See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/238566346

Managing cognitive load in ICT-based learning

Article · October 2009

CITATIONS READS
12 552

1 author:

Slava Kalyuga
UNSW Sydney
153 PUBLICATIONS   12,338 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Relations between example-based learning and PBL: from the perspective of Element Interactivity View project

Rethinking Cognitive Load Theory View project

All content following this page was uploaded by Slava Kalyuga on 04 March 2016.

The user has requested enhancement of the downloaded file.

 Managing cognitive load in ICT-based learning

Slava Kalyuga
School of Education, University of New South Wales
Sydney, NSW 2052, Australia

Thus, the extensive knowledge base reduces

ABSTRACT working memory limitations by allowing experts to
process information more efficiently. Available
The paper describes theory- and research-based knowledge structures and associated cognitive
principles and design guidelines for handling characteristics may significantly change the
cognitive load in e-learning environments by effectiveness of various instructional methods.
adapting instructional methods, and presentation Therefore, in order to be efficient, ICT-based
formats to levels of learner prior knowledge in a learning formats and methods need to be tailored to
task domain. The suggested approaches and cognitive characteristics of learners.
techniques are based on contemporary knowledge of
human cognitive architecture, cognitive load theory,
and, most importantly, on extensive empirical CHALLENGES
studies of the instructional consequences of learner
prior knowledge (expertise reversal effect) in Most ICT-based learning materials continue to be
controlled experimental conditions. designed in a fixed way with novice learners as
assumed intended audience. However, recent studies
Keywords: Cognitive Load, Working Memory, of the expertise reversal effect (see [1], [2] for
Learner Knowledge Base, Expertise Reversal Effect, overviews) have indicated that designs and
Adaptive Learning Environments techniques that are effective with novices can lose
their effectiveness and even have negative
consequences when used with more experienced
learners. The major ICT design implication of these
INTRODUCTION studies is that information presentation and design
techniques need to change as learners acquire more
The design of effective ICT-based learning expertise in a domain.
environments should take into account how the
human mind works and what are its cognitive Tailoring instruction to individual learners is a very
limitations. Mental resources we use when learning complex problem due to multiple learner
and performing different tasks are limited by the characteristics, technical, organizational and other
capacity and duration of working memory that issues. The existing developmental projects in e-
represents a major factor influencing the learning are focused mostly on technical issues of
effectiveness and efficiency of learning. If more tailoring content to learner preferences, interests,
than a few chunks of information are processed choices, history of previous on-line behavior etc.
simultaneously, working memory may become and are not based on fundamental cognitive
overloaded and inhibit learning. characteristics of learners. This paper discusses
theory- and research-based cognitive principles and
On the other hand, our long-term memory is not guidelines for managing cognitive load in ICT-
limited in capacity and duration and considerably based learning environments by adapting them to
influences the operation of working memory. It levels of learner prior knowledge and skills. The
allows us to handle many interacting elements of suggested approaches and techniques are based on
information in terms of larger units (chunks) in contemporary knowledge of human cognitive
working memory thus reducing cognitive load and architecture and extensive empirical studies.
making high-level cognitive activities possible.
This paper reviews empirical studies of the expertise
reversal effect in ICT-based learning environments books, but no specific knowledge in many
and their implications for the design of learner- professional domains.
tailored instructional systems. It starts by
introducing a general theoretical framework for the The learner domain-specific knowledge in long-
described approach followed by the review of term memory and associated levels of expertise
cognitively efficient evidence-based instructional reduce working memory limitations and guide high-
techniques, procedures, and different forms of level cognitive activities. The available knowledge
information presentations for learners with different base is considered as the most important cognitive
levels of expertise. Finally, the paper suggests characteristic that influences learning and cognitive
procedures and methods for dynamic online performance. Understanding the key role of long-
tailoring of learning tasks and information term memory knowledge base in our cognition is
presentation formats to levels of learner expertise. essential to successful management of cognitive
load in ICT-based learning.

