been emphasized. New stress must

be placed on the active role of the individual in learning.

From this point of view, there is no one best method of teaching all

students, although there may be one best logical organization of the sub-

ject matter. It is possible for reception-learning treatments to lead to

discovery learning, contingent on the teacher's ability to build on the

learner's previous knowledge and upon what the instruction causes the

learner to do. A variety of teaching methods will be needed by each

student, depending on his background and its relationship to the subject

matter he is taught.

In sum, learning with understanding can occur with discovery treat-

ments or with reception treatments. The important point is what these

treatments cause the learner to do.

One way to summarize the above conception is as follows. Although a

student may not understand sentences spoken to him by his teacher, it is

highly likely that a student understands sentences that he generates himself.

In this context, generation and understanding are closely related; possibly

one causes the other, or perhaps the terms are synonymous.

182 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
Demonstrations of learning as a generative process

The above argument may be elaborated by presenting two brief demon-

strations, one perceptual and one verbal. For the first demonstration, look

at the black irregularly shaped objects in Figure 1. What does one see?

AA V IA

Fig. 1. Look at the black objects above. What are they? (adapted from W. Brown

and H. Gilhousen, College Psychology, 1950, courtesy Prentice-Hall.)

One probably sees only black irregularly shaped objects because previous

experience has trained one to treat black as foreground on white pages,

and also because one was instructed to process the information as black

irregularly shaped objects.

Now process the information in Figure 1 differently. Process the black

as background and the white as foreground. One will see the word PLAY

in capital letters. The point here is that the way in which one processes

structurally organized information is crucial to the meaning derived from

it. The reader might reflect on this example the next time his students

don't get the point. They may be generating black objects instead of words.

For a demonstration from the verbal area, read the paragraph in Fig-

ure 2 and generate a title for it. The words seem to apply to many tasks,

The procedure is actually quite simple. First you arrange things into differ-

ent groups depending on their makeup. Of course, one pile may be sufficient

depending on how much there is to do. If you have to go somewhere else due

to lack of facilities that is the next step, otherwise you are pretty well set. It

is important not to overdo any particular endeavor. That is, it is better to do

too few things at once than too many. In the short run this may not seem

important, but complications from doing too many can easily arise. A mistake

can be expensive as well. The manipulation of the appropriate mechanisms

should be self-explanatory, and we need not dwell on it here. At first the

whole procedure will seem complicated. Soon, however, it will become just

another facet of life. It is difficult to foresee any end to the necessity for this

task in the immediate future, but then one never can tell. [Bransford &

Johnson, 1972.]

Fig. 2

November 1974 183

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 but identifying a specific title is difficult for most readers. The paragraph

seems very vague-until one is told that the title of the passage is "Washing

Clothes." The title provides the cue that allows one to relate the vague

terms contained in the passage to one's knowledge of washing clothes.

Now we will turn to the research literature on generative structures and

processes.

Recent Research

The recent research is divided into two sections: (a) structural organiza-

tion and (b) processing and coding of information. Representative studies

only will be presented and no attempt will be made to be exhaustive in

citing relevant research in the field.

Structural organization

The example given above, which dealt with the washing of clothes, was

taken from a study reported by Bransford and Johnson in the December,

1972 issue of the Journal of Verbal Learning and Verbal Behavior. They

found (p. 723) that giving the title for the story before the story was read

greatly enhanced its comprehension and recall, while giving the title after

the story was read did not increase comprehension or retention. The

structural organization in this vague story is too weak to allow much

generative processing of its information.

The second research study to be cited is by Bower and Winzenz (1969).

They presented students with strings of digits such as the following:

First String (17) (683) (9452) (7) (56)

Second String (176) (8) (394) (5275) (6)

The digits of each string and their order remained the same across the

strings, while the grouping changed from the first to the second string.

The results showed that altering the organization of the string did not

improve learning across altered strings, although the digits and their order

of presentation remained the same. Apparently, each string was learned as

though it were a completely new one. By changing the structural organiza-

tion, it seems that a new task is presented to the learner. In fact, proactive

interference could have been predicted in this task, based on the similarity

of the digits across the strings. Because proactive interference among digits

did not occur, it could mean that the groups of digits are treated as the

units to be learned. Concepts of the units learned and stored need to be

examined. These units may be larger than have been assumed.

