Content uploaded by Alexander Renkl
Author content
All content in this area was uploaded by Alexander Renkl on Aug 11, 2015
Content may be subject to copyright.
Content uploaded by Alexander Renkl
Author content
All content in this area was uploaded by Alexander Renkl on Aug 26, 2014
Content may be subject to copyright.
This article was downloaded by: [Universitaetsbibliothek Freiburg]
On: 11 August 2015, At: 02:44
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place,
London, SW1P 1WG
Educational Psychologist
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/hedp20
Cognitive Load Theory and Instructional Design: Recent
Developments
Fred Paas , Alexander Renkl & John Sweller
Published online: 08 Jun 2010.
To cite this article: Fred Paas , Alexander Renkl & John Sweller (2003) Cognitive Load Theory and Instructional Design: Recent
Developments, Educational Psychologist, 38:1, 1-4, DOI: 10.1207/S15326985EP3801_1
To link to this article: http://dx.doi.org/10.1207/S15326985EP3801_1
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
PAAS, RENKL, SWELLERINTRODUCTION
Cognitive Load Theory and Instructional Design:
Recent Developments
Fred Paas
Educational Technology Expertise Center
Open University of The Netherlands, Heerlen
Alexander Renkl
Department of Psychology
University of Freiburg, Germany
John Sweller
School of Education
The University of New South Wales, Sydney, Australia
Cognitive load theory (CLT) originated in the 1980s and un-
derwent substantial development and expansion in the 1990s
by researchers from around the globe. As the articles in this
special issue demonstrate, it is a major theory providing a
framework for investigations into cognitive processes and in-
structional design. By simultaneously considering the struc-
ture of information and the cognitive architecture that allows
learners to process that information, cognitive load theorists
have been able to generate a unique variety of new and some-
times counterintuitive instructional designs and procedures.
The genesis of this special issue emerged from an interna-
tional symposium on CLT that was organized at the 2001 Bi-
annual Conference of the European Association for Research
on Learning and Instruction, Fribourg, Switzerland. Most of
the articles that follow are based on contributions to that sym-
posium and discuss the most recent work carried out within
the cognitive load framework. Before summarizing those ar-
ticles, we provide a brief outline of CLT.
Althoughtheinformationthatlearnersmustprocessvaries
on many dimensions, the extent to which relevant elements
interact is a critical feature. Information varies on a contin-
uum from low to high in element interactivity. Each element
of low-element interactivity material can be understood and
learned individually without consideration of any other ele-
ments. Learning what the usual 12 function keys effect in a
photo-editing program provides an example. Element
interactivity is low because each item can be understood and
learned without reference to any other items. In contrast,
learning how to edit a photo on a computer provides an exam-
ple of high-element interactivity. Changing the color tones,
darkness, and contrast of the picture cannot be considered in-
dependently because they interact. The elements of high-ele-
ment interactivity material can be learned individually, but
they cannot be understood until all of the elements and their
interactionsareprocessedsimultaneously. As a consequence,
high-elementinteractivitymaterialisdifficulttounderstand.
Element interactivity is the driver of our first category of
cognitive load. That category is called intrinsic cognitive
load becausedemandsonworking memory capacity imposed
by element interactivity are intrinsic to the material being
learned. Different materials differ in their levels of element
interactivityandthusintrinsiccognitiveload,and they cannot
be altered by instructional manipulations; only a simpler
learning task that omits some interacting elements can be
chosen to reduce this type of load. The omission of essential,
interacting elements will compromise sophisticated under-
standing but may be unavoidable with very complex, high-el-
ement interactivity tasks. Subsequent additions of omitted
elements will permit understanding to occur. Simultaneous
processingofall essential elements must occur eventually de-
spite the high-intrinsic cognitive load because it is only then
that understanding commences.
One may argue that this aspect of the structure of informa-
tionhas driven the evolution of human cognitive architecture.
An architecture is required that can handle high-element
interactivity material. Human cognitive architecture met this
requirement by its combination of working and long-term
EDUCATIONAL PSYCHOLOGIST, 38(1), 1–4
Copyright © 2003, Lawrence Erlbaum Associates, Inc.
Requestsforreprintsshouldbe sent to JohnSweller,SchoolofEducation,
University of New South Wales, Sydney, NSW 2026, Australia. E-mail:
j.sweller@unsw.edu.au
Downloaded by [Universitaetsbibliothek Freiburg] at 02:44 11 August 2015
memory. Working memory, in which all conscious cognitive
processing occurs, can handle only a very limited num-
ber—possibly no more than two or three—of novel interact-
ing elements. This number is far below the number of
interacting elements that occurs in most substantive areas of
human intellectual activity. Alone, working memory would
only permit relatively trivial human cognitive activities.
