Upload
philippe-c-duchastel
View
214
Download
1
Embed Size (px)
Citation preview
Cognitive design for instructional designAuthor(s): PHILIPPE C. DUCHASTELSource: Instructional Science, Vol. 19, No. 6 (1990), pp. 437-444Published by: SpringerStable URL: http://www.jstor.org/stable/23369878 .
Accessed: 28/06/2014 15:20
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 ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
Springer is collaborating with JSTOR to digitize, preserve and extend access to Instructional Science.
http://www.jstor.org
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
Instructional Science 19: 437—444 (1990) 437 © K lu wer Academic Publishers, Dordrecht - Printed in the Netherlands
FORUM
Cognitive design for instructional design
PHILIPPE C. DUCHASTEL LearnTech, 612 George, Birmingham, MI 48009, USA
Abstract. There is strong interest in the field of instructional design in building expert systems that
can provide advice to inexperienced instructional designers. This paper questions whether the expert
systems model to advice-giving is in fact appropriate for a design process such as instructional design. An alternate approach based on case-based reasoning and the critic approach to advice-giving is con
sidered to be better adapted to the cognitive needs of this task. The cognitive constraints of the task are
used to orient the design of an instructional design wotkbench. The concept of an ID workbench illus
trates the direct application of cognitive science to a complex practical task in the area of design.
Introduction
The aim of this paper is to lay the foundations for the cognitively-based architec
ture of an instructional design workbench. An ID workbench is a computer-based
system that serves as the primary drafting board for the design of instruction. It
provides a framework within which instruction is designed, and provides access to all the tools and information that are needed to accomplish this task in an effi
cient manner. It should also serve as a skill-building tool for the novice, enabling
one's experience to be broadened through access to the instructional design
knowledge contained in the tool.
Two existing systems are early prototypes of an ID Workbench. One, called IDioM for Instructional Development Method (now renamed the ID Library), is a workbench used by Apple Computer Inc. for its training development (Gustafson and Reeves, 1988). The other, IDE (for Instructional Design Environment), was
developed at Xerox as a workbench for its own trainers (Russell et al., 1988). Both are considered in their own context as ID Workbenches and used as such, even if not on an intensive basis. Each has interesting features, borrowed in the
design proposed later. Each is also considered to have limitations, which could
potentially be overcome, hence their characterization here as prototypes. Our aim is to design the ideal ID Workbench, within the constraints imposed
by the current state of near-term computer technology (the technology expected to be available shortly) and, even more importantly, by the current state of the art
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
438
in instructional design. These two factors time-stamp what is 'ideal' in both the
practical sense of building such an ID Workbench, and in the more abstract sense
of what is theoretically possible. One interesting aim later will be to 'unconstrain' the design to see what might come along in the future.
We deal here with the cognitively-based architecture of an ID Workbench. By that we mean that we are concerned not so much with the technical details of the
system (although these are important), as we are with the task requirements of the
design process and how these interact with the designer's own level of experience as he or she engages in instructional design. We will thus be concerned with the
cognitive needs of the designer faced with an instructional design task and how
these needs can orient the architecture of the ID Workbench. Such a needs-based
approach is always present in the design of any system, but often implicitly so. What we want to do here is make the cognitive needs explicit and have them be
the main driver of the Workbench architecture.
This brings us to a discussion of the main competitor of an ID Workbench as
the designer's main job aid, the ID expert system. We believe the expert system
approach has drawbacks which make it inappropriate for instructional design.
The expert system approach to instructional design
Expert systems technology has been around for quite some time now, and it has
been applied to numerous application domains. In the field of instructional
design, it was thought that the ID process could be formally represented as a set
of rules that would drive an ID expert system (Merrill, 1987). That goal is still
one which is currently being actively pursued (Merrill, 1990). There are reasons to believe, however, that that particular approach may be ill-founded.
Expert systems are part of the larger class of consultation systems that provide
specific information to users to help them with particular problems. Other
systems within that larger class are advisory systems, which can coach a user in
performing a given task, critiquing systems, which comment on the performance of a task, and tutorial systems, which teach how to perform a task. These systems are all similar to one another in that they assist a user in one way or another in the
performance of a task. Another commonality between them is that they all
encourage growth in the skill being performed, the tutorial systems being
explicitly aimed at that function. Thus consultation systems have an assistance
function and a skill improvement function.