THEORETICAL FRAMEWORK Cognitive load theory (see [5], [6] for an overview)
and closely related cognitive theory of multimedia
A contemporary model of our cognitive architecture learning (see [7], [8], [9] for recent overviews)
includes two major components: working memory consider learning design implications of the above
and long-term memory. Working memory provides human cognitive architecture. Based on
temporary storage and transformation of verbal and theoretically and empirically established
pictorial information that is currently in the focus of instructional principles, they make specific
our attention (e.g., constructing and updating mental prescriptions for managing cognitive load in
representations of a situation or task). Processing learning and instruction. These theories define
limitations of working memory influence several different types and sources of cognitive
significantly the effectiveness of performance, load: effective (e.g., intrinsic) and ineffective
particularly in complex tasks. Our working memory (extraneous) cognitive load. These types of
becomes overloaded if more than a few chunks of cognitive load are associated with different
information are processed simultaneously ([3], [4]). instructional design methods and techniques.
For a simple example, we all experience this Examples of cognitive load factors that may
cognitive overload when trying to keep in memory influence effectiveness of ICT-based learning
an unfamiliar telephone number or add two four- environments are levels of element interactivity in
digit numbers in the absence of a pen and paper. learning materials, their spatial and temporal
configurations, redundant representations of
Long-term-memory represents a large store of information, etc.
organized information with effectively unlimited
storage capacity and duration. It contains a huge The effective cognitive load is associated with
number of organized knowledge structures cognitive resources directed towards achieving
(schemas) that effectively determine our capabilities certain learning objectives. When this type of
to function successfully in complex environments. cognitive load is involved, the learner attends to the
Generally, schemas are organized knowledge learning elements, attempts to establish connections
structures that are used for mentally categorizing between them and construct a coherent mental
and representing concepts and procedures in long- representation in working memory. Because this
term memory. Most of our cognitive activities are load is essential for comprehending the material and
based on available domain- and task-specific constructing new knowledge, it is vital to maximize
knowledge base. We know what to do when buying its level within limits of working memory capacity.
things at a supermarket, eating at a restaurant, filling On the other side, the irrelevant extraneous
in a car at a gas station. We easily understand fiction cognitive load represents invested cognitive
books we read, however certainly encounter huge resources that are not essential for achieving
problems when reading specialist books in learning goals and are caused by the instructional
unfamiliar domains. This is because we have design features of specific learning tasks.
massive knowledge base for dealing with our
natural and social environment in everyday life Major sources of excessive extraneous cognitive
which is usually sufficient for understanding fiction load that may inhibit performance and learning with
multimedia applications are spatially and/or load by improving the design of presentation
temporally split elements of information that need to formats or task procedures may be critical for
be integrated for understanding; an excessive step- learning.
size and/or rate of information presentations that
introduce too many new elements of information There are different sources of cognitive load related
into working memory too fast to be organized and to different modes and modalities of ICT-based
comprehended; insufficient user support or guidance information presentations (verbal and pictorial
for lower prior knowledge learners; excessive representational modes, or auditory and visual
redundant support overlapping with available information modalities). When learners process text
knowledge structures of more experienced learners. and visuals that could not be understood in isolation,
the integration of verbal and pictorial
Based on a large number of studies ([10], [11]) representations is required. When text and pictures
within a cognitive load framework, it has been are not appropriately located or synchronized in
established that learning procedures and techniques time, integrating these referring representations may
that are beneficial for learners with low levels of increase cognitive load and inhibit learning.
prior knowledge may become ineffective for more Instructional design techniques dealing with such
knowledgeable learners, and vice versa (the split attention situations may enhance learning.
expertise reversal effect). The effect is related to Using dual-mode presentations (e.g., auditory
increased cognitive load for more knowledgeable explanations of a visual diagram) is an alternative
learners due to processing redundant for these approach to eliminate split attention. Examples of
learners instructional components. other means for dealing with potential cognitive
overload are eliminating redundant components of
The main implication of the expertise reversal effect presentations or segmenting presentations.
is the need to tailor instructional techniques and However, the instructional efficiency of different
procedures to changing levels of learner expertise in formats of information presentation depends on
a domain. In order to design adaptive procedures levels of learner expertise in specific task domains.
capable of tailoring instruction in real time, it is
necessary to have sufficiently rapid online measures The general guidelines for minimizing extraneous
of learner expertise. Such measures should also have cognitive load in ICT-based learning environments
a sufficient diagnostic power to detect different include providing learners with direct access to
levels of expertise. The idea of rapid diagnostic required knowledge base, avoiding diversion of
approach is based on evaluating knowledge learner cognitive resources on redundant and/or
structures that learners are able to activate rapidly irrelevant cognitive activities, managing step-size
and apply to a briefly presented problem situation. and rate of information presentation, and
eliminating spatial and temporal split of related
sources of information.
EVIDENCE-BASED METHODS
In the most general form, the main instructional
For efficient performance and/or learning, total implication of cognitive load theory could be
cognitive load imposed on cognitive system should expressed as the need to avoid anything that gets in
not exceed limited working memory capacity. When the way of learning. Some specific design
a learning task is characterized by a high degree of implications in respect to ICT-based learning
element interactivity relative to the learner level of include (see also [7], [9] for more details):
expertise, it may require a heavy intrinsic (effective)
cognitive load to comprehend the situation. In this • enrich on-screen text with visual
case, an additional extraneous cognitive load caused representations;
by an inappropriate design can leave insufficient
cognitive resources for efficient performance and/or • present visualizations and corresponding textual
learning because total cognitive load may exceed the explanations simultaneously rather than
learner working memory capacity. The available successively to avoid temporal split-
cognitive resources may be inadequate for attention;
sustaining the required level of total cognitive load.
Elimination or reduction of extraneous cognitive
• present related sources of information close to applied to different forms of dynamic visualizations
one another on screen (e.g., embed the text such as instructional animations, simulations, and
into the graphic, avoid covering or games. For example, continuous animations may be
separating information that must be too cognitively demanding for novice learners due
mentally integrated for learning, design to a high degree of transitivity. Less knowledgeable
space for guidance or feedback close to learners may benefit more from a set of equivalent
problem statements); static diagrams. However, animations could be
relatively more beneficial for more experiences
• avoid irrelevant graphics, stories, interesting but learners who have acquired a sufficient knowledge
irrelevant details, irrelevant sounds and base for dealing with issues of transitivity and
music, nonessential words and lengthy text; limited working memory capacity. Optimal forms of
tailoring visual dynamic representations to levels of
• use visual representations explained by audio learner expertise require selecting appropriate levels
narration rather than on-screen text; of visual dynamics.

