LEARNING
Joseph Krajcik and Maria Varelas, Section Coeditors
Visual Representations in Science
Education: The Influence of Prior
Knowledge and Cognitive Load
Theory on Instructional Design
Principles
MICHELLE PATRICK COOK
School of Education, Clemson University, Clemson, SC 29634-0702, USA
Received 26 October 2005; revised 22 March 2006; accepted 18 April 2006
DOI 10.1002/sce.20164
Published online 20 June 2006 in Wiley InterScience (www.interscience.wiley.com).
ABSTRACT: Visual representations are essential for communicating ideas in the science
classroom; however, the design of such representations is not always beneficial for learners.
This paper presents instructional design considerations providing empirical evidence and
integrating theoretical concepts related to cognitive load. Learners have a limited working
memory, and instructional representations should be designed with the goal of reducing
unnecessary cognitive load. However, cognitive architecture alone is not the only factor
to be considered; individual differences, especially prior knowledge, are critical in determining what impact a visual representation will have on learners’ cognitive structures and
processes. Prior knowledge can determine the ease with which learners can perceive and
interpret visual representations in working memory. Although a long tradition of research
has compared experts and novices, more research is necessary to fully explore the expert –
C 2006 Wiley
novice continuum and maximize the potential of visual representations.
Periodicals, Inc. Sci Ed 90:1073 – 1091, 2006
Correspondence to: Michelle Patrick Cook; e-mail: [email protected]
C
2006 Wiley Periodicals, Inc.
1074
COOK
INTRODUCTION
Historically, educational research has emphasized verbal learning while interest in visual
learning has lagged behind. As the amount of information acquired through visual mediums multiplies, the ability to understand, evaluate, and produce visual representations has
become increasingly important in education (Ferk, Vrtacnik, Blejec, & Gril, 2003). Visual
representations attract attention and maintain motivation. They provide an additional way of
representing information and foster the obtainment of knowledge that students may not get
from text alone (Mayer, Bove, Bryman, Mars, & Tapangco, 1996). More specifically, visual
representations enhance information retention of associated text (Peeck, 1993), improve
problem solving, and facilitate the integration of new knowledge with prior knowledge
(Roth, Bowen, & McGinn, 1999).
Visual representations are especially critical in the communication of science concepts
(Mathewson, 1999). They provide a means for making visible phenomena that are too small,
large, fast, or slow to see with the unaided eye. Similarly, visual representations illustrate
invisible or abstract phenomena that cannot be observed or experienced directly (Buckley,
2000). In science, graphics are also used to display data, organize complex information, and
promote a shared understanding of scientific phenomena (Kozma, 2003). These graphics
are often used to present multiple relationships and processes that are difficult to describe.
Although their importance in science is well documented in the literature, little research has
been conducted on the role of visual representations in science instruction (Roth, McGinn,
& Bowen, 1998). Teachers, researchers, and instructional designers believe that visual
representations have a great deal of potential as meaning-making resources, yet in practice,
these graphics do not always live up to this potential (Roth et al., 1999).
In science education today, presentations that combine visual and verbal information are
widely used for displaying instructional material. However, educational research has been
slow to identify how to effectively design these learning materials. To create instructional
materials that make use of multiple modes of representation requires a deeper understanding
of the cognitive basis of what makes for an effective representation of scientific concepts.
A recent line of research that bases instructional design principles on learners’ cognitive
structures and processes has yielded some promising results. Specifically, the work on
cognitive load theory by Sweller and others has resulted in empirical evidence for the
design principles derived from this theory.
To comprehend the role of visual representations in science teaching and learning, we
must consider not only the way they are designed, but also the way they are interpreted by
different learners (Pozzer-Ardenghi & Roth, 2005). Students may have more difficulty understanding graphics than initially assumed (Wu, Krajcik, & Soloway, 2001). Even though
a particular graphic may be designed to be cognitively useful, it may turn out to be functionally useless unless the learner perceives the information in the intended manner. Many of the
design principles derived from cognitive load theory are not generally applicable because
of differences among learners. These individual differences, especially prior knowledge,
are critical in determining what impact visual representations and their design will have
on learners’ cognitive structures and processes. Therefore, this paper will review the major
design principles, providing empirical evidence and integrating theoretical concepts related
to cognitive architecture, and account for the role of the prior knowledge in determining the
effectiveness of these designs.
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1075
PRIOR KNOWLEDGE
Prior knowledge is an important determinant of learning (Johnson & Lawson, 1998)
and has been studied extensively in science education. From misconception research, there
is widespread agreement that learners construct concepts from prior knowledge (Novak,
1990). However, prior knowledge not only influences subsequent conceptual learning, but
also influences perception and attention. Therefore, variations in how learners interpret
visual representations are also largely due to their existing knowledge. Learners use prior
knowledge to select relevant information from graphics, add information from their prior
knowledge, and ultimately, develop a mental model (Braune & Foshay, 1983).
A long tradition of research has compared differences in cognitive structures and processes of experts and novices (Chi, Feltovich, & Glaser, 1981; Chi, Glaser, & Rees, 1982).
Although many of these studies focused predominately on text representations and problem
solving, their findings are applicable to visual representations as well. Novices tend to have
fragmented knowledge, where pieces of information are only weakly connected (diSessa,
2004). Because they lack coherent and integrated existing knowledge, novice understanding of visual representations tends to be constrained to surface features. For example, in
most representations of DNA replication, color is used to distinguish the original DNA
strand from the new one. In a study by Patrick and her colleagues (2005), novice learners
attended to the color difference as evidenced by eye tracking and interview responses, but
were unable to interpret meaning from it. These novices attended to the salient features
and recognized them as relevant, but did not have the cognitive resources to explore the
underlying themes of DNA replication. In addition, since their mental models do not go
beyond the perceptual level of processing, novices are not able to easily coordinate features
within and across multiple representations to develop an understanding of the underlying
concepts (Kozma, 2003).
On the other hand, experts have more domain knowledge and are able to understand
the important core principles represented by a graphic (Chi et al., 1981). In other words,
they concentrate more on the information which is relevant for constructing of an effective
mental model (Schnotz, Picard, & Hron, 1993). The attention given to relevant information
seen by experts occurs because they possess a large number of schemas specific to the
domain. Even when they are exposed to novel information, experts are able to use relevant
prior knowledge as a starting point for interpretation (Larkin, McDermott, Simon, & Simon,
1980). For example, in a study investigating expert and novice differences in interpreting
a novel mechanics situation with weights and pulleys, Larkin (1983) found that novices
search for superficial physical features, like the presence of a rope, to interpret the problem.
Novices focused on functional relations (what direction the weights were being pulled),
whereas experts sought out geometrical relations, such as the presence of a pivot point, to
help simplify the problem.
Differences in the use of visual representations by experts and novices can be linked
to cognitive architecture. Information-processing theories assume that individuals have a
limited working memory, and when overloaded, learning will not take place. Primarily, it
is the prior knowledge of the learner that determines how much information can be held
simultaneously in working memory. One particular information-processing theory in which
prior knowledge can be easily integrated into its conceptual framework is the cognitive load
theory.
Science Education DOI 10.1002/sce
1076
COOK
COGNITIVE LOAD THEORY
Cognitive load theory provides a theoretical foundation for designing instructional materials to best enhance learning. The basic premise of this theory is that learning will be
hindered if the instructional materials overwhelm a learner’s cognitive resources. Research
has indicated that several features of human cognitive architecture are especially important
in instructional design. Specifically, cognitive load theory is based on a cognitive architecture consisting of a limited working memory that interacts with an unlimited long-term
memory (Chandler & Sweller, 1992; Sweller, van Merrienboer, & Paas, 1998). The ease
with which information can be processed in working memory is a primary concern of cognitive load theory. While a low-level perceptual stream of information is constantly entering
sensory buffers in the perceptual systems of the brain, there is limited capacity to further
process this information and link it to prior knowledge.