The third empirical study on structural organization (Marks, Doctorow,

& Wittrock, 1974) involves the teaching of reading to public school

children. Two hundred twenty-two students were randomly assigned to two

conditions that varied the meaningfulness of 20% of the words in a com-

merically published story designed to teach reading. From the generative

184 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
hypothesis presented above, it was predicted that the semantic properties

of the words would have decided effects on story comprehension and reten-

tion. The reason for the prediction is related to an all-or-none hypothesis.

By increasing word meaning, the organization of the narrative should be

discovered on an all-or-none basis. As you can see from Table 1, story

TABLE 1

MEANS AND STANDARD DEVIATIONS OF THE PERCENTAGE OF ITEMS CORRECT

ON THE POSTTESTS OF READING COMPREHENSION AND RETENTION

Test of Test of

Comprehension at Comprehension above Cloze

Subjects' Reading Level Subjects' Reading Level Retention Test

Reading

Level High Low High Low High Low

Frequency Frequency Frequency Frequency Frequency Frequency

Treatment Treatment Treatment Treatment Treatment Treatment

High 73 48 67 47 58 36

(14) (19) (14) (17) (15) (16)

Middle 74 47 57 29 53 31

(14) (16) (14) (13) (16) (15)

Low 71 45 61 36 55 20

(14) (22) (14) (15) (15) (16)

51 29

(14) (14)

comprehension and retention were nearly doubled by changing the mean-

ing of 20% of the words in the story from less-frequent ones (e.g., lad)

to more-frequent ones (e.g., boy). The use of unfamiliar terms can

severely retard comprehension of the structural organization of a story.

One implication is that a few unfamiliar terms may have a similar decided

effect on mathematics learning.

The fourth study (Wittrock, Doctorow, & Marks, forthcoming) was

designed to test further the generative conception of learning presented

above by using a familiar organization (a familiar story) to teach the

meaning of new words to 482 elementary school children. Again it was

reasoned that a familiar structural organization enables a child to

"discover" the meanings of new words without defining them for him.

From Table 2 one can see the results from this research study, using

commonly available teaching materials. All children read the same stories

twice; however, the high-meaningful group read a familiar version first,

one that used highly meaningful words, and then read the same story but

with many new unfamiliar synonyms that did not change the meaning of

the story. The control group read the version with the low-meaningful

words twice. On the test of definitions of unfamiliar words, the group that

read the familiar version of the story first did much better than the control

group.

No words were ever defined for either group. Neither were any of the

synonyms temporally associated with each other. Instead, a familiar struc-

November 1974 185

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 TABLE 2

USING WORD MEANING TO IMPROVE

READING AND LISTENING COMPREHENSION

Mean Scores for the Four Experimental Groups

Reading Listening

Reading

Levels Tests 1) High Freq. 2) Low Freq. 3) High Freq. 4) Low Freq.

Comprehension 7.0 4.2 8.1 3.9

Low Vocabulary 7.3 5.3 7.8 5.7

Retention 19.2 8.9 20.2 9.0

Comprehension 6.9 4.0 7.9 4.2

Middle Vocabulary 10.9 7.0 10.2 7.1

Retention 45.5 25.2 52.4 24.5

Comprehension 7.3 4.8 7.8 4.7

High Vocabulary 12.7 10.5 13.5 10.1

Retention 53.6 26.6 61.2 26.4

tural organization, a story, was introduced and used as a meaningful con-

text to teach definitions of new words in a relatively painless, efficient way.

The idea of a familiar organization can also be used to teach the definitions

of unfamiliar mathematical terms.

It is proper to end this section on structural organization by mentioning

the research of David Ausubel, who pioneered in this area with his books

and articles on meaningful verbal learning. Ausubel's studies on learning

Buddhism as a new organization that can be imposed on one's previously

learned religion continue to stir interest in structural organizations.