Where the systems differ is in their emphases on certain aspects of the situa
tion. Table 1, adapted from Duehastel and Brahan (1989), summarizes some of
these emphases (note that the critiquing systems class has been subsumed within
others, for it is either advisory or tutorial in nature depending on circumstances).
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
439
Table 1 : Distinctions between systems
Expert
Systems
Advisory
Systems
Tutoring
Systems
Accomplish task
Help user
accomplish task
Teach user
to accomplish task
User
assists
system
System assists
user
System teaches
user
Real task Real task Simulated task
System control
of process
User
control
of process
Mixed
control
of process
Mild Strong Very strong
Our concern here is mainly with contrasting the first two approaches, the
expert systems (ES) one and the advisory systems (AS) one. The ES approach is
more heavily system-centered than the AS approach. Even though it is the user
who frames the problem to be solved and who finally accepts or rejects the advice
proposed, it is the ES that solves the problem. The user merely inputs the parame ters of the problem and provides any additional information needed.
The ES approach works well with highly technical and potentially critical
problems involving very fine knowledge, such as in medical diagnosis or tax
planning, and with generally inexperienced professional staff. In such situations, users are not reluctant to hand over the task to the ES and to input the masses of
information needed to specify the exact nature of the problem to be solved.
However, as the problem becomes less critical or less technical, and especially as the users overcome any inexperience they may have started out with, the ES
approach's inherent system control of the process makes it unusable in a practical sense: professionals simply will not use the system. Thus, we see that a sense of
control over the analytical process involved in problem-solving is one very
important cognitive need of the professional. Instructional design, by and large, falls into the latter category of problem, that
in which criticality is low and for which a sense of experience is rapidly built up. The implication of this is that the success of an ES approach to instructional
design will likely remain marginal at best.
The alternative approach is the advisory systems one, in which the user
remains much more in the driving seat (especially important in a design task), even though assistance may be offered or requested to help with the accomplish
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
440
ment of the task. Advisory systems are task-centered, just like expert systems, but
they are also much less directive and intrusive in the unfolding of the process under way.
What specifically are the drawbacks of the ES approach in the instructional
design area? There are a number of them:
1. ID is a synthetic process which is not aimed at the derivation of one solution
(as say medical diagnosis is), but rather at the derivation of a whole host of
solutions (possibly a different one for each major segment of instruction).
2. In ID, there may not be a best solution (or a series of best solutions); rather, there may be a set of appropriate solutions which are more or less equivalent in
terms of eventual effectiveness. If that set is large, then a system that points out
inappropriate solutions rather than appropriate ones might be preferable. Expert systems do the latter.
3. ID knowledge as a body of expertise is far from refined enough to handle the
complex situations involved in juggling the many factors to be considered in
the process of designing instruction. Simplifying the approach by disregarding some of the factors might well result in a design that is repetitive and thus
boring.
4. The high volume of content to be included in an instructional product, and thus
processed by a consultation system, makes the rigid data input mechanisms of
expert systems tedious to use.
All of these negative aspects of the ES approach to ID render the approach
rather unpalatable to the experienced instructional designer, at least in terms of
the ES serving as a productivity/quality tool in the design of instruction. It is often argued that, even so, an instructional design ES will be useful to the
inexperienced designer who, while he or she may find it all somewhat tedious, will gradually learn what is involved in quality ID and thus become a designer who can eventually bypass the use of the tool. That would however appear to be a
lame excuse for the ES approach to ID productivity and quality. On the one hand, what is needed (or at least would be very useful) is an automated tool that a
designer of any level of experience can use on a daily basis to improve his or her work (either in terms of higher quality or of reduced time). And on the other
hand, there are better ways to train inexperienced instructional designers than to
have them second-guess the principles operating behind the scenes within an ES.
The task demands of instructional design
Instructional design should be regarded as a problem solving process involving a
large-scale interplay of possibilities and evaluations which eventually leads to a
design embodied in a product, namely the course of instruction (Pirolli, 1989).