According to this theory, the burden placed on working memory can be reduced by either
increasing its capacity or reducing its cognitive load. Working memory has two components, a visuo-spatial sketchpad and a phonological loop, that initially process visual and
verbal information independently. Two largely independent working memory processing
systems mean information load that might overwhelm one of these processing systems
can be managed when divided across two of these systems. Therefore, using more than
one presentation modality can increase the capacity of working memory (Kirschner, 2002).
Mayer and his colleagues have found this pattern in numerous studies (Mayer, 2001). In
one example, students listening to an explanation of bicycle tire pump operation performed
moderately well on retention tests and poorly on transfer tests; students viewing an animation on the same topic performed poorly on retention and transfer tests. On the other hand,
students viewing the narrative animation performed well on both retention and transfer tests
(Mayer & Anderson, 1991).
Another way to reduce cognitive load is by constructing cognitive schemas. According
to schema theory, knowledge is stored in long-term memory in schemas so that it is organized and accessible when needed (Chi et al., 1982). Schemas help to organize and link
relevant information together, increasing the likelihood that relevant information will later
be available for related learning tasks (Glaser, 1990). Although a schema can hold a large
amount of information, it is processed as a single unit in working memory. In this manner,
cognitive schemas reduce the burden on a working memory system limited to only a few
elements of information at one time (Kalyuga, Chandler, & Sweller, 1999; Kirschner, 2002).
In addition, the load on working memory can be decreased by automating rules. Schemas
vary in their degree of automation; however when they can be processed unconsciously in
working memory, capacity can be freed (Sweller et al., 1998).
Prior knowledge can easily be integrated into schema theory. High levels of existing
knowledge imply that schemas have previously been constructed and can be retrieved easily.
These schemas help select what information will be attended to and link it to existing schema
(Valcke, 2002). Therefore, having more available schemas will reduce the cognitive load
placed on working memory. If there are few schemas available to process the information,
which is a typical scenario for novices, working memory is more likely to be overloaded.
For example, when experts and novices were asked to complete a sorting task involving
representations of chemical equations, coordinate graphs, animations at the molecular level,
and video segments of wet laboratory experiments, experts were better able to sort the
representations into larger, more meaningful clusters that were organized conceptually
(Kozma & Russell, 1997). On the other hand, novices, without the more abundant, better
organized schema, face cognitive load issues and their interpretations are constrained to
surface structures. Instead of creating conceptually meaningful clusters such as “gas laws”
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1077
or “collision theory,” these students organized these representations into groups such as
“graphs of concentrations” or “molecules moving about.”
Instruction with visual representations can burden the limited capacity of working memory. Working memory may be affected by the inherent nature of the material and the manner
in which it is presented (Kirschner, 2002). When the intention of instruction is learner understanding, it is assumed that the subject material will have a high-intrinsic load. Intrinsic load
is imposed by characteristics inherent in the subject matter, such as element interactivity.
Elements that cannot be isolated and must be learned simultaneously in working memory
will result in high-intrinsic load. If the material is low in element interactivity, the individual
elements can be learned easily without imposing a heavy load on working memory.
The amount of pre-existing knowledge a learner has will vary the level of intrinsic load
(Sweller et al., 1998). With expertise, interacting elements that would otherwise overwhelm
working memory can be incorporated into a schema that acts as a single element in working
memory. For example, a graphic representing the diffusion of starch, iodine, and glucose
across dialysis tubing is likely to be highly interactive. Typically, this kind of graphic represents diffusion at both the molecular and macroscopic levels with multiple representations
to indicate change occurring over time. In order to understand such a graphic, students will
need to simultaneously take into account the following elements: the relative amounts of
glucose, starch, iodine, and water molecules at each time period, concentration gradient,
changes in color at the macroscopic level, and changes in the size of the dialysis tubing at
the macroscopic level. While experts are able to simultaneously process these elements, for
novices, these elements exceed working memory capacity.
Any available working memory resources remaining after dealing with intrinsic cognitive
load can be allocated to deal with extraneous and germane cognitive load. Both extraneous
and germane cognitive load can be altered by instructional design. Germane cognitive load is
imposed by information and activities that contribute to the process of schema construction
and automation. Extraneous load is the effort required to process poor instructional designs
(Kirschner, 2002). Any cognitive resources devoted to processing extraneous material not
germane to the learning task at hand consumes working memory resources and decreases
the capacity for learning (Kalyuga et al., 1999). Therefore, well-designed visual representations should seek to decrease extraneous load while increasing germane load (Paas & van
Merrienboer, 1993), so that the potential of the instructional situation is maximized.
Understanding the relationship among the three types of cognitive load is especially
important. When the material imposes a low-intrinsic load due to the expertise of the learner
or the ability of the material to be processed in smaller chunks, the quality of instructional
design is less likely to have an impact because there is enough memory space remaining
to compensate for poor design (Sweller et al., 1998). However, the goal of instruction
with visual representations should not be to minimize the level of total cognitive load. In
fact, research has shown that performance can degrade when cognitive load is excessively
low (Paas, Renkl, & Sweller, 2004). Graphics that appear simple in content and design
can easily be underevaluated by the learner (Weidenmann, 1989). In this scenario, where
extraneous load is not an issue, an appropriate goal would be to encourage germane cognitive
load.
On the other hand, excessively high loads can also impede learning. Cognitive load theory
is primarily concerned with controlling the high load associated with complex information
and tasks to facilitate learning (Paas et al., 2004). In these situations, it is important to
consider the source of the load. When the intrinsic load inherent in the nature of the subject
material is high, schema formation will require more effort. Therefore, it will likely be
necessary to reduce extraneous cognitive load in order to reduce the total cognitive load
to more manageable levels. When extraneous load is reduced, more resources are free for
Science Education DOI 10.1002/sce
1078
COOK
germane load. As a consequence of acquiring and automating schema, intrinsic load is in
turn reduced.
Prior knowledge has a role in categorizing information and tasks as imposing intrinsic,
extraneous, or germane load. As previously suggested, a learner’s level of expertise can
determine the ease with which interacting elements can be processed simultaneously in
working memory without cognitive overload. In general, the less prior knowledge a learner
has, the more prone he or she is to cognitive overload. However, experts are not immune
to the effects of cognitive load. For example, cognitive load that is necessary for schema
construction for a novice learner may extraneous for an expert.
INSTRUCTIONAL DESIGN CONSIDERATIONS
Because working memory load is an important consideration for experts and novices alike,
the cognitive load theory has lead to the development of a number of instructional design
guidelines. These design principles intend to facilitate schema construction and automation
by reducing working memory load. The following design principles will be presented
along with an explanation of when and how each should be used: multiple representations,
dual-mode presentations, split-attention material, narration, redundant material, animation,
material with interacting elements, and instructional guidance.
Multiple Representations
In the typical instructional design, learners are faced with multiple representations of
information in the visual modality. Multiple graphics are commonly used in chemistry,
where it is important to understand the differences between symbolic, microscopic, and
macroscopic representations. Multiple representations may serve to complement one another with regard to information or processes, to constrain the interpretation of one another,
or to construct new connections between one another (Tsui & Treagust, 2003). They require
coherence formation; learners must create referential connections between corresponding
features of different representations. Experts are able to coordinate features within and
across multiple representations and develop an understanding of underlying concepts. They
are also better able to transform representations or provide an equivalent representation for
a given graphic (Kozma 2003).