From an S-R perspective, Benton Underwood (1957) also fostered

work in this area with his classical research on the importance of proactive

processes in learning and retention. Either associationistic or cognitive

approaches are capable of handling the data of these studies. In fact, when

these two approaches were put to an explicit test in a study dealing with

the retention of sentences (Anderson & Bower, 1971), the associationistic

model was supported.

The controversy that Plato and Aristotle began over the roles of ab-

stractions and particulars in learning has not yet been resolved. No end

to it is in sight.

Processing of information

Under this heading are the research studies in which learners generate

groups or other large units, such as stories, sentences, and images, from the

stimuli they have been given. The results of these studies occasionally are

dramatic.

For example, Bower and Clark (1969) gave subjects 12 different lists

of 10 unrelated nouns. One group of subjects was asked to learn each list

in whatever manner each person wished, keeping the order of the words

the same. The experimental group was asked to make a meaningful story

from each list. The results need little elaboration. Statistical tests are not

186 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
necessary either, because the mean gain from the control to the experi-

mental group was 79%-from 14% to 93%! The generation of a story,

a thematic organization, greatly reduced interference among the 12 lists

and facilitated the retention of the words and their serial order.

A number of researchers (Rohwer, 1973; Anderson, 1970; & Bobrow,

1970) have found that processing words into a sentence also facilitates

their retention. The facilitating effect of processing words into sentences

will not be discussed any further here, because of the apparent similarity

of this type of processing to the generation of stories.

The next type of generative processing is imagery, which Alan Paivio

(1971) and others have recently studied extensively. Bull and Wittrock

(1973) have completed a study on the teaching of word definitions to

90 fifth graders, using three learning conditions. The instructions for the

groups differed as follows: (a) generate (draw) an image of the word and

its definition; (b) trace a picture (image given) representing the word

and its definition; or (c) learn the verbal definition by copying it. The

group means for retention of the definition ranked from high to low in

the order given above. The group that generated their own images re-

membered the definitions best one week later. There was no statistically

significant difference between the other two groups. Again, the generative

processing hypothesis fits the data. It does not seem that the mode of

generation (story, sentence, or imagery) regularly discriminates among

performance. What does seem to be important is that some type of gen-

eration be called on.

Processes and structures

To test the latter possibility, Wittrock and Goldberg (in press) com-

bined into one study the above three different types of generative processes

commonly studied in the literature: (a) imagery, (b) stories, and (c)

sentences, in addition to a control group. College subjects were randomly

assigned to these four conditions and given lists of words to remember.

The words varied in meaningfulness and in imagery value. The results are

presented in Figure 3. No important mean differences were found among

the three generative processes, although the control group did less well

than the other three groups. However, we did find that the imagery and

meaningfulness values of the words used in the study influenced retention.

The history of associations to words is an important factor in determining

their effect on retention.

Wittrock and Goldberg (in press) have replicated this study with

public school children. They found that verbal processing conditions, sen-

tences or stories, do facilitate retention in children, when compared with

the imagery conditions. Perhaps, with children, the nonverbal processes

need to be primed more than the verbal ones. This issue is still being

November 1974 187

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 6

IMM 5.29

5.03

4.86

S IHML 5.10

S4 . 5 1 4.72 4.76 4.51

TIMI 4.60

IMI 46 4.07 4.26

4 - IM 3.81

83.81

3.44 3.78

Imagery Story Sentence Control

Instructional Sets

Fig. 3. Results from Wittrock and Goldberg Study

investigated. The point is that generative processing of information has

effects on retention.