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
441
The wide range of possibilities generated and evaluated during an instructional
design effort suggest that prototyping to enable performance modeling would be
an important part of an automated tool to assist instructional design. The problem solving process involved in ID, if we consider it at a very general
level, is one which involves the mental juggling of numerous factors, which must
be combined into an overall situational assessment. This is then matched against the technological possibilities offered by the cumulative experience that the
designer can draw upon. The designer gradually refines the match until an opti mal, or at least acceptable, solution emerges. The optimality/acceptability of the
solution to emerge is judged against a number of usually implicit criteria attached
to the eventual product to be crafted, such as its content validity, understandabil
ity, coherence, attractiveness, and so on.
What this emphasizes is the fact that the process of instructional design is very much a knowledge intensive one, a process that is heavily grounded in the per sonal design experience of the designer. One of the task requirements of ID is
thus to re-expose oneself to one's own previous experience, as well as to delve
into the cumulative experience in the field, in whatever form it may be available. An ID workbench should facilitate this process of contact with previous experi ence, one's own and that of other professionals in the field.
As a process, instructional design is a procedure that embodies a set of fairly
general systems rules (eg. Gagné, 1985) that orient the activity of the designer, without however being too explicit at the micro-level of design. In other words, at
this level, instructional design offers guidance, without any firm set of prescrip tive rules beyond those accessible to common understanding (such as 'teach any
pre-requisites first'). Anything beyond general guidance falls into the realm of
instructional theory and thus into debate (see e.g., Reigeluth, 1983,1987). The instability of instructional theory is likewise underlined by Merrill (1990),
who points to the desirability of nevertheless attempting to construct a declarative
knowledge base of instructional rules that could be used in an advice-giving sys tem. The assumption underlying such a view is that a generally agreed-upon set
of rules capturing the process of instructional design can be constructed over
time. However, with the wide variety of theoretical viewpoints current in the
instructional design field, it proves difficult for that assumption to be upheld. The implication is that the ID process cannot be easily reduced to a procedural
implementation of a theory-based view of the field. Thus, following a procedural formalism is not one of the task requirements of ID, except at a very general level
that does not capture the full richness of the process. Even if rules are currently lacking (or considered unstable or overly general as
a set), instructional design is nevertheless being carried out by instructional
designers on a daily basis. What most likely is occurring is a form of case-based
reasoning in which the designer's experience is put to use in the alternatives
generation phase of design. In other words, design solutions from the past are
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
442
examined to determine the degree to which their design problem spaces parallel
the current problem space. Analogical reasoning could thus be the driving force
behind instructional design rather than a more orderly rule-based reasoning. If
that is a likely task requirement for ID, and that is what is being suggested here, then an important addition to an instructional design workbench would be a
library of design solutions which the inexperienced instructional designer could
draw from in tackling his or her own current design problem. This is what is
meant by contact with the field's cumulative experience. Yet another facet of instructional design is the large number of elements
involved in the design of a course of instruction: tasks, objectives, expository ele
ments, interactive transactions, test items, and so on. All of these elements must
be inter-related in a coherent manner and the instructional designer must ensure
that overall coherence is maintained This is a particular need with very large scale design programs that include a great many instructional objectives. This
recognized task requirement for consistency and coherence can be aided by a
checking process that points out areas of potential difficulty in these respects. In summary, the main task demands for instructional design appear to be:
- toying out many different design options in a generation-evaluation repetitive
cycle; - contact with the prior accumulated experience of the field;
- access to a library of design solutions that can be adopted and adapted; - checking for consistency and coherence across the multiple elements of the
design.
These task demands are pertinent at the cognitive level of accomplishing the
task. However, as mentioned earlier, one essential task demand, and certainly one
of the most important, lies at the conative level: the sense on the part of the
designer that he or she is the one in creative charge of the design process. Without
the latter, the design process will not proceed effectively, nor efficiently.
Designing an Instructional Design Advanced Workbench
Now that certain essential task requirements for practical instructional design work have been made explicit, it is possible to design the architecture of a com
puter-based workbench to support these functions. We call this workbench
ID AW, for Instructional Design Advanced Workbench.