Long-term memory provideshumanswith the ability to vastly
expand this processing ability. This memory store can con-
tain vast numbers of schemas—cognitive constructs that in-
corporate multiple elements of information into a single
element with a specific function.
Schemas can be brought from long-term to working mem-
ory. Whereas working memory might, for example, only deal
with one element (e.g., a cognitive load that can be handled
easily), that element may consist of a large number of lower
level, interacting elements. Those interacting elements may
farexceedworkingmemorycapacity if each elementhadtobe
processed. Their incorporation in a schema means that only
one element must be processed. If readers of this article are
giventheproblemof reversingthelettersof thelastwordofthe
last sentence mentally, most will be able to do so. A schema is
availableforthiswrittenwordalongwithlower level schemas
fortheindividuallettersandfurtherschemasforthesquiggles
that make up the letters. This complex set of interacting ele-
ments can be manipulated in working memory because of
schemas held in long-term memory. The automation of those
schemas so that they can be processed unconsciously further
reduces the load on working memory. It is by this process that
human cognitive architecture handles complex material that
appears to exceed the capacity of working memory.
CLT is concerned with the instructional implications of
this interaction between information structures and cognitive
architecture. As well as element interactivity, the manner in
which information is presented to learners and the learning
activities required of learners can also impose a cognitive
load. When that load is unnecessary and so interferes with
schema acquisition and automation, it is referred to as an ex-
traneous or ineffective cognitive load. Extraneous cognitive
load is a second category of cognitive load. Many conven-
tional instructional procedures impose extraneous cognitive
load because most instructional procedures were developed
withoutanyconsiderationorknowledgeofthestructureofin-
formation or cognitive architecture. For example, any in-
structionalprocedurethatrequireslearnerstoengageineither
a search for a problem solution or a search for referents in an
explanation (i.e., when Part A of an explanation refers to Part
B without clearly indicating where Part B is to be found) is
likely to impose a heavy extraneous cognitive load because
working memory resources must be used for activities that
are irrelevant to schema acquisition and automation. The arti-
cles in this special issue are concerned with this second cate-
gory of cognitive load, extraneous cognitive load, and,
indeed, cognitive load theorists spend much of their time de-
vising alternative instructional designs and procedures that
reduce extraneous cognitive load compared to convention-
ally used procedures.
Extraneous cognitive load is primarily important when in-
trinsic cognitive load is high because the two forms of cogni-
tive load are additive. If intrinsic cognitive load is low, levels
of extraneous cognitive load may be less important because
total cognitive load may not exceed working memory capac-
ity. As a consequence, instructional designs intended to re-
duce cognitive load are primarily effective when element
interactivity is high. When element interactivity is low, de-
signs intended to reduce the load on working memory have
little or no effect.
Thelastformofcognitiveloadisgermane oreffective cog-
nitive load. Like extraneous cognitive load and unlike intrin-
sic cognitive load, germane cognitive load is influenced by
the instructional designer. The manner in which information
is presented to learners and the learning activities required of
learners are factors relevant to levels of germane cognitive
load. Whereas extraneous cognitive load interferes with
learning, germane cognitive load enhances learning. Instead
ofworking memory resources being used to engage in search,
for example, as occurs when dealing with extraneous cogni-
tive load, germane cognitive load results in those resources
being devoted to schema acquisition and automation. Note
that increases in effort or motivation can increase the cogni-
tiveresources devoted to a task. If relevant to schema acquisi-
tion and automation, such an increase also constitutes an
increase in germane cognitive load.
Intrinsic, extraneous, and germane cognitive loads are ad-
ditive in that, together, the total load cannot exceed the work-
ing memory resources available if learning is to occur. The
relations between the three forms of cognitive load are asym-
metric.Intrinsiccognitive loadprovidesabase loadthatisirre-
ducible other than by constructing additional schemas and
automating previously acquired schemas. Any available
working memory capacity remaining after resources have
beenallocatedtodealwithintrinsiccognitive load can beallo-
cated to deal with extraneous and germane load. These can
workintandeminthat,forexample, a reduction in extraneous
cognitive load by using a more effective instructional design
can free capacity for an increase in germane cognitive load. If
learning is improved by an instructional design that reduces
extraneous cognitive load, the improvement may have oc-
curredbecausetheadditionalworkingmemorycapacityfreed
bythereductioninextraneouscognitiveloadhasnowbeen al-
locatedtogermanecognitiveload.Asaconsequenceoflearn-
ing through schema acquisition and automation, intrinsic
cognitive load is reduced. A reduction in intrinsic cognitive
load reduces total cognitive load, thus freeing working mem-
ory capacity. The freed working memory capacity allows the
learnertousethenewly learned material inacquiringmoread-
vancedschemas. A new cycle commences; over many cycles,
very advanced knowledge and skills may be acquired.