A clear picture of how different organizational structures and different

processes of organizing information fit together is provided in a study by

Wittrock and Carter (forthcoming), which used materials adapted from

a study by Bower, Clark, Lesgold, and Winzenz (1969). Figure 4 presents a

ALUMINUM

GRANITE MARBLE COMMON GASOLINE

MINERALS KEROSINE CHEMICALS LIMESTONE COPPER FERTILIZERS BRASS STONES FUELS

NON-PETROLEUM TOPAZ MERCURY IRON ZINC SILVER METALS SULPHUR PLATINUM

STEEL ALLOYS LEAD PEAT PHOSPHATE EMERALD SAPHIRE SOLDER COAL

PRECIOUS SANDSTONE GOLD POTASH NITRATE PETROLEUM ARSENIC URANIUM MASONRY

DIAMOND RARE SLATE IODINE MEDICINAL

RUBY COKE BRONZE OIL PEWTER

Fig. 4. Randomly Ordered Hierarchy (Wittrock and Carter Study)

188 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 randomly ordered hierarchy for the concept of minerals. This hierarchy

can be rearranged into a properly ordered one, with the word "minerals"

at the top. In addition to the random hierarchy, a proper hierarchy

was developed as shown in Figure 5. Last, an unordered hierarchy

MINERALS

METALS STONES CHEMICALS FUELS

RARE COMMON ALLOYS PRECIOUS MASONRY FERTILIZERS MEDICINAL PETROLEUM PETROLEUM

PLATINUM ALUMINUM BRONZE SAPPHIRE LIMESTONE NITRATE IODINE OIL COAL

SILVER COPPER STEEL EMERALD GRANITE POTASH ARSENIC GASOLINE COKE

GOLD LEAD BRASS DIAMOND MARBLE PHOSPHATE SULPHUR KEROSINE PEAT

URANIUM IRON PEWTER RUBY SLATE

MERCURY ZINC SOLDER TOPAZ SANDSTONE

Fig. 5. Proper Hierarchy (Wittrock and Carter Study)

was prepared (see Figure 6) by putting conceptually unrelated words

into a hierarchy. These words were comparable in frequency values to the

words in the other two hierarchies.

GRASSHOPPER

CHEESE ZERO KERCHIEF SHIVER

MAYOR RANDOM FRIGATE MITT NOCTURNAL REGISTER ABDOMEN FEEBLE ADMONITION

POIGNANT ENAMEL CITADEL LITTLE METAPHOR MANUFACTURER INDUSTRY SCISSORS DESERT

CLUB CITY OFFICE MOUSE BIRCH LECTURE IMPRINT BLADE PUNGENT

SAUCER GENUINE GROUND SANCTUARY MAJORITY UNORGANIZED SLEDGE PHEASANT PULLEY

CABINET SCHOOL SEASON TECHNICIAN PERFORMER

ANNUAL SETTLEMENT PEST PERENNIAL PORTION

Fig. 6. Unordered Hierarchy (Wittrock and Carter Study)

November 1974 189

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 We gave these hierarchies to college subjects under two different proc-

essing conditions. In the first, or control condition, the learners were asked

to copy the hierarchies. The second, or generative processing condition,

emphasized rearranging the hierarchies until an order was found that

"made sense."

The interesting results are found in Figure 7. With the rote processing

indicated in the bottom line, increasing the structural organization increases

22 21.9 Generative
21.4 Processing

20-

S18-

16.9 Reproductive

16 C Processing

14

13.1

12 -

11.2

E 10

S8

6 6.1

4U

Unrelated Random Proper

Hierarchical Organization

Fig. 7. Results from Wittrock and Carter Study (Forthcoming)

retention in nearly a linear fashion. However, the generative processing

instructions had sizable effects on retention at all levels. Generative proc-

essing had the most marked effect when the organization was in the learn-

ers' repertoires, but not made explicit, as in the randomly ordered hier-

archy (Figure 4). Because generative processing facilitated retention at all

three levels of structural organization, it makes sense to separate these

two types of factors as has been done in the outline of this paper. This

separation helps to understand sometimes conflicting results on discovery

learning.

The data in Figure 7 summarize an important relationship between

structural (content) organization and generative, processing of informa-

tion. It was this study that led to the conclusion that learning with

understanding is a generative process.

190 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 The above line of research gives a new understanding of what is involved

in meaningful learning. Meaningful learning requires a structural or content

organization and enough relevant background knowledge to enable one

to discover the new organization when it is not made explicit. Whether

or not the organization of the content is made explicit does not seem to

be crucial. It is more important that the learner generatively process any

kind of organization he is given. The combination of appropriate back-

ground and generative processing of the new information should facilitate

his understanding of the organization presented to him.