The proposal embodied in ID AW is that productivity enhancement in ID lies in
giving the designer access to support that will guide without constraining. Three
categories of support are needed for productive ID: information, tools, and
advice, as illustrated in Figure 1.
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
443
f \ TOOLS
°o° Task analysis
Story boarding Lesson prototyping Simulation
V )
( INFORMATION
Alternatives
Taxonomies ISD process
Example lessons
V
Structural design Media selection
Transaction analysis
J V. J
Figure 1. The principal components of ID AW
The provision of such support in an efficient, easily-used manner will require a
computer workstation with two major characteristics:
- advanced design features in software engineering and in cognitive engineering; - substantial content relevant to ID.
The IDAW concept is not modest, but also, is not overly distant. Most of the
technology for prototyping IDAW is available, although the more advanced ideas
in the advice area need expansion and more development work. The design speci fications for IDAW are outlined in Duchastel (1990).
Prospects for IDAW
Some of the initial specifications for IDAW presented above can readily lead to a
prototype design within the scope of current technology. Others are more open to
risk in this respect. The latter ones in particular, concerned with intelligent advice-giving, will require rather more R & D to better define and prototype. The
general IDAW concept, however, appears sound in that it would prove useful
even without much of the functionality described under the advice set. This latter
functionality can be incrementally built into IDAW as progress is made.
We recommend a program of research to pursue the line of investigation embodied in the IDAW concept. The incremental nature of the elements of
IDAW make prototyping IDAW a relatively straightforward effort; an initial
IDAW shell with some of the functionality presented in this document would be
sufficient to generate further interest in the field of instructional design, as well as
in the community of eventual IDAW users (professional instructional designers). As the IDAW concept is demonstrated and evaluated in practice, further func
tionality can be continually added to the system until the process of refinement
discussed earlier takes over to fine tune it.
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions
444
The IDAW concept is a direct challenge to the currently fashionable expert
systems approach to instructional design. It was argued above that various task constraints prevent in theory the practical use of expert systems for instructional
design. This is but one perspective on the issue: other perspectives are certainly around. However, until systems are built and tried out in practical ID contexts,
perspectives will remain invalidated.
What is required now is that various challenging perspectives be tried out and
compared. Thus, it is recommended that, alongside the current development of ID
expert systems, the concept of IDAW be pursued through a prototyping effort that will indicate the level of its usefulness for practical instructional design.
References
Duchastel, P. (1990). A cognitive-based analysis of instructional design for instructional design tool
building. Internal Document, McDonnell Douglas Training Systems, Denver, Colorado.
Duchastel, P. and Brahan, J. (1989). Functionality considerations in a generic advisor system. In B.
Moulin (Ed.), L'informatique cognitive des organisations. Lyon, France: L'Interdisciplinaire, 1989,287-296.
Gagné, R. (1985). The conditions of learning and theory of instruction. (4th edition) New York: Holt, Rinehart and Winston.
Gustafson, K. and Reeves, (1988). IDioM: an instructional design workstation. Gottlieb Duttweiler
Conference. Merrill, D. (1987). An expert system for instructional design. IEEE Expert, 2(2), 25-40.
Merrill, D., Li, Z. and Jones, M. (1990). Second generation instructional design. Educational
Technology, January 1990.
Pirolli, P. (1989). On the art of building: putting a new instructional design into practice. In H. Bums and J. Parlett (Eds.), Proceedings of the 2nd Intelligent Tutoring Systems Research Forum. San
Antonio, TX. April 1989.
Reigeluth, C. (1983). Instructional design theories and models: an overview of their current status.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Reigeluth, C. (1987). Instructional theories in action: lessons illustrating selected theories and mod
els. Hillsdale, NJ: Lawrence Erlbaum Associates.
Russell, D., Moran, T. and Jordan, D. (1988). The Instructional Design Environment. In J. Psotka, L
Massey and S. Mutter (Eds.), Intelligent Tutoring Systems: lessons learned. Hillsdale, NJ:
Lawrence Erlbaum Associates.
This content downloaded from 91.213.220.138 on Sat, 28 Jun 2014 15:20:27 PMAll use subject to JSTOR Terms and Conditions