For most novice learners, coordinating the representations is difficult. Novices expend
much of their cognitive resources interpreting the graphic and are left with few resources to
link the representations. Typically, novice learners do not make use of multiple representations, usually relying on a familiar or simple one. If switching between representations
occurs, most often the learner is having difficulty understanding the representations utilized
(Seufert, 2003). Even when novices attempt to interconnect representations, they often
concentrate on surface features with no awareness of the underlying relevant features. For
these students, translating and transforming representations is difficult, because it requires
underlying knowledge about a concept.
Some instructional tools have been designed to help students more effectively coordinate multiple representations to construct deeper understanding. For example, eChem is a
computer-based visualizing tool that allows students to build molecular models and view
multiple representations simultaneously (Wu et al., 2001). Explicitly linking representations, in this case symbolic and microscopic representations, and allowing them to be viewed
simultaneously reduced cognitive load. If learners are required to hold a representation in
their working memory while they search for corresponding features in another representation, resources will quickly become overloaded. However, by making information explicit,
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1079
learners were more likely to create referential connections between the representations. A
rotation feature also helped the learners transform between representations. Externalizing
the mental rotation process helped free cognitive resources to be used in learning.
Dual-Mode Effect
Instructional presentations that take advantage of both visual and verbal (in the form of
text or audio) modalities are more beneficial than presentations that rely on either visual
or verbal information alone. The facilitation of graphics on learning from text can be
explained by Paivio’s dual coding theory. According to dual coding theory, visual and
verbal information are processed in independent subsystems of working memory (Paivio,
1986). The visuo-spatial sketchpad takes the visual input and ultimately creates a visual
mental model; the phonological loop takes the verbal input and ultimately creates a verbal
mental model. The two different kinds of mental models are finally mapped onto each other
(Mayer, 2003). By using the capacity of both systems, more information can be processed
than would otherwise be possible with only one of those systems. Therefore, learning from
text with graphics will be richer than learning from text alone or graphics alone.
Many studies have confirmed the benefits of dual-mode presentations (Levie & Lentz,
1982; Levin, Anglin, & Carney, 1987). In one study, dual-mode instructions presented
to electrical engineering students were superior to visual-based instruction (Tindall-Ford,
Chandler, & Sweller, 1997). In another study, students who received corresponding graphics
and narration performed better on transfer tests than students who received only narration
(Moreno & Mayer, 1999). However, in a more recent study, Schnotz and Bannert (2003)
indicated that presenting graphics may not always be beneficial for the acquisition of knowledge from text-based materials. They suggest that pictures can also have negative effects
and may interfere with mental model construction.
Split-Attention Effect
Although dual-mode presentations have proven to be effective, it is important that the
complementary visual and verbal information is presented in a way that best facilitates student learning. When multiple sources of information are unintelligible in isolation, learners
must mentally integrate the sources before they are understood (Sweller et al., 1998). Integrating the information usually involves holding small segments of verbal information in
working memory while searching the graphic for the matching element. When the design
of the graphic does not foster the coordination of visual and verbal material, integration
can be difficult as the learner’s attention is split between the two modes of information.
This process of integration imposes a heavy extraneous cognitive load for novice learners,
especially when the material is high in element interactivity. Only after mental integration
occurs can schema acquisition begin. Because learners are assumed to have a limited working memory and more cognitive resources are required to process split-attention materials,
the resources available for learning are decreased (Kalyuga, Chandler & Sweller, 2000). For
example, findings in a study involving text and graphic representations of bicycle tire pump
operation suggested that because some subjects were unable to make referential linkages
between the two representations after using most of their cognitive resources on spatially
linking them together (Mayer & Sims, 1994).
It is possible to improve the instructional design of split-attention materials and decrease
extraneous load by facilitating the mental integration of disparate sources of information.
One way to reduce the cognitive load inducing search for a graphical element referenced
in the verbal information is to present related material contiguously in space and time
Science Education DOI 10.1002/sce
1080
COOK
(Wu & Shah, 2004). When material is presented contiguously in time and space, learners
are better able to form associations between visual and verbal material (Chandler & Sweller,
1992). For example, students performed better on transfer tests when textual information on
lightning formation was integrated into explanative drawings than when text and drawings
were presented sequentially or when text and drawings were presented simultaneously in
time but physically spaced apart (Mayer, Steinhoff, Bower, & Mars, 1995). In another study,
students exposed to geographic materials with text integrated into the graphic performed
better on test items than students exposed to materials that were not physically integrated
(Purnell, Solman, & Sweller, 1991). Likewise, other researchers have found that graphics
and associated narration also should be temporally and spatially coordinated to facilitate
integration (Mayer & Anderson, 1992). For example, presenting narration before or after
animations on the human respiratory system was shown to impede learning; students receiving concurrent presentations outperformed students receiving sequential presentations
on transfer tests (Mayer & Sims, 1994).
Another way to reduce search time is to color code related graphical and textual elements.
Research on computer-based instructions in electrical engineering indicates that by coloring
the elements of the graphic the same color as the referential text, students were better
able to integrate the information presented in two modalities. Mental load rating scales
indicated that this alternative to split-attention designs was effective due to reductions in
cognitive load (Kalyuga et al., 1999). In a study involving narration, as students were
listening to a narration, flashing was used to indicate where on the graphic the narration was
referring. This technique can help students coordinate visual and verbal material requiring
extensive search that they might not otherwise have the cognitive resources to coordinate.
Overall, the research literature suggests that split-format designs should be avoided to
prevent cognitive load problems. Unless the learner has adequate prior knowledge or the
content material does not impose a heavy intrinsic load, the additional processing capacity
provided with dual-mode presentations will occur only if cognitive resources do not have to
be devoted to extensive search processes associated with coordinating graphical and textual
information (Jeung, Chandler, & Sweller, 1997). Therefore, to obtain the benefits of dualmode presentations, split-attention designs should be avoided by physically and temporally
integrating the material.
Modality Effect
Modality may be another way to reduce unnecessary cognitive load resulting from splitattention materials. Under split-attention conditions, when two sources are unintelligible in
isolation, presenting verbal information through narration rather than text is advantageous.
Because text is initially processed in the visual subsystem of working memory, it competes
with the graphic for visual attention. Although the words may eventually be translated into
sounds in verbal working memory, this initial competition decreases the likelihood that the
learner will attend to the relevant elements of the graphic and text (Mayer, Heiser, & Lonn,
2001). Typically when auditory information is presented in the verbal subsystem, the load
on working memory is reduced (due to an increase in working memory capacity) allowing
more resources to be devoted to learning.
However, audio presentations may not always be effective, especially if they are too
lengthy or too complex. Spoken texts are linear in nature, preventing students from jumping
back and forth as they could with written text. Also, auditory text may not be permanent
(if it cannot be played again) and if the presentation is too quick and too complex, it
could exceed the capacity of working memory. Even when presentations are learner paced,
research has shown that replacing written text with narration may not enhance learning
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1081
(Tabbers, Martens, & van Merrienboer, 2004). However, it is likely that the lack of improvement in learning in self-paced instruction with narration is due to the visual-only
learners also having more time to integrate the text and graphics. When the instruction
is system paced, the benefits of narration over written text are clear. In addition, audio
presentations may become redundant when presented to more experienced learners. Some
researchers have found that the relative advantage of spoken text decreases as students gain
more experience (Kalyuga et al., 2000).