Individual differences

Interest in the highly important study of aptitude-treatment-interactions

(ATI) and individual differences in processing variables has grown. One

way to develop ATI research is to study the relevant processes that the

learner engages in when he learns a given subject matter. Hypotheses about

these processes and the development of protocols and specific tests for

them are needed.

Suppes (1973) and several authors in the JOURNAL FOR RESEARCH IN

MATHEMATICS EDUCATION have studied the step-by-step processes of

learners engaged in mathematical tasks, such as addition and subtraction.

Also, several interesting ATI studies on learners' abilities and verbal proc-

esses (summarized by Aiken, 1971) were noted in the JOURNAL FOR

RESEARCH IN MATHEMATICS EDUCATION. Studies on these processes have

great promise.

Brain research

Another important line of research is relevant to describing the process

variables and individual differences involved in cognition and in the learn-

ing of mathematics. Recent research on the human brain indicates there

is a differentiation of functions between its right and left hemispheres that

may tell something about how right or left hemispheres process informa-

tion differently from each other. This research may also tell how spatial

and verbal processes are involved in learning.

Figure 8 (based on Bogen, 1969b, p. 150) summarizes some of the

educationally relevant functions and processes that seem to be lateralized

in the brain, especially after the age of five years. Some of the categories

within a column are repeated because of the overlapping comparisons made

in the different studies whose findings are the basis for Figure 8.

From such findings it seems that the brain has at least two modes of

learning, remembering and solving problems. If we have multiple modes of

learning, this valuable information should be used in the individualization

of instruction. Another possible meaning of the above research is that

November 1974 191

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 Left Hemisphere Right Hemisphere

propositional thought appositional thought

language spatial relations

verbal perceptual

symbolic visiospatial

temporal processing part-whole processing

(Gestalt perception)

logical or analytic analogic or relational

propositional thought visual imagery

linear nonlinear

Fig. 8. Functions That Seem to Be Lateralized in the Brain

instruction should be introduced in the dominant or preferred mode and

elaborated in the second mode. Again, research on the processing of

information is a fruitful area to investigate.

Higher-order processes

If it is possible to isolate higher-order processes, used in learning or in

problem-solving strategies, can we teach students to improve their uses of

them? The studies on imagery and verbal elaboration indicate that it is

possible. The answer is also yes if one accepts the following data.

In 1967, a study was reported in the Journal of Educational Psychology

on the teaching of replacement strategies and nonreplacement strategies to

young children (Wittrock, 1967). Physical objects were used to represent

the hypothesis to be tested. In the nonreplacement strategy each card

representing a hypothesis was turned facedown when it was eliminated

from the set of testable hypotheses. In that study, the children learned and

transferred simple problem-solving strategies to new problems, including

ones where the cards and props no longer appeared.

Higher-order strategies can be taught to children, in the JOURNAL FOR

RESEARCH IN MATHEMATICS EDUCATION several studies on the teaching

of strategies appear. This is an important area to pursue.

Motivation

Recent research in cognition also provides a new perspective regarding

how reinforcement may operate on students. Two areas of research will be

mentioned here: (a) achievement motivation and (b) the delay-retention

effect.

Weiner, Heckenhausen, Meyer, and Cook (1972) and Weiner (1972)

indicate that the effects of reinforcement depend on which of the learner's

attributes, e.g., his effort or his innate intellectual ability, he believes is

being reinforced. If the student infers that his effort at learning mathe-

matics is responsible for his success, he is more likely to persevere with

similar mathematical problems than if he infers that his success or failure

192 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
is attributable to factors over which he has no control, such as his innate

ability or lack of it. Weiner's cognitive interpretation of reinforcement has

practical utility for teachers of mathematics. A student may erroneously

assume his failure to learn is due to factors beyond his control, when addi-

tional effort would solve the mathematics problem and increase motiva-

tion to learn more advanced mathematics.