• use animated visualizations with brief audio Interactive simulations may provide appropriate
narrations rather than on-screen textual environments for exploring hypotheses and
explanations; receiving immediate feedback, thus enhancing the
development of critical thinking and problem-
• present static or animated visualizations with solving skills. However, high levels of working
narration-only instead of duplicating the memory load could be responsible for instructional
narration with onscreen text. failures of many simulations. Many instructional
simulations and games represent purely exploratory
When designing an instructional guidance on how to learning environments with limited guidance for
use the hardware that involves material with high learners. From cognitive load perspective, random
levels of element interactivity, a self-contained search procedures that novice learners have to use in
instruction that does not require the use of the such environments may impose excessive levels of
computer or other hardware could be superior to cognitive load thus interfering with meaningful
instructional formats that involve continual learning. Optimizing levels of instructional guidance
interactions with the hardware. represent an essential means for managing cognitive
load and enhancing learning outcomes in such
environments.
Sophisticated ICT-based learning environments
include various forms of interactivity and respond
dynamically to learner actions. They involve TOWARD ADAPTIVE ICT-BASED
multiple representations, linked information LEARNING
networks, and high levels of learner control. Such
environments are expected to promote active A major instructional implication of the expertise
construction and acquisition of new knowledge. reversal effect is the need to tailor dynamically
High levels of cognitive load in interactive learning instructional techniques and procedures, levels of
environments could be caused by a large number of instructional guidance to current levels of learner
variables involved in corresponding cognitive expertise. In ICT-based instructional systems, the
processes; by uncertainty and non-linear levels of expertise may change noticeably as
relationships between these variables; and by learners develop more experience in a specific task
temporary delays. In many situations, learners have domain. Therefore, the tailoring process needs to be
to carry the burden of deciding when to use dynamic, i.e. consider learner levels of expertise in
additional instructional support (if available) and real time as they gradually change during the
what forms of support to request. While more learning sessions. Personalized adaptive
advanced learners could handle such burden, it may environments may provide learner-centered
go beyond cognitive resources available to less experiences that are specifically tailored to
experienced learners. individual learners or groups.