Most of the empirical evidence supporting the superiority of audiovisual presentations
relates to knowledge acquisition. For example, students viewing an audiovisual presentation
on soldering theory outperformed students receiving a visual-only presentation (Kalyuga
et al., 1999). In addition, a series of studies performed by Mayer and his colleagues (Mayer &
Anderson, 1991; Mayer & Moreno, 2002) suggested that students receiving animations
with narration (on lighting formation, car braking systems, or bicycle tire pump operation) outperformed students viewing the same animation with on-screen text in recall and
problem-solving transfer tests.
While most studies have used learning outcomes as measures of cognitive load, one study
in particular has focused on providing insight into cognitive processes. Brunken and his
colleagues (2004) investigated whether audiovisual presentations would lead to higher cognitive load placed on the verbal subsystem than visual-only (graphic and text) presentations.
In their first study, a secondary, visual-monitoring task (used as a measure of cognitive load)
was performed simultaneously to the primary task. The learners performed better on the
secondary task when the primary task was presented in an audiovisual format. To confirm
their results, an auditory secondary task was used in their next study. As they predicted,
students viewing the visual-only presentation outperformed the audiovisual group on the
auditory task. Their results suggest that in audiovisual presentations, load is transferred
from an overloaded visual subsystem to a relatively underloaded auditory system.
Redundancy Effect
Integrated dual-mode presentations have been suggested when visual and verbal information is unintelligible in isolation; otherwise, the omission of redundant material is preferable.
For example, when a graphic can be understood by itself, adding explanatory verbal information is redundant. By presenting the same information in two different modes, the student
is required to process the learning material twice, using up unnecessary cognitive resources
and decreasing learning. For example, in a graphic with text and graphical representations
of heart function, the text material did nothing more but to restate the parts of the heart
and path of blood flow (Chandler & Sweller, 1991). Since the information was redundantly
presented, in this case an integrated presentation of text and graphic was not ideal. Likewise,
offering graphics with text in paper form and on a computer results in less learning than if
the learning material is only offered in paper form (Sweller & Chandler, 1994). The results
of these studies contradict earlier research conducted by Rieber (1990) which suggests that
instructional material is more effective when it is presented redundantly in graphics and
text.
Numerous studies have also investigated presenting redundant verbal information in the
form of written and spoken text. Mayer and his colleagues found that students viewing
narrated animations of lightning formation performed better on transfer test than students
viewing narrated animation with on-screen text (Mayer et al., 2001). In addition, Moreno
and Mayer (1999) found that when using an environmental science game, students receiving
narration performed better than students receiving narration with on-screen text. Similar
findings have resulted from research on soldering theory (Kalyuga et al., 1999). Research
Science Education DOI 10.1002/sce
1082
COOK
has suggested that when verbal information is presented in both written and spoken text,
the visual subsystem of working memory is overloaded. When this happens, learners are
less likely to be able to carry out the cognitive processes required for schema formation
since they have to split their attention between two visual sources, the graphic and the text.
When different versions of the same material are presented, an unnecessary burden
is placed on working memory. However, this determination depends on both the nature
of the material and the level of expertise. Kalyuga, Chandler, and Sweller (1998) found
that novice electrical trainees could not understand a wiring diagram in isolation, whereas
more experienced students could. They suggested that the verbal material was redundant
and could be eliminated for the experts. Other researchers have found that additions to
biological text intended to increase learner understanding benefited novices while impeding the learning of experts (McNamara, Kintsch, Songer, & Kintsch, 1996). These results
support the expertise reversal effect. In general, as expertise increase, material that is essential for novices may become redundant for experts, imposing extraneous load (Sweller,
2004).
Animations
A growing trend exists to use highly illustrated materials like animations (Lowe, 2003).
Animations, by their nature, present multiple images over time which allow viewers to perceive dynamic phenomena much as they would in the physical world. These dynamic
images can be grounded in events that can be seen with the unaided human eye, but
in the science classroom—more often than not—they represent phenomena that unfolds
very slowly or very rapidly or is an abstract concept not directly connected to physical objects. While the use of animations has been perceived favorably by learners, research has not shown a clear advantage for the using animations for conceptual learning
in all instances (Hegarty, 2004). Therefore, animations will be most powerful in representing concepts where change over time is a critical component (Ainsworth & VanLebeke,
2004).
Research has shown animations to be helpful in the visualization of dynamic processes,
such as motion and trajectory (Rieber, 1990). In numerous studies, animations have been
shown to improve student’s understanding of chemical processes at the molecular level when
compared to static images (Sanger, Brecheisen, & Hynek, 2001; Sanger & Greenbowe, 2000;
Yang, Andre, & Greenbowe, 2003). For example, students who watched two animations
before performing laboratory experiments were less likely to choose responses suggesting the particle motion stops after equilibrium is reached (Sanger et al., 2001). Likewise,
students viewing animations illustrating chemical reactions performed better on content
tests compared with students viewing static diagrams (Yang et al., 2003). Also regarding
motion concepts, Rieber (1990) found that students viewing animations on Newton’s laws
of motion were better able to retain, retrieve, and apply the content material.
When the goal is to help students understand how processes work, it is important to show
how the features of the graphic interact and interrelate (Mayer & Gallini, 1990). These
animations allow students to build mental representations of unseen processes and facilitate
science learning (Yang et al., 2003). It has been suggested that animations, by reducing
the level of abstraction of spatial and temporal information, provide a complete model
for building a mental representation (Mayer & Gallini, 1990) and can reduce the load of
cognitive processing. On the other hand, this model must be inferred by students viewing
static graphics.
Although most research findings suggest that animations are better than static graphics
in representing concepts involving change over time, the results from animation studies
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1083
have been inconsistent. In some cases, animations may not provide benefits over static
graphics (Wu & Shah, 2004). As with static graphics, cognitive resources are required
to extract relevant information, incorporate it into schema, and construct mental models.
However, animations are often complex, transitory, and fast paced. These characteristics
can hinder novice learning, further increasing the cognitive load required to process the
graphic (Tversky, Morrison, & Betrancourt, 2002).
For example, in a study by Schnotz and Grzondzeil (1996), animations of the earth’s
rotation were not shown to have a positive effect on understanding. Learners who used
animations performed lower on content questions than students who viewed static images.
Schnotz and Grzondzeil suggested that a deeper processing of the material could not take
place because of the fleeting display. Likewise, in a study by Narayanan and Hegarty
(2002), animated graphics did not improve conceptual understanding compared with their
static equivalents. When a learner’s cognitive resources are overwhelmed, simple graphics
with less detail may be more effective than realistic ones (Dwyer, 1969) provided that the
learner is able to extract relevant information. They are more likely to prevent the learner
from being distracted. However, simplifying the animation is not always an appropriate
solution. Although animations tend to be more complex than static graphics, sometimes the
apparent simplicity of the animation may also prevent learning. This apparent simplicity of
dynamic information, along with the externalization of dynamic information, could reduce
the mental effort learners expend on valuable processing activities (Bodemer, Ploetzner,
Bruchmuller, & Hacker, 2005; Lowe, 2003).
Also, novices may find it difficult to extract relevant information at such a rapid pace,
therefore they may require a slower pace or successive viewings. For example, animations of
molecular drawings had to be shown successively at least three times in order for students to
have enough time to interpret the animation (Sanger et al., 2001). One way of managing the
quick pace of animations is through interactive controls. With such controls, learners could
review segments of the animations and view individual frames as static images. However,
because interactivity can sometimes go from imposing germane load to extraneous load,
the addition of controls is not always beneficial (Chandler, 2004). Without the appropriate
meta-cognitive ability, novices are unable to effectively use such controls (Lowe, 2004).