Such alterations in expertise also have profound instruc-
tional implications that were realized in the late 1990s. Until
2PAAS, RENKL, SWELLER
Downloaded by [Universitaetsbibliothek Freiburg] at 02:44 11 August 2015
that time, research had focused on rather static situations in
which novices were confronted with high-interactive materi-
als resulting in a fixed level of intrinsic cognitive load, which
couldnot be altered by instructional manipulations. Although
it was stated theoretically, the changes in cognitive load that
occurred as a function of increasing learner’s expertise were
not considered from an instructional perspective. Within this
static focus, two instructional goals can be characterized. Ini-
tially, cognitive load research was aimed at the development
of instructional techniques to reduce extraneous cognitive
load. The goal specificity, worked examples, completion,
split-attention,redundancy,andmodalityeffectsarethefruits
of these research efforts. Under the assumption of a fixed in-
trinsic load and working memory capacity, the successful re-
duction of extraneous load naturally leads to the hypothesis
that the freed capacity could be deployed for techniques that
increase germane cognitive load. Employing example vari-
ability and prompting imagination are instructional tech-
niques that have been used to substitute extraneous load with
germane load.
With the publication in the late 1990s of research on levels
of expertise in instructional design, a second, more dynamic
line of cognitive load research began to materialize. The dy-
namic approach provides an opportunity for researchers to
consider intrinsic load as a property of the task–subject inter-
action, which is open to instructional control. Typically, re-
search within this line studies instructional techniques that
take into account the alterations in the cognitive load that oc-
cur as learners’ levels of expertise increase to facilitate the
transition from novice to expert. The dynamic line’s main
outcome can be summarized as the expertise reversal effect,
indicating that instructional techniques that are effective with
novices can lose their effectiveness and even become ineffec-
tive when used with more experienced learners.
In one way or another, the articles in this special issue re-
flect this theory. The first three articles are all directly con-
cerned with this new, major concern of CLT: How should
instructional design be altered as a learner’s knowledge in-
creases? Schematic information held in long-term memory
will, as just indicated, have dramatic consequences on the
characteristics of working memory. What, in turn, are the in-
structional consequences?
The article by van Merriënboer, Kirschner, and Kester ad-
dresses this issue by beginning with the premise that learners
should be presented realistic tasks despite the fact that, when
dealing with complex areas, realistic tasks presented to nov-
ices with only limited schematic knowledge are likely to im-
pose a heavy cognitive load. Van Merriënboer et al. suggest
two forms of scaffolding to take into account when consider-
ing the alterations in cognitive load that occur with experi-
ence in a domain. The intrinsic aspects of cognitive load can
be reduced by the scaffold of simple-to-complex sequencing,
whereas the extraneous aspects can be reduced by providing
the substantial scaffolding of worked examples initially, fol-
lowed by completion problems and then full problems. (As
mentioned next, Renkl & Atkinson describe a related fading
procedure.) In addition, van Merriënboer et al. indicate that
the timing of essential information presented to students can
be critical from a cognitive load perspective, with inappropri-
ate timing unnecessarily increasing load. They suggest that
general, overarching supportive information be presented
first so that learners can construct a schema to be used
throughout the task, whereas specific procedural information
should be presented only at the particular point when it is re-
quired. Lastly, the authors present their four-component in-
structional design model that integrates the various
instructional design principles outlined in their article.
The use of worked examples rather than solving the equiv-
alent problems is one of the earliest and probably the best
known cognitive load reducing technique. Renkl and
Atkinson are concerned with the role of worked examples
when learning to solve particular classes of problems and,
specifically,howthatroleshouldchangeaslearners’levelsof
expertise increase. They suggest that in the earliest stages of
learning, when intrinsic cognitive load is high because few
schemas are available, learners should study instructions;
during intermediate stages when schema formation has freed
some working memory capacity, they should study worked
examples and increase germane load by using self-explana-
tions; in the final stages, there should be sufficient working
memory capacity to permit more problem solving. Renkl and
Atkinsondescribe the fading technique to facilitate the transi-
tion from the intermediate to final stages. Complete worked
examples are faded by successively eliminating sections of
the worked example until eventually only a full problem re-
mains.Theintermediate,fadedworkedexamplesarecomple-
tion problems that are discussed in the van Merriënboer et
al.’s article. This fading technique was found to be superior to
the traditional procedure of alternating worked examples and
problems.