The cognitive load framework could be effectively A possible adaptive methodology could be based on
the empirically established interactions between this task domain. Appropriately defined cognitive
levels of learner expertise and instructional methods efficiency indicators were used for the initial
(the expertise reversal effect), and on real-time selection of optimal levels of instructional guidance,
monitoring of expertise using rapid diagnostic as well as for continuous monitoring of learner
methods. For example, completion tasks and faded progress and tailoring instruction to changing levels
worked examples could be used for providing of task-specific expertise. With both approaches,
appropriate levels of instructional support that are results indicated that learner-adapted conditions
optimal for learners with different levels of resulted in significantly better knowledge gains than
expertise. As learners acquire more experience in a non-adapted conditions. However, there were no
domain, reduced levels of guidance and more significant differences found between the two
independent exploratory-based learning could be adaptation procedures when they were used in the
involved. same study and could be meaningfully compared.

Preliminary studies have indicated that using rapid Thus, dynamically adapting task selection
dynamic performance indicators in adaptive procedures and levels of instructional guidance to
methodologies is a viable approach to the problem levels of learner task-specific expertise using rapid
of tailoring e- learning environments to levels of diagnostic methods enhanced learning outcomes and
learner task-specific expertise. The rapid diagnostic supported previous results ([12]. [13]). Despite
methods were used for optimizing levels of differences in performance assessment methods,
instructional guidance and cognitive load in several definitions of instructional efficiency, and task
adaptive learning environments in algebra and selection algorithms, learner-adapted conditions
kinematics [11]. All these environments used a were superior to non-adapted formats in all these
similar adaptive procedure. At the beginning of a studies.
session, each learner was allocated to an appropriate
level of guidance according to the outcome of the Incorporating learner control approaches into
initial rapid diagnostic test. Depending on the adaptive instruction represents alternative ways of
outcomes of the rapid diagnostic probes during dynamic tailoring of instruction to levels of learner
instruction, the learner was allowed to proceed to expertise. Shared-responsibility, advisory, and
the next stage of the session or was required to adaptive guidance models could be effectively used
repeat the same stage and then take the rapid test in adaptive multimedia learning environments. For
again. At each subsequent stage of the tutoring example, a shared control model demonstrated
session, a lower level of instructional guidance was higher learning outcomes than a fully system-
provided to learners, and a higher level of the rapid controlled condition [14]. The shared control model
diagnostic tasks was used at the end of the stage. effectively combined system- and learner-controlled
Important advantages of this approach to learner- environments by first selecting a subset of tasks
adapted learning environments are its transparency based on learner performance and cognitive load
and relative simplicity. indicators, and then presenting this subset to the
learner who made the final decision.
In some of the studies, the allocation of learners to
appropriate stages of instructional guidance was The quality of adaptive environments depends on
based on levels of task-specific expertise as the accuracy of information about levels of learner
measured by the rapid online first-step or rapid knowledge and skills in specific task domains.
verification tests. In other studies, the rapid Using traditional multiple-choice tests and tracing
measures of task-specific expertise were combined user interactions with the system may not produce
with measures of cognitive load based on subjective sufficient levels of diagnostic precision. Applying
ratings of task difficulty. Since expertise is modern artificial intelligence approaches and
associated not only with higher levels of developing sophisticated fine-grained production
performance but also with lower cognitive effort, rule-based learner models allowed a significant
combining both measures was expected to produce a increase in the precision of adaptive methodologies
better indicator of learner task-specific expertise. [15].