In summary, animations can provide learners with explicit dynamic information that may
not be available in static graphics, but have additional and different information processing
demands. Learners, especially those without prior knowledge, selectively process the animation and extract knowledge on the basis of perceptual salience, not thematic importance
(Lowe, 2003). Therefore, even if an animation represents dynamic information about motion and trajectory, the superiority of the animation may be limited to aspects with a high
level of perceptual salience. Features with low-perceptual salience tend to be neglected,
regardless of how important they are conceptually. Student attention must be cued to the
motion and trajectory details contained in the animation, so that they are able to extract
relevant information for interpretation (Rieber, 1990).
Element Interactivity
When the elements to be learned from the material are highly interactive and cannot
be processed in isolation without compromising understanding, the instructional material
will impose a high-intrinsic load (Sweller & Chandler, 1994). If the number of interacting
elements exceeds what can be processed by working memory simultaneously, it is unlikely
that learning will occur. However, the level of interactivity will depend of the learner’s level
of expertise. If the learner has sufficient prior knowledge, the interacting elements could
be incorporated into a schema and processed as a single unit. For novices, the elements
Science Education DOI 10.1002/sce
1084
COOK
may need individual attention before interactions between the elements can be understood.
However, if the element interactivity is high for the novice learner, cognitive resources will
be quickly overloaded and learning will not occur (Kalyuga, Ayres, Chandler, & Sweller,
2003).
High-intrinsic load can be altered by isolating the interacting elements to provide a simpler learning task. This intervention will allow novices to construct schemas to process the
highly interactive material, even though they cannot process all the elements in working
memory simultaneously. For example, Pollock and his colleagues (2002) artificially reduced
element interactivity and presented the elements in isolation in the first phase of their study.
Students were able to avoid working memory limitations and construct partial schemas for
the information presented, but full understanding of the material was compromised since
students were unable to interrelate the elements at this stage. It was not until the second
phase of instruction that students began relate the interacting elements. This two-stage instructional method was superior to a method in which interacting elements were presented
to learners during both stages. However, the superiority of this method disappeared when
used with learners with sufficient prior knowledge (Kalyuga et al., 2003). Therefore, isolating interactive elements allows novice learners to reduce working memory load; however,
it provides no additional benefits to experienced learners.
Instructional Guidance
Instructional guidance plays an important role in learning from visual representations,
particularly when instruction requires active construction of knowledge. Sometimes the free
exploration of multimedia presentations can impose heavy cognitive load. Many studies
have investigated the effects of instructional guidance provided by the learning materials.
In one study, learners exposed to an unstructured simulation on the laws of mechanics did
not perform as well as those working with a guided simulation (Rieber & Parmley, 1995).
In another study,
Ried and colleagues (2003) found that learners who receive explanations of floating
and sinking that were embedded in the animation outperformed learners who did not receive those explanations. Moreno (2004) examined the effects of providing explanatory and
corrective feedback to guide students in their comprehension of instructional materials in
botany. Her findings indicate that explanatory feedback promoted higher transfer scores and
reduced cognitive load compared to corrective feedback. In general, instructional guidance
can provide assistance to learners, allowing them to complete tasks that would otherwise
be out of reach.
The level of learner expertise determines how much instructional guidance may be warranted for an instructional task. Insufficient or excessive instructional guidance can negatively affect the learning process. Novices do not receive guidance in the form of schemas
in long-term memory to assist with learning. Therefore, instructional guidance will be more
beneficial in helping students with little prior knowledge acquire schemas and minimize
cognitive load (Sweller, 2004). On the other hand, experts may not need instructional guidance because they already possess internal guidance in the form of schemas. As with other
instructional design interventions, if the instructional guidance provides redundant information, additional cognitive resources will be required to integrate the redundant information.
To prevent cognitive overload in students with sufficient domain knowledge, instruction
should rely more on already developed schemas for guidance, rather than instructional
guidance (Kalyuga et al., 2003). Experts can use their prior knowledge to compensate for
this lack of instructional guidance (Mayer, 2003).
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1085
DISCUSSION
Presentations with visual representations are widely used for displaying learning materials, however not all of these presentations necessarily lead to better learning results. Many
theories address this issue of learning from verbal and visual displays and offer guidelines
for instructional design; however, the most influential to date is the cognitive load theory.
The cognitive load theory provides insight into the obstacles students face when learning
new material. The cognitive load theory emphasizes the limited capacity of working memory and suggests that instructional materials should be designed with the goal of reducing
unnecessary cognitive load (Purnell et al., 1991). Cognitive resources devoted to processing
extraneous material not germane to the learning task take away from the ability of the learner
to maximize the potential of the instructional representation. The design principles derived
from this theory have ample empirical evidence suggesting their superiority in facilitating
learning by reducing working memory load and encouraging schema construction.
For most learners in the science classrooms, explicitly linking multiple representations
used in instructional materials will be important. Multiple representations of information
using the same modality, or multiple representations of information in visual and verbal
modes, must be physically and/or temporally integrated so that students may extract information out of all the representations, leading to a more complete understanding. While
dual-mode presentations have been shown to be more beneficial than single-mode presentations, research has also indicated that presenting verbal information in spoken form
rather than written form will more likely increase the capacity of working memory. As for
visual information, the use of static graphics rather than animations tends to be beneficial
for enhancing learning, except when representing motion or trajectory. For all instructional
presentations, instructional guidance is likely to help learners perform at levels that would
have otherwise been impossible. Although these instructional design considerations are informative, research has indicated that when these guidelines are used with more experienced
learners, they can lose their effectiveness or even impair learning (Kalyuga et al., 2003).
Table 1 provides a summary of the design principles along with a rationale for use based
upon cognitive research.
Recently, strong evidence has emerged suggesting that the effectiveness of these design
principles is dependent on several learner characteristics. Research has suggested specific
characteristics, such as spatial ability, cognitive ability, and developmental level, are important determinants of understanding from instructional design; however, prior knowledge
seems to be the most influential. A learner’s expertise is an important factor mediating the
relation between cognitive architecture and learning. Students are not cognitively passive
as they approach learning; they construct an understanding from visual representations on
the foundation of their existing knowledge. Learners engage in numerous active cognitive processes when viewing a graphic, and each process is influenced by learners’ prior
knowledge.
Specifically, Mayer has developed the cognitive theory of multimedia learning, which
requires learners to engage in five active cognitive processes while viewing and interpreting
explanative graphics (Mayer & Anderson, 1991; Mayer & Moreno, 2003). These cognitive
processes, which allow learners to construct a mental representation of presented material,
involve selecting the relevant words and images for processing in working memory, organizing the relevant words and images to build coherent verbal and visual mental models,
and integrating the mental models with one another and with relevant prior knowledge.
In this theory, a learner’s prior knowledge is one of the strongest factors influencing the
interpretation of representations. Mayer argues that relevant prior knowledge facilitates the
referential connections made between the visual and verbal mental models. However, prior
Science Education DOI 10.1002/sce
1086
COOK
TABLE 1
Overview of Instructional Design Considerations and Rationale for Use
Instructional Design Consideration
Multiple representations using the visual
modality or visual/verbal modalities should be
explicitly linked in time and space
Dual-mode presentations are typically
advantageous over single-mode presentations
Present verbal information through narration
rather than written text
Animations have tremendous potential when
representing dynamic phenomena, but in
many cases the benefits of animation are not
realized
Highly interactive elements should be presented
in isolation for novice learners
Instructional guidance can help learners actively
construct an understanding of the concepts
Redundant information should be avoided
Rationale
Reduces cognitive load required to
integrate multiple sources of
information
Increases the capacity of working
memory
Eliminates competition for visual
attention
Animations are often complex and fast
paced, requiring more cognitive
resources for processing
Reduces working memory load,
eliminating the need for
simultaneous processing of the
elements
Minimizes the cognitive load required
to construct schemas
Avoids using cognitive resources for
processing information multiple
times, especially for learners with
more prior knowledge
knowledge also influences what visual and verbal representations will be selected for processing in working memory and how those representations will be organized into coherent
visual and verbal mental models.
