John F. (John Fralick) Rockart.

Computers and the learning process online

. (page 1 of 2)
Online LibraryJohn F. (John Fralick) RockartComputers and the learning process → online text (page 1 of 2)
Font size
QR-code for this ebook


LIBRARY

OF THE

MASSACHUSETTS INSTITUTE
OF TECHNOLOGY






Center for Information Systems Research

Massachusetts Institute of Technology

Alfred P. Sloan School of Management

50 Memorial Drive

Cambridge, Massachusetts, 02139

617 253-1000



COMPUTERS AND THE LEARNING PROCESS



Jolin F. Rockart



REPORT CISR-15
SLOAN WP 802-75




August 1975



FO REWARD



This p^per is an ^typical C.I.E.R. paper. It is included in the
series, however, since it is felt to be of interest to a segment of
C.I.S.R. sponsors for the reasons noted below.

The paper was presented at the 10th Educom Fall Conference held
in Toronto, Ontario, Canada several months ago and is a summary of the
book Coir\puters qnd the Learning Process in Higher Education to be
published by McGraw-Hill in September, 1975. Written by John F. Rockart
and Michael S. Scott Morton, the book presents several research results
and a resulting framework for the use of computers in higher education
(l(_^v(' loped by Morton and Rockart.

The paper is included in the C.I.S.R. series because:

(1) Some C.I.S.R. sponsors are computer manufacturers interested
in utilizing their products in the area of education. The
paper therefore is relevant to them.

(2) An increasing number of computer-user organizations utilize tht>ir
on-line systems for internal training and education, and it is
hoped that the work is of some interest to them also.

(3) Finally, it may be of general interest to computer-minded
executives — providing a new slant on a field that is certain
to grow - if slowly — computers and education.



0725161i



COMPUTERS AND THE LEARNING PROCESS

by

John F. Rockart



In order to discuss the impact of computers on learning, it is important
to understand the learning process - the manner in which people obtain and
assimilate knowledge. I would argue that one must have a working model of
the learning process if one is to design computer systems to assist it.

A significant point is that we all do have implicit models of what learning
is, and therefore, how computers can assist this process. And each of us uses
his or her implicit model when we design computer systems to assist learning.
The real need in the field, I believe, is to make these models explici t.
Only then can we lay bare the assumptions upon which each of us bases the
fundamental design of our computer-based learning aids. These assumptions,
as noted later, are critical — for they hugely influence the type and
exact form of the computer tools that we design to assist the learner.

This paper presents some conclusions that Michael Scott Morton and I
have come to as a result of reviewing the literature in the field of technology-
assisted learning, and also of working in the field for several years (albeit
at intervals). A much more expanded version of this paper, which I trust
will fill in the many chinks these few pages leave open, will be available
soon in the book Computers and the Learning Process in Higher Education , by
Michael and myself. This will be published in the Carnegie Commission series
on Higher Education.

The search for a precise model of the learning process is a difficult
one. Unfortunately, the majority of the published material with regard to
the learning process, while intellectually rewarding, is of little assistance



- 2



to the designer of computer-based learning aids. What is needed is an opera-
tional statement of the learning process — one that can be acted upon in
the design of course materials, the design of a pedagogical strategy, and
the assessment of a place of computer technology in that strategy.

A precise "model" enables one to describe and partition the possible
impact of various technologies on segments in the learning process. Of equal
importance, it enables others to test and offers conclusions about the
effects of the technology with reference to the stated model. Finally, others
can test the author's conclusions against his own model of the learning
process, and determine whether differences in perspective are based on
differing perceptions of learning or differing perceptions of technology .

Neither Professor Morton nor I would claim that our particular view
of the learning process, and how it can be aided by computers, is "right."
Rather, I present here an explicit model of the learning process which we
have found operational and useful as a point of departure from which to build
systems to aid education.

The State of Learning Theory

Over the years learning has been considered primarily the domain of
psychology. Since the mid-1880 's when the first really useful work on
memory was published (Ebbinghaus, 1895) , there has been a series of attempts
at explaining learning phenomenon. Unfortunately, many of these attempts
have been at or near the level of grand theory and have been in conflict
with each other. A good summary of these theories is found in Hilgard and
Bower (1966) .

In general, two major theories of the learning process have dominated
the field through the last few decades. These are known as the stimulus -
response (SR) theory, and the cognitive theory. A third major theoretical



- 3 -



approach to learning, that of Piaget , has also entered this field more recently
but will not be discussed here. As Figure 1 shows, the two major theories
can be separated on three major characteristics. The stimulus-response
school (perhaps best known through the experiments and written works of
Pavolv and Skinner) , suggest that what we perceive as learning is
merely a "chained muscular response." In effect, the SR school suggests
that people learn habits and learn new things by applying the closest old
habit until one fits the situation. If there is no old learned habit that
comes close to fitting the situation, one learns through trial and error.
The cognitive school, on the other hand, (represented by pyschologists
such as Tolman and Bruner) suggests that what goes on is really a central
brain process in which the learner learns cognitive structures — similar
to computer programs. New things are learned by "insight," which is a
process of comparing new situations with old cognitive structures. Exactly
how this occurs is unexplained. Clearly, at least to me, the latter theory
is somewhat more appealing. Skinner's (1972) writings, which reflect the
SR theory, remove all free will from mankind. But personal choice of theory,
intuitively or logically, is not important. What is important is the
usefulness of these theories in designing learning aids.