Kalyuga, Ayres, Chandler, and Sweller review research
directly concerned with the consequences of differing levels
of expertise on cognitive load effects. They indicate that
manyinstructional design recommendations proceed without
an explicit reference to learner knowledge levels. Research is
reviewed demonstrating that a large number of CLT effects
that can be used to recommend instructional designs are only
applicable to novices and can disappear and even reverse as a
function of increasing expertise. Kalyuga et al. provide an
overview of this so-called expertise reversal effect by coordi-
nating and unifying multiple empirical observations of the in-
teractions between instructional techniques and levels of
learnerexpertise and show that the effect has a plausible theo-
retical explanation within a cognitive load framework.
Whereas the first three articles deal with issues tradition-
ally considered by cognitive load theorists, Gerjets and
Scheiter are concerned with procedures in which learners
rather than instructors make instructional decisions. CLT
usually has assumed that instructors rather than novice learn-
ersshould decide what should be studied and how it should be
INTRODUCTION 3
Downloaded by [Universitaetsbibliothek Freiburg] at 02:44 11 August 2015
studied. The worked example effect in which studying
worked examples can be superior to solving the equivalent
problems provides the clearest example. Nevertheless, as the
first three articles indicate, there now is strong evidence that,
as levels of expertise increase, it is appropriate to decrease in-
structor control and increase learner control. Under these cir-
cumstances,Gerjets and Scheiter’s analysis with its emphasis
on learner control is timely. They criticize the fact that CLT
researchtypicallyassumes a one-to-one mapping between in-
structional design and a resulting pattern of extraneous and
germane cognitive loads without taking into account other
moderating variables, such as learner goals that interfere with
this direct mapping. An extension to CLT is proposed along
with the moderating factors of the configuration of teacher
and learner goals and the learner’s processing strategies that
are used to accomplish these goals. Data from four experi-
ments on hypertext instruction are summarized to support the
claim that CLT should take these factors into account when
making predictions for instructional material.
In their article, Mayer and Moreno show why CLT pro-
vides a very fruitful perspective in the area of multimedia
learning. All too often, learners in multimedia environments
experience cognitive overload when dealing with the com-
plexity of text and pictorial presentations. Five overload sce-
narios are described; more importantly, theory-based and
empirically proven solutions for each of these overload prob-
lems are offered. At the conclusion of their article, Mayer and
Moreno suggest that techniques for measuring cognitive load
are one of the most important issues that need to be addressed
by CLT if it is to continue to provide a robust framework for
instructional design. The last two articles, by considering this
vital methodological issue, provide beacons to the future.
Brünken, Plass, and Leutner introduce a dual-task ap-
proach to the measurement of cognitive load in multimedia
learning as a promising alternative to existing methods. They
argue that learners’ performance on a visual secondary reac-
tion time task can be used as a direct measure of the cognitive
loadinducedbymultimediainstruction.Theysummarizetwo
experimentsthatreproducedthemodalityeffectintwodiffer-
ent multimedia learning environments as a cognitive load ef-
fect, thereby demonstrating the feasibility of the dual-task
approach. This approach may provide a viable alternative to
the most commonly used measure of cognitive load, subjec-
tive task ratings.
The final article discusses the conceptual and practical
issues associated with cognitive load measures. Paas,
Tuovinen, Tabbers, and Van Gerven provide an overview
of the different operationalizations of cognitive load and
their advantages and disadvantages. Because a valid mea-
surement of cognitive load is essential to the endeavor to
further advance the empirical basis of cognitive load theory,
their review of recent developments of cognitive load mea-
surement is both important and timely. Finally, Paas et al.
point out that assessing cognitive load is also helpful in the
online adaptation of learning tasks in computer-based envi-
ronments.
In its ability to generate a large range of novel, the-
ory-based instructional design procedures, CLT is
unique. Furthermore, because the ability of any scientific
theory to generate applications tends to validate the origi-
nal theory, the existence of the applications generated by
CLT validates not only CLT but also many of the con-
structsofcognitive psychology, such as schema construc-
tion and the distinction between working and long-term
memory. The articles in this special issue demonstrate
that CLT is continuing its role of using cognitive psychol-
ogy principles to generate novel instructional design pro-
cedures.
4PAAS, RENKL, SWELLER
Downloaded by [Universitaetsbibliothek Freiburg] at 02:44 11 August 2015