Schnotz and Bannert’s integrative model of text and picture comprehension explains the
differences found between experts and novices by proving insight into how learners exposed
to a new concept will process the information found in the visual representation (Schnotz,
2002; Schnotz & Bannert, 2003). According to this model, learners comprehend graphics
by constructing multiple mental representations. Initially, the graphic is processed at a
perceptual level and the learner creates a visual mental representation of surface features.
Then, the learners begin to construct a mental model in which the surface level interpretation
is linked to a higher level conceptual understanding of the material. In this mental model,
irrelevant perceptual information is omitted and the abstracted material is linked to prior
knowledge related to the subject.
Students with little prior knowledge often fall short of creating more abstract mental models; the only internal representation constructed by these learners remains at the perceptual
level. With little prior knowledge, novices focus on surface features of graphics when interpreting the concepts represented (Seufert, 2003). However, in many cases, the most salient
features of a graphic (such as color, shape, labels, etc.) may not be the most relevant or
important for building an understanding of the representation. Because students with little
prior knowledge have difficulty distinguishing between relevant and irrelevant information,
visualizations can easily confuse these learners (Hegarty, Carpenter, & Just, 1991; Linn,
2003). Reliance on surface-level interpretations may constrain the understanding of novices
in other ways as well. Unlike experts, when working with multiple representations, novices
are unable to notice that representations with different surface structures can present the
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1087
same underlying theme. Likewise, students with little prior knowledge are less able to provide an equivalent representation for a given concept (Kozma, 2003; Kozma & Russell,
1997). Visual representations can be powerful tools in the science classroom, but unfortunately, novice learners must surmount more obstacles in order for these graphics to facilitate
their understanding of concepts (Perkins & Unger, 1994).
Experts are able to go beyond superficial features because they have an abundance of
relevant prior knowledge. This prior knowledge has been stored in long-term memory
in schemas so that it is organized and easily accessible when needed (Chi et al., 1982).
Although a schema can hold a large amount of information, it is processed as a single unit
in working memory. Therefore, when relevant prior knowledge is integrated into working
memory to facilitate connections between visual and verbal mental models, it is less likely
to overburden working memory (Kirschner, 2002). Because experts have well-developed
schemas, they attend to different information than novices (Chi et al., 1981). Experts link
their initial visual and verbal representations to underlying principles of the content, and
develop a more comprehensive mental model (Snyder, 2000).
CONCLUSIONS
Although the use of visual representations has shown much promise in instructional settings, there is also a risk that they will prove to be an impediment to learning (Mayer, 2001).
Without proper design and consideration for individual differences among learners, visual
representations may be no better than textual information alone or, at worst, may actually interfere with learning or lead to misconceptions about scientific phenomena. Current
cognitive research has provided guidance in the development of heuristics for the design
of visual representations. Specifically, cognitive load theory provides insight into how the
brain processes visual information, and in turn, provides an explanation of how these images
can be designed to optimize their effectiveness in an instructional context to meet the needs
of diverse learners (Linn, 2003).
Cognitive load theory is consistent with current knowledge regarding memory, learning, and problem solving. Regardless of the prior knowledge of the leaner, heuristics for
instructional design guided by cognitive load theory stem from the need to maximize the
cognitive resources available to be used in the learning. It is interesting to note that many of
the instructional design effects generated by cognitive load theory are contrary to traditional
practice. For example, when instructional materials involve text and graphics, it is standard
practice for those representations to be presented separately in a neat and organized manner rather than integrated. Likewise, most instructional material with text and graphics is
presented visually to ensure consistency, rather than in a mixed media format involving
graphics and narration. Finally, instructional material is often presented in several different
ways at the same time to foster comprehension rather than avoiding redundant information.
These instances, where traditional practice is contrary to the guidelines derived by cognitive
load theory, only attest to the strength of this theory.
Although cognitive research on graphics has proven useful, some researchers believe it
is limited since it does not take into account the sociological perspective focused on how
people use visual representations (Roth, Pozzer-Ardenghi, & Han, 2005). For example, Roth
and his colleagues recognize the role of cognitive architecture and mental representations
when learning from graphics, but would rather focus their attention on the practices of
competent learners and investigate how they share that information with others. They suggest
interpretation of graphics is a social practice shared in community of learners and learning
to use graphics involves acquiring practices. While this framework should not be as an
alternative to cognitive frameworks, investigating the social perspective can offer a more
Science Education DOI 10.1002/sce
1088
COOK
complete understanding of how students interpret graphics. This approach can be especially
helpful when the poor performance of student in interpreting graphics cannot be easily
explained by deficits in cognitive processing.
Although cognitive load theory provides a framework for instructional design considerations (Pass et al., 2004), several limitations exist which suggest areas of future research.
A vast majority of the research on design considerations guided by cognitive load theory
has been conducted in laboratory settings with college students. More work is needed to
validate instructional design heuristics in the science classroom with diverse, younger populations. In addition, much of the experimental evidence supporting the design heuristics
is based upon the amount of knowledge acquired from the instructional materials. Recall
and transfer indicate two different levels of processing of the instructional materials (Elliot,
McGregor, & Gable, 1999), with recall indicating surface processing or rote memorization
of information (Zimmerman & Pons, 1986) and transfer indicating deep processing or integration of the new information with prior knowledge and experience (Weinstein & Mayer,
1986). In many studies, there is a lack of empirical evidence to back up transfer assumptions
(Simpson, 1994). Even when transfer is addressed, as in many of Mayer’s studies, measures
emphasize student ability to access learned material for problem solving more than their
ability to apply what was learned to a new task. In addition, since a central assumption
underlying these heuristics is that they reduce cognitive load, it would be useful to include
more direct measures of cognitive load in future research (Mayer, 2005).
While most researchers recognize the importance of individual differences, the role of
prior knowledge in cognitive load theory and the resulting principles for instructional design
is still in its infancy. While most of the research relates to differences between novices and
experts, in reality, the situation is not that simple. Students cannot merely be categorized
as experts or novices; instead they represent a continuum of prior knowledge. Although
research has shown that students with less prior knowledge are usually more prone to
cognitive load influence, it is unclear whether more or less proficient novices differ in the
same manner as experts and novices. More research is necessary to fully explore the expert –
novice continuum and to improve the specificity of the design principles derived from the
cognitive load theory. Such research could allow for the tailoring of instructional design to
meet the needs of differentially prepared students.
REFERENCES
Ainsworth, S., & VanLabeke, N. (2004). Multiple forms of dynamic representation. Learning and Instruction, 14,
241 – 255.
Bodemer, D., Ploetzner, R., Bruchmuller, K., & Hacker, S. (2005). Supporting learning with interactive multimedia
though active integration of representations. Instructional Science, 33, 73 – 95.
Braune, R. F., & Foshay, W. R. (1983). Towards a practical model of cognitive information processing: Task
analysis and schema acquisition for complex problem-solving situations. Instructional Science, 12, 121 – 145.