Delay-retention effect

For many years, it has been believed that reinforcers should be given

(a) immediately, (b) discriminately, and (c) frequently, during acquisi-

tion of behavior. Teacher-education students studying educational psy-

chology have been taught those "truths" about using reinforcement and

feedback. However, recent data on the delay-retention effect (Sassenrath

& Yonge, 1969) indicate that there is now reason to doubt the impor-

tance of immediate feedback. One reason for the doubt is that immediate

feedback sometimes reduces learning when it is compared with delayed

feedback, perhaps because it stops problem solving and other cognitive

processes.

Learners do construct meaning from their memory. Their attitudes do

determine, in part at least, what the effects of the reinforcer are and what

they will learn and remember. The research summarized above means

to the author that learning is a constructive, generative process.

Implications for the Teaching of Mathematics

From the research and theory developed above, implications for teach-

ing and research in mathematics learning will be discussed. In prepara-

tion for writing this section, many articles in issues of the JOURNAL FOR

RESEARCH IN MATHEMATICS EDUCATION were consulted. In addition,

research articles and papers by many others were consulted. Descriptions

of the excellent mathematics curricula that are now being used throughout

the public schools of the United States, such as the SMSG materials pre-

pared under Begle's lead were also consulted.

In those sources it is apparent that many researchers in mathematics

education are delving into the same issues that have been discussed pre-

viously. Suppes (1973) specifies the step-by-step processes involved in

addition and subtraction; Atkinson (1973) considers the learner's de-

tailed history in the design of an instructional program; ATI research

uses intellectual processing variables; and mathematics learning is being

conceptualized from a cognitive perspective (Scandura, 1971).

The two implications for mathematics learning and teaching that will

be emphasized involve the organizational structures and processes dis-

cussed throughout this paper. In structural organization, mathematics

education has made great advances in recent times, largely because of

November 1974 193

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 the carefully organized instructional materials prepared by SMSG and

other curriculum projects. Their authors and researchers have emphasized

hierarchical organization and meaning. The organization of mathematics

around its most fundamental concepts and the teaching of them first to

young children is a most important contribution to education. Research in

ATI, brain processes, and individual differences is needed to determine if

multiple structural organizations of mathematics will improve instruction.

In the area of processing of information, recent research in cognition

has offered important ideas regarding mathematics learning. To sum-

marize these ideas, the learning of mathematics should be viewed as a

generative process. This involves relating the structural organization to the

learner's experience and encouraging the learner to process generatively the

information. An example follows:

Before a mathematics teacher presents a theorem or equation beginning

with a statement such as "Let x = 2y," he should consider what his state-

ments will cause the learner to do. What is the purpose of letting x = 2y?

What is the learner to generate from the equation? How does the equation

relate to the learner's experience-his previous mathematics? What verbal

and spatial processes are involved in the new material?

To answer questions such as these, it is important to study the

step-by-step mathematical and higher-order cognitive processes involved

in every phase of the learning of mathematics. It may seem trivial to

focus on the details, but it is through the study of these specific proc-

esses that a deeper understanding of the learning of mathematics will

develop. Some of the success of the recently developed curricular mate-

rials is due to their understanding of this point. A similar improvement

can be made in the teaching of mathematics by conceptualizing the cogni-

tive processes in it as generative ones.

Socrates taught the slave boy to construct a proof. The boy's thought

processes were engaged in the problem, even though Socrates closely di-

rected the boy's development of the proof. B. F. Skinner dislikes Socrates'

style of teaching for the very reason the author likes it. The instruc-

tion left some room for thought within an organized structure. It left

something important for the learner to do-something to process, to

generate.

Accepting the notion that learning with understanding is a generative

process requires explicit changes in methods of teaching. In a classroom,

one cannot assume that students are at the same level for storing and

retrieving information, or that all students process information in the

same manner. It is important to determine the individual's specific level

of knowledge and his ways of processing information.

The author's research on human learning indicates that the ways teachers

introduce new material, the ways they relate it to the student's experi-

ences, and the ways they stimulate the student to generate meanings are

194 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
crucially important to learning. With proper attention to the introduction

and sequencing of new material, far fewer reinforced repetitions may be

needed for learning to occur.