Unfortunately, it is almost impossible to use these theories pragmatically.
The only viewpoint one can take, and have explicit action flow from it, is;
a strict SR stance. If one does so, one is thrust in lock-step mode toward
the exclusive use of Skinnerian programmed learning. Whether explicitly or
implicitly because of this, much of the early work done in the field of
computers and learning followed these lines. Yet this approach seems far
too single-minded to encompass the entire field of learning with the rich-
ness and variation which we believe it holds.



4 -



Thus existing learning theory left us somewhat unimpressed. These
theories suggested only generalized (and/or single-minded) approaches to
the technology. They left too few specific knobs to turn. They failed
to suggest specific areas of potential success for learning technology.
And they failed to provide guidance for the most useful action steps that
could be taken to utilize computers/ as opposed to other learning aid,
by the individual professor with a particular course to teach,

Dave Kolb, one of our colleagues of M.I.T., has summarized the
pragmatic failures of these theories well :

"Because of my early psychological training in learning theory
my first impulse was to turn for the answer to these questions to
the basic psychological literature on this subject. To my dismay
I found that, in spite of the high scientific quality of this work,
it was immensely difficult to apply this research on reinforcement
theory, discriminative learning and such to the kind of nractical
decisions involved in the design of university teaching." (Kolb, 1974, p.l)

This same frustration led us to review the work of practitioners in
the field of learning, as well as the theorists, in an attempt to elicit
a useful learning model. Five sets of critical variables with regard to
learning emerged from this search. They are:

The stages of the learning process

The characteristics of the material to be learned

The characteristics of the learner



The characteristics of the teacher
The learning environment

Of these five categories, the first two appeared most significant for
our purposes. The last two are very complex variables which we found terribly
difficult to model. At the time of the development of our model, also, the
characteristics of the learner — the third variable in the above list — was
not as well researched as it is at the time of the writing of this paper.
Therefore, it was not included.



We thus limited our modeling of the learning process to two variables —
the two that seemed to us (1) significant, (2) operationally describable,
and (3) least apt to be changed by the conclusions from the study. These
two variables are the "stages of the learning process" and "characteristics

of material to be learned."

The Stages of the Learning Process

The search for an operationally useful statement of the learning process
is not an easy one. We finally centered on a model developed by behavioral
scientists — but one which could be generalized to the learning process
in all fields.

The behavioral approach to describing the learning process which serves
as the basis for our own model follows the work of Kolb (1971) . Reacting
to the frustration of the inapplicability of learning theory for the practical
educator, and needing something to guide his efforts in the educational
process, Kolb turned to the so-called experiential learning model (Figure 2) —
a model that, adapted somewhat, we have used as the core of our work.
Developed primarily out of the experience of sensitivity-training practitioners
(Schein and Bennis, 1965) , the model has gained increasing acccpt(i .



- 8 -



For any particular course or department, this breakdown of material
can easily be further divided. Category 2 (skills) can (perhaps should)
be broken down into several highly distinct types of skills. Indeed it is
necessary to subdivide these categories further when dealing with a
particular course.

But, across courses and departments, we find that this four-part categori-
zation serves as a good vehicle for understanding the applicability of
technology to particular curriculums. We found it meaningful to several of
our colleagues who were able, in a survey, to easily classify the material
taught in their courses within the bounds of this classification.

It does not take much imagination or deep thought to hypothesize that
the relative emphasis on these four categories of material will vary with
particular university departments and with differing degree-level programs
(e.g., undergraduate, master's, doctoral). This, in turn, will mean that
the type of learning technology that can be effectively employed will also
differ from course to course. The type of learning technology that can be-
effectively employed will therefore also differ. Yet the four classes of
material appear to be a resonable division of material types.

An Operationally-Useful Learning Model

Our basic learning model, then, has two variables — learning stages
and material classes — that together produce 16 cells as shown in Figure 4.
It is this two dimensional structure which must be confronted when asking
the question as to where the the computer fits in the learning process.
The process involved in learning in each of the cells is different , we
submit; and the technology utilized must be fitted to these differences.
Therefore, the computer fits in differing ways into each of these cells;
in some it is not very useful at all today.