Brunken, R., Plass, J. L., & Leutner, D. (2004). Assessment of cognitive load in multimedia learning with dual-task
methodology: Auditory load and modality effects. Instructional Science, 32, 115 – 132.
Buckley, B. C. (2000). Interactive multimedia and model-based learning in biology. International Journal of Science
Education, 22(9), 895 – 935.
Chandler, P. (2004). The crucial role of cognitive process in the design of dynamic visualizations. Learning and
Instruction, 14(3), 353 – 357.
Chandler, P., & Sweller, J. (1991). Cognitive load theory and the format of instruction. Cognition and Instruction,
8, 293 – 332.
Chandler, P., & Sweller, J. (1992). The split-attention effect as a factor in the design of instruction. British Journal
of Educational Psychology, 62(2), 233 – 246.
Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and representation of physics problems by
experts and novices. Cognitive Science, 5, 121 – 152.
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1089
Chi, M. T. H., Glaser, R., & Rees, E. (1982). Expertise in problem solving. In R. Sternberg (Ed.), Advances in the
psychology of human intelligence (pp. 7 – 75). Hillsdale, NJ: Erlbaum.
diSessa, A. A. (2004). Metarepresentations: Native competence and targets for instruction. Cognition and Instruction, 22(3), 293 – 331.
Dwyer, F. M. (1969). The effect of varying the amount of realistic detail in visual illustrations designed to
complement programmed instruction. Programmed Learning, 6(3), 147 – 153.
Elliot, A. J., McGregor, H. A., & Gable, S. (1999). Achievement goals, study strategies, and exam performance:
A meditational analysis. Journal of Educational Psychology, 91, 549 – 563.
Ferk, V., Vrtacnik, M., Blejec, A., & Gril, A. (2003). Students’ understanding of molecular structure representations. International Journal of Science Education, 25(10), 1227 – 1245.
Glaser, R. (1990). The reemergence of learning theory within instructional research. American Psychologist, 45,
29 – 39.
Hegarty, M. (2004). Dynamic visualizations and learning: Getting to the difficult questions. Learning and Instruction, 14, 343 – 351.
Hegarty, M., Carpenter, P. A., & Just, M. A. (1991). Diagrams in the comprehension of scientific text. In R. Barr,
M. L. Kamil, P. B. Mosenthal, & P. D. Pearson (Eds.), Handbook of reading research (Vol. 2, pp. 641 – 668).
New York: Longman.
Jeung, H. J., Chandler, P., & Sweller, J. (1997). The role of visual indicators in dual sensory mode instruction.
Educational Psychology, 17(3), 329 – 343.
Johnson, M. A., & Lawson, A. E. (1998). What are the relative effects of reasoning ability and prior knowledge
on biology achievement in expository and inquiry classes? Journal of Research in Science Teaching, 35(1),
89 – 103.
Kalyuga, S., Ayres, P. L., Chandler, P. A., & Sweller, J. (2003). The expertise reversal effect. Educational
Psychologist, 38(5), 23 – 31.
Kalyuga, S., Chandler, P., & Sweller, J. (1998). Levels of expertise and instructional design. Human Factors, 40(1),
1 – 17.
Kalyuga, S., Chandler, P., & Sweller, J. (1999). Managing split-attention and redundancy in multimedia instruction.
Applied Cognitive Psychology, 13(4), 351 – 371.
Kalyuga, S., Chandler, P., & Sweller, J. (2000). Incorporating learner experience into the design of multimedia
instruction. Journal of Educational Psychology, 92(1), 126 – 136.
Kirschner, P. A. (2002). Cognitive load theory: Implications of cognitive load theory on the design of learning.
Learning and Instruction, 12(1), 1 – 10.
Kozma, R. (2003). The material features of multiple representations and their cognitive and social affordances for
science understanding. Learning and Instruction, 13(2), 205 – 226.
Kozma, R., & Russell, J. (1997). Multimedia and understanding: Expert and novice responses to different representations of chemical phenomena. Journal of Research in Science Teaching, 34(9), 949 –
968.
Larkin, J. H. (1983). The role of problem representation in physics. In D. Gentner & A. L. Stevens (Eds.), Mental
models (pp. 75 – 98). Hillsdale, NJ: Erlbaum.
Larkin, J. H., McDermott, J., Simon, D. P., & Simon, H. A. (1980). Expert and novice performance in solving
physics problems. Science, 208, 1335 – 1342.
Levie, W. H., & Lentz, R. (1982). Effects of text illustrations: A review of research. Educational Communication
and Technology Journal, 30, 195 – 232.
Levin, J. R., Anglin, G. J., & Carney, R. N. (1987). On empirically validated functions of pictures in prose. In
D. M. Willows & H. A. Houghton (Eds.), The psychology of illustration. Volume 1: Basic research (pp. 51 – 85).
New York: Springer-Verlag.
Linn, M. (2003). Technology and science education: Starting points, research programs, and trends. International
Journal of Science Education, 25(6), 727 – 758.
Lowe, R. K. (2003). Animation and learning: Selective processing of information in dynamic graphics. Learning
and Instruction, 13(2), 157 – 176.
Lowe, R. K. (2004). Interrogation of a dynamic visualization during learning. Learning and Instruction, 14(3),
257 – 274.
Mathewson, J. H. (1999). Visual-spatial thinking: An aspect of science overlooked by educators. Science Education,
83(1), 33 – 54.
Mayer, R. E. (2001). Multimedia learning. Cambridge, UK: Cambridge University Press.
Mayer, R. E. (2003). Elements of a science of e-learning. Journal of Educational Computing Research, 29(3),
297 – 313.
Mayer, R. E. (2005). The Cambridge handbook of multimedia learning. Cambridge, UK: Cambridge University
Press.
Science Education DOI 10.1002/sce
1090
COOK
Mayer, R. E., & Anderson, R. B. (1991). Animations need narrations: An experimental test of a dual-coding
hypothesis. Journal of Educational Psychology, 83(4), 484 – 490.
Mayer, R. E., & Anderson, R. B. (1992). The instructive animation: Helping students build connections between
words and pictures in multimedia learning. Journal of Educational Psychology, 84(4), 444 – 452.
Mayer, R. E., Bove, W., Bryman, A., Mars, R., & Tapangco, L. (1996). When less is more: Meaningful learning
from visual and verbal summaries of science textbook lessons. Journal of Educational Psychology, 88(1),
64 – 73.
Mayer, R. E., & Gallini, J. K. (1990). When is an illustration worth ten thousand words? Journal of Educational
Psychology, 82, 715 – 726.
Mayer, R. E., Heiser, J., & Lonn, S. (2001). Cognitive constraints on multimedia learning: When presenting more
material results in less understanding. Journal of Educational Psychology, 93(1), 2001.
Mayer, R. E., & Moreno, R. (2002). Aids to computer-based multimedia learning. Learning and Instruction, 12(1),
107 – 119.
Mayer, R. E., & Moreno, R. (2003). Nine ways to reduce cognitive load in multimedia learning. Educational
Psychologist, 38(1), 43 – 52.
Mayer, R. E., & Sims, V. K. (1994). For whom is a picture worth a thousand words? Extensions of a dual-coding
theory of multimedia learning. Journal of Educational Psychology, 86(3), 389 – 401.
Mayer, R. E., Steinhoff, K., Bower, G., & Mars, R. (1995). A generative theory of textbook design: Using annotated
illustration to foster meaningful learning of science text. Educational Technology Research and Development,
43(1), 31 – 43.