Plato's ideas about nativism, and about inheritance of abstractions are

not involved in the research on the generative hypothesis. But the author's

research supports this hypothesis that the learner must actively construct

meaning if he is to learn with understanding.

Today, after more than 2000 years of Aristotelian-dominated thought

about human learning, Plato's ideas about teaching are still very much

alive-let's put these ideas to use.

REFERENCES

Aiken, L. R., Jr. Verbal factors and mathematics learning: a review of research.

JOURNAL FOR RESEARCH IN MATHEMATICS EDUCATION, 1971, 2, 304-313.

Anderson, J. R. & Bower, G. H. On an associative trace for sentence memory.

Journal of Verbal Learning and Verbal Behavior, 1971, 10, 673-680.

Anderson, R. C. Control of student mediating process during verbal learning and

instruction. Review of Educational Research, 1970, 40, 349-369.

Atkinson, R. C. Ingredients for a theory of instruction. In M. C. Wittrock (Ed.)

Changing Schools: Alternatives from Educational Research. Englewood-Cliffs, New

Jersey: Prentice-Hall, 1973, 65-82.

Behr, M. J. Interactions between structure-of-intellect factors and two methods of

presenting concepts of modular arithmetic-a summary paper. JOURNAL FOR

RESEARCH IN MATHEMATICS EDUCATION, 1970, 1, 29-42.

Bobrow, S. A. Memory for words in sentences. Journal of Verbal Learning and

Verbal Behavior, 1970, 9, 363-372.

Bobrow, S. A. & Bower, G. H. Comprehension and recall of sentences. Journal of

Experimental Psychology, 1969, 80, 455-461.

Bogen, J. E. The other side of the brain II: an appositional mind. Bulletin of the

Los Angeles Neurological Societies, 1969, 34, 191-220.

Bower, G. H. Organization and memory. Paper read at annual meeting of the

Western Psychological Association, San Diego, California, April, 1968.

Bower, G. H., & Clark, M. C. Narrative stories as mediators for serial learning.

Psychonomic Science, 1969, 14, 181-182.

Bower, G. H., Clark, M. C., Lesgold, A. M., & Winzenz, D. Hierarchical retrieval

schemes in recall of categorized word lists. Journal of Verbal Learning and Verbal

Behavior, 1969, 8, 323-343.

Bower, G. H., & Winzenz, D. Group structure, coding, and memory for digit series.

Journal of Experimental Psychology Monographs, 1969, 80, (2, Pt. 2), 1-17.

Bransford, J. D., & Johnson, M. K. Contextual prerequisites for understanding:

Some investigations of comprehension and recall. Journal of Verbal Learning and

Verbal Behavior, 1972, 11, 717-726.

Bull, B. L., & Wittrock, M. C. Imagery in the learning of verbal definitions. British

Journal of Educational Psychology, 1973, 43, 289-293.

Levy-Agresti, J., & Sperry, R. W. Differential perceptual capacities in major and

minor hemispheres. Proceedings of the national academies of science, 1968, 61,

1151.

Marks, C. B., Doctorow, M. J., & Wittrock, M. C. Word frequency and reading

comprehension. Journal of Educational Research, 1974, 67, 259-262.

November 1974 195

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions
 O'Day, E. F., Kulhavy, R. W., Anderson, W., & Malczynski, R. J. Programmed

instruction techniques and trends. New York: Appleton-Century-Crofts, 1971.

Paivio, A. Imagery and verbal processes. New York: Holt, Rinehart, and Winston,

1971.

Rohwer, W. D., Jr. Children and adolescents: Should we teach them or let them

learn? In M. C. Wittrock (Ed.) Changing schools: Alternatives from educational

research. Englewood-Cliffs, New Jersey: Prentice-Hall, 1973, 103-123.

Rossi, Sheila & Wittrock, M. C. Clustering versus serial ordering in recall among

four-year-old children. Child Development, 1967, 38, 1139-1142.

Rossi, Sheila & Wittrock, M. C. Developmental shifts in verbal recall between mental

ages two and five. Child Development, 1971, 42, 333-338.