Critical levels of efficiency were defined for each However, implementing these methodologies
class of tasks as criteria for achieving proficiency in requires complex computational modeling
 procedures. Therefore, their application has been CA: Pfeiffer.
limited to several well defined and relatively simple [8] Mayer, R. E. (2001). Multimedia learning.
for modeling domains (e.g., programming and Cambridge, MA: Cambridge University Press.
mathematics). On the other hand, the models that [9] Mayer, R. E. (Ed.). (2005). Cambridge
are used in most adaptive hypermedia and web- handbook of multimedia learning. New York:
based environments are based on several discrete Cambridge University Press.
coarse-grained levels of learner expertise. An [10] Kalyuga, S. (2006). Instructing and testing
important advantage of the suggested rapid advanced learners: A cognitive load approach.
diagnosis-based approach to the design of learner- NY: Nova Science Publishers.
adapted environments is combining high levels of [11] Kalyuga, S. (2008). Managing cognitive load in
diagnostic precision with simplicity of adaptive multimedia learning. Hershey, PA: IGI
implementation. Global.
[12] Camp, G., Paas, F., Rikers, R., & van
Achieving higher levels of expertise is associated Merriënboer, J. J. G. (2001). Dynamic problem
with flexible performance in new situations. selection in air traffic control training: A
Extending the described approaches and techniques comparison between performance, mental effort,
to developing adaptive forms of expertise represents and mental efficiency. Computers in Human
an important direction for future research. Behavior, 17, 575-595.
[13] Salden, R. J. C. M., Paas, F., & van
Merriënboer, J. J. G. (2006a). A comparison of
REFERENCES approaches to learning task selection in the
training of complex cognitive skills. Computers
[1] Kalyuga, S. (2005). Prior knowledge principle. in Human Behavior, 22, 321-333.
In R. Mayer (Ed.), Cambridge Handbook of [14] Corbalan, G., Kester, L., & van Merriënboer, J.
Multimedia Learning (pp. 325-337). New York: J. G. (2006). Towards a personalized task
Cambridge University Press. selection model with shared instructional
[2] Kalyuga, S. (2007). Expertise reversal effect and control. Instructional Science, 34, 399-422.
its implications for learner-tailored instruction. [15] Anderson, J. R., Corbett, A. T., Fincham, J. M.,
Educational Psychology Review, 19, 509-539. Hoffman, D., & Pelletier, R. (1992). General
[3] Baddeley, A. D. (1986). Working memory. New principles for an intelligent tutoring architecture.
York: Oxford University Press. In V. Shute & W. Regian (Eds.), Cognitive
[4] Cowan, N. (2001). The magical number 4 in approaches to automated instruction (pp. 81-
short-term memory: A reconsideration of mental 106). Hillsdale, NJ: Erlbaum.
storage capacity. Behavioral and Brain .
Sciences, 24, 87-114.
[5] Clark, R. C., Nguyen, F. & Sweller, J. (2006).
Efficiency in learning: Evidence-based
guidelines to manage cognitive load. San
Francisco, CA: Wiley.
[6] Sweller, J., van Merriënboer, J., & Paas, F.
(1998). Cognitive architecture and instructional
design. Educational Psychology Review, 10,
251-296.
[7] Clark, R. C., & Mayer, R. E. (2007). E-learning
and the science of instruction. San Francisco,

View publication stats