McNamara, D. S., Kintsch, E., Songer, N. B., & Kintsch, W. (1996). Are good texts always better? : Interactions
of text coherence, background knowledge, and levels of understanding in learning from text. Cognition and
Instruction, 14, 1 – 43.
Moreno, R. (2004). Animated pedagogical agents in educational technology. Educational Technology, 44(6),
23 – 30.
Moreno, R., & Mayer, R. E. (1999). Cognitive principles of multimedia learning: The role of modality and
contiguity. Journal of Educational Psychology, 91(2), 358 – 368.
Narayanan, N. H., & Hegarty, M. (2002). Multimedia design for communication of dynamic information. International Journal of Human-Computer Studies, 57, 279 – 315.
Novak, J. (1990). Concept mapping: A useful tool for science education. Journal of Research in Science Teaching,
27, 937 – 994.
Paas, F., Renkl, A., & Sweller, J. (2004). Cognitive load theory: Instructional implications of the interaction
between information structures and cognitive architecture. Instructional Science, 32, 1 – 8.
Paas, F., & van Merrienboer, J. J. G. (1993). The efficiency of instructional conditions: An approach to combine
mental effort and performance measures. Human Factors, 35(4), 737 – 743.
Paivio, A. (1986). Mental representations: A dual coding approach. Oxford, UK: Oxford University Press.
Patrick, M. D., Carter, G., & Wiebe, E. N. (2005). Visual representations of DNA replication: Middle grades
students’ perceptions and interpretations. Journal of Science Education and Technology, 14(3): 353 – 365.
Peeck, J. (1993). Increasing picture effects in learning from illustrated text. Learning and Instruction, 3, 227 – 238.
Perkins, D. N., & Unger, C. (1994). A new look in representations for mathematics and science learning. Instructional Science, 22, 1 – 37.
Pollock, E., Chandler, P., & Sweller, J. (2002). Assimilating complex information. Learning and Instruction, 12(1),
61 – 86.
Pozzer-Ardenghi, & Roth, W.-M. (2005). Making sense of photographs. Science Education, 89(2), 219 – 241.
Purnell, K. N., Solman, R. T., & Sweller, J. (1991). The effects of technical illustrations on cognitive load.
Instructional Science, 20, 443 – 462.
Rieber, L. P. (1990). Using computer animated graphics in science instruction with children. Journal of Educational
Psychology, 82(1), 135 – 140.
Rieber, L. P., & Parmley, M. W. (1995). To teach or not to teach? Comparing the use of computer-based simulations
in deductive versus inductive approaches to learning with adults in science. Journal of Educational Computing
Research, 13(4), 359 – 374.
Reid, D. J., Zhang, J., & Chen, Q. (2003). Supporting scientific discovery learning in a simulation environment.
Journal of Computer Assisted Learning, 19, 9 – 20.
Roth, W.-M., Bowen, G. M., & McGinn, M. K. (1999). Differences in graph-related practices between high
school biology textbooks and scientific ecology journals. Journal of Research in Science Teaching, 36(9), 977 –
1019.
Roth, W.-M., McGinn, M. K., & Bowen, G. M. (1998). How prepared are preservice teachers to teach scientific
inquiry? Levels of performance in scientific representation practices. Journal of Science Teacher Education,
9(1), 25 – 48.
Science Education DOI 10.1002/sce
VISUAL REPRESENTATIONS
1091
Roth, W.-M., Pozzer-Ardenghi, L., & Han, J. Y. (2005). Critical graphicacy: Understanding visual representation
practices in school science. Dordrecht: Springer.
Sanger, M. J., Brecheisen, D. M., & Hynek, B. M. (2001). Can computer animations affect college biology students’
conceptions about diffusion & osmosis? The American Biology Teacher, 63(2), 104 – 109.
Sanger, M. J., & Greenbowe, T. J. (2000). Addressing student misconceptions concerning the flow in aqueous
solutions with instruction including computer animations and conceptual change strategies. International Journal
of Science Education, 22(5), 521 – 537.
Schnotz, W. (2002). Towards an integrated view of learning from text and visual displays. Educational Psychology
Review, 14(1), 101 – 120.
Schnotz, W., & Bannert, M. (2003). Construction and interference in learning from multiple representations.
Learning and Instruction, 13(2), 141 – 156.
Schnotz, W., & Grzondziel, H. (1996). Knowledge acquisition with static and animated pictures in computer-based
learning. Paper presented at the Annual Meeting of the American Educational Research Association, New York.
Schnotz, W., Picard, E., & Hron, A. (1993). How do successful and unsuccessful learners use texts and graphics?
Learning and Instruction, 3, 181 – 199.
Seufert, T. (2003). Supporting coherence formation in learning from multiple representations. Learning and
Instruction, 13, 227 – 237.
Simpson, T. J. (1994). Message into medium: An extension of the dual coding hypothesis. Paper presented at the
26th Annual Conference of the International Visual Literacy Association, Tempe, AZ.
Snyder, J. L. (2000). An investigation of the knowledge structures of experts, intermediates and novices in physics.
International Journal of Science Education, 22, 979 – 992.
Sweller, J. (2004). Instructional design consequences of an analogy between evolution by natural selection and
human cognitive architecture. Instructional Science, 32, 9 – 31.
Sweller, J., & Chandler, P. (1994). Why some material is difficult to learn. Cognition and Instruction, 12(3),
185 – 233.
Sweller, J., van Merrienboer, J. J. G., & Paas, F. G. W. C. (1998). Cognitive architecture and instructional design.
Educational Psychology Review, 10, 251 – 296.
Tabbers, H. K., Martens, R. L., & van Merrienboer, J. J. G. (2004). Multimedia instructions and cognitive load
theory: Effects of modality and cueing. British Journal of Educational Psychology, 74, 71 – 81.
Tindall-Ford, S., Chandler, P., & Sweller, J. (1997). When two sensory modes are better than one. Journal of
Experimental Psychology: Applied, 3(4), 257 – 287.
Tsui, C.-Y., & Treagust, D. F. (2003). Genetics reasoning with multiple external representations. Research in
Science Education, 33, 111 – 135.
Tversky, B., Morrison, J. B., & Betrancourt, M. (2002). Animation: Can it facilitate? International Journal of
Human-Computer Studies, 57, 247 – 262.
Valcke, M. (2002). Cognitive load: Updating the theory? Learning and Instruction, 12(1), 147 – 154.
Weidenmann, B. (1989). When good pictures fail: An information-processing approach to the effect of illustration.
In H. Mandl & J. R. Levin (Eds.), Knowledge acquisition from text and pictures (pp. 157 – 170). Amsterdam:
North-Holland.
Weinstein, C. E., & Mayer, R. E. (1986). The teaching of learning strategies. In M. Wittrock (Ed.), Handbook of
research on teaching (pp. 315 – 327). New York: Macmillan.
Wu, H.-K., Krajcik, J. S., & Soloway, E. (2001). Promoting understanding of chemical representations: Students’
use of a visualization tool in the classroom. Journal of Research in Science Teaching, 38(7), 821 – 842.
Wu, H.-K., & Shah, P. (2004). Exploring visuospatial thinking in chemistry learning. Science Education, 88,
465 – 492.
Yang, E., Andre, T., & Greenbowe, T. J. (2003). Spatial ability and the impact of visualization/animation on
learning electrochemistry. International Journal of Science Education, 25(3), 329 – 349.
Zimmerman, B. J., & Pons, M. M. (1986). Development of a structured interview for assessing student use of
self-regulated learning strategies. American Educational Research Journal, 23, 614 – 628.
Science Education DOI 10.1002/sce