Sassenrath, J. M., & Yonge, G. D. Effects of delayed information feedback and feed-

back cues in learning and retention. Journal of Educational Psychology, 1969, 60,

174-177.

Scandura, J. M. A theory of mathematical knowledge: Can rules account for creative

behavior? JOURNAL FOR RESEARCH IN MATHEMATICS EDUCATION, 1971, 2, 183-

196.

Suppes, P. Facts and fantasies of education. In M. C. Wittrock (Ed.) Changing

schools: Alternatives from educational research. Englewood-Cliffs, New Jersey:

Prentice-Hall, 1973, 6-45.

Underwood, B. J. Interference and forgetting. Psychological Review, 1957, 64, 49-60.

Weiner, B. Theories of motivation: From mechanism to cognition. Chicago, Markham

Press, 1972.

Weiner, B., Heckenhausen, H., Meyer, W. U., & Cook, R. E. Causal ascriptions and

achievement behavior: A conceptual analysis of effort and reanalysis of locus of

control. Journal of Personality and Social Psychology, 1972, 21, 239-248.

Wittrock, M. C. Replacement and non-replacement strategies in children's problem

solving. Journal of Educational Psychology, 1967, 58, 69-74.

Wittrock, M. C., & Carter, J. F. Organization as a generative process. In preparation.

(Copy available from M. C. Wittrock).

Wittrock, M. C., & Goldberg, S. M. Imagery and meaningfulness in free recall: Word

attributes and instructional sets. Journal of General Psychology. (in press)

Wittrock, M. C., Doctorow, M. J., & Marks, C. B. Story meaningfulness and com-

prehension, vocabulary, and retention. In preparation. (Copy available from M. C.

Wittrock).

Statement of ownership, management, and circulation (Act of August 12, 1970: Section 3685. Title 39.

United States Code). (1) Title of publication: The JOURNAL FOR RESEARCH IN MATHEMATICS

EDUCATION. (2) Date of filing: 1 October 1974. (3) Frequency of issue: Four times yearly-November,

January, March, and May. (4) Location of known office of publication: 1906 Association Drive, Reston,

Virginia 22091. (5) Location of the headquarters or general business offices of the publishers: Same as #4.

(6) Names and addresses of publisher, editor, and managing editor: Publisher, National Council of Teachers

of Mathematics, same as #4; Editor, J. F. Weaver, University of Wisconsin, Madison, Wisconsin 53706;

Manager Editor, none. (7) Owner: National Council of Teachers of Mathematics, same as #4. (8) Known

bondholders, mortgagees, and other security holders owning or holding 1 percent or more of total amount

of bonds, mortgages, or other securities: none. (10) The purpose, function, and nonprofit status of this

organization and the exempt status for federal income tax purposes have not changed during preceding

12 months. (11) Extent and nature of circulation. Average no. copies each issue during preceding 12 months:

A. Total no. copies printed, 5,551; B. 1. Paid circulation sales through dealers and carriers, street vendors,

and counter sales, none; B. 2. Paid circulation, mail subscriptions, 5,159; C. Total paid circulation, 5,159;

D. Free distribution by mail, carrier, or other means, none; E. Total distribution, 5,159; F. Office use,

left-over, unaccounted, spoiled after printing, 392; G. Total, 5,551. Actual no. copies of single issue pub-

lished nearest to filing date: A. Total no. copies printed, May 1974, 5,424; B. 1. Paid circulation, sales

through dealers and carriers, street vendors, and counter sales, none; B. 2. Paid circulation, mail sub-

scriptions, 5,076; C. Total paid circulation, 5,076; D. Free distribution by mail, carrier, or other means,

none; E. Total distribution, 5,076; F. Office use, left-over, unaccounted, spoiled after printing, 348; G.

Total, 5,424. I certify that the statements made by me above are correct and complete: James D. Gates,

Business Manager.

196 Journal for Research in Mathematics Education

This content downloaded from 160.36.178.25 on Sun, 13 Mar 2016 00:56:17 UTC
All use subject to JSTOR Terms and Conditions