- 9 -



The fundamental proposition this suggests is that one cannot think of
the use of "the computer in learning" as a whole without doing a great
disservice to the variety of material and the complexities of the differing
stages of the learning process. The computer must be expected to be
utilized in different modes in each area (or each cell of the model shown
in Figure 4) . Some of the cells, I would submit, are almost impregnable
by "computer aids to learning" as they are currently available today.
Others are ripe for differing types of computer assistance to the learning
process.

We now have before us a learning model divided into 16 cells. The
following sections (1) briefly note the many "learning aids" available
to assist learning, (2) analyze the learning demands of each of tlie cells
of the learning model, and then (3) attempt to match the appropriate
technology to these demands. In this paper, because of space and time,
this is done in a very cursory manner. The process is much more full
explicated in our book.

Learning Aids . It is clear that today we have a large set of avail-
able mechanisms to assist learning. These include the professor (who can
lecture, lead a class discussion, etc.); books; video tape; computers
(which can be used in tutorial mode (more commonly called "|>roi)r\immed
Instruction"), for drill and practice, for gaming, simulation, etc.).
Each of these mechanisms can be rated as to its effectiveness on a well-
defined set of attributes. For example, some mechanisms are very economical
per fact presented. Some have the attribute of decentralized availability —
text books have this; computers in most cases today do not. Other attributes
such as emotional impact , sensory impact , the ability to telescope time , etc .
are possessed to a greater or lesser extent by each learning mechanism. We
have defined a set of sixteen of these attributes (Figure 5) and find that



- 10



the learning aids available today vary widely along many dimensions with
regard to each of these attributes. The major learning aids and our ranking
of them on each of the sixteen attributes (where l=high and 10=low) are
shown in Figure 6.

Learning "demands" of Each Cell . The matrix shown in Figure 4 suggests
that the 16 cells are each somewhat unique and thus that the type of learn-
ing that takes place in each cell requires differing types of learning
assistance. Just as the learning aids noted above have different attributes,
the cells of the learning matrix need different types of learning assistance ■
and must be matched by the choice of that (those) learning aid(s) which
best fit these needs.

Four examples may best illustrate the concept of matching the correct
learning aid with the learning process requirement of each particular cell.
The upper left cell of the matrix (the acquiring of facts) has two key
learning attributes needs as shown in Figure 7. There are many facts to
be learned, therefore, the learning tool which aids this learning should
therefore be (1) cheap, and (2) available whenever and wherever the student
wishes to turn this learning process. Clearly the learning aid which
ranks high on these attributes (#15 and 16, and from Figure 5) is the
textbook . Relative to all the other learning aids it is clearly dominant
at the present time with regard to economy of presentation of facts and
decentralized availability.

Moving to the lower left corner of the matrix, a different situation
exists with regard to the acquisition of frontier concepts. Here one
is looking for a learning aid which can structure ill-structured material
for the learner, is able to promulgate new material in a timely way, and
can easily adjust to individual needs with regard to ill-structured
material. The dominate learning aid here is clearly the professor , most
probably in lecture or class discussion mode.



11



Now let us shift to the lower right-hand corner of the matrix. For
the testing out of frontier concepts in areas other than the one in which
they were learned, it is clear that the learner needs the availability
of a large data base, methods of manipulating this data base, and —
highly desirable — a method of "trying out" the concepts in a way
which does no harm to anyone. This testing out process has been done in
the past in the "real world" quite often resulting in some damage to
both the testees and the tester. We would suggest that the ideal learning
mechanism for this cell today, is not the real world as a test ground, but
rather computers and simulation models. In this way the student can
adequately test out his ideas in a simulated world if he has access to
enough data and enough computing power. The learning aid of choice thus
becomes computers .

Moving back to the second cell from the left at the top of the matrix,
in order to embed facts, there is a need for feedback to the learner —
as immediate a: possible — in order that he can know how well he has
performed on answering questions which test whether he has or has not
acquired the facts which he is supposed to have learned. This is usually
done today through "homework." Unfortunately, the lag in time from the
point of which the student completes his homework to the time he receives
written feedback from the professor or teaching assistant is often quite
long. It is also desirable if there is "learner control" (the ability for
the learner, for example, to omit imbedding exercises on material he has
previously learned) associated with the homework process. Under current
paper and pencil methods, this is difficult to do as the learner usually
is required to perform all of the exercises given. A computer-based
"drill and practice" interactive system in which the student is "tested,"
allowed to answer, and given immediate feedback as to the correctness of
his answer, is much the preferable mechanism in this cell. With continuing
decreases in computer costs, we expect that it will, in the relatively near
future, become the mechanism of choice for embedding f.icts, skill;., etc.



- 12 -



This discussion of "matching" appropriate learning aids to the need of
each cell can be generalized somewhat to develop groups of cells which will
benefit from the same general treatment. Figure 8 suggests this generali-
zation. Five distinct areas are shown. We believe that area I belongs
primarily to the textbook. As a g


1

Online LibraryJohn F. (John Fralick) RockartComputers and the learning process → online text (page 1 of 2)