Interactive and dynamic visualizations in teaching and learning of anatomy: A cognitive load perspective

Authors

  • Mohammed K. Khalil,

    Corresponding author
    • Learning Systems Institute, Florida State University, 2000 Levy Avenue, Suite 320, Tallahassee, FL 32310
    Search for more papers by this author
    • Fax: 850-644-4179

    • Dr. Khalil is an assistant in research at the Learning Systems Institute (LSI) at Florida State University (FSU). He taught gross anatomy for several years and currently his research interests are in the areas of learning and instructional technology. He is interested in advancing medical education with innovative strategies and technologies. Dr. Paas is an associate professor of educational technology at Educational Technology Expertise Centre, Open University, The Netherlands. He has developed research expertise in cognition and instruction, instructional design for complex learning, cognitive load, measurement, adaptive education, and multimedia instruction. Dr. Johnson is the associate director for research at the LSI at FSU. He has designed, developed, researched, and evaluated many computer-based prototypes and programs related to science, engineering, and medicine. He is currently working on several research projects studying human expertise. Dr. Payer is the director for the anatomy, embryology, and imaging course and the director of the year 1 medical curriculum at the FSU College of Medicine. His primary interests are in developing continuous quality improvement, enhancing student learning, and increasing student satisfaction while also reducing operational costs. Dr. Payer is a member of the AAA Educational Affairs Committee

  • Fred Paas,

    Search for more papers by this author
    • Dr. Khalil is an assistant in research at the Learning Systems Institute (LSI) at Florida State University (FSU). He taught gross anatomy for several years and currently his research interests are in the areas of learning and instructional technology. He is interested in advancing medical education with innovative strategies and technologies. Dr. Paas is an associate professor of educational technology at Educational Technology Expertise Centre, Open University, The Netherlands. He has developed research expertise in cognition and instruction, instructional design for complex learning, cognitive load, measurement, adaptive education, and multimedia instruction. Dr. Johnson is the associate director for research at the LSI at FSU. He has designed, developed, researched, and evaluated many computer-based prototypes and programs related to science, engineering, and medicine. He is currently working on several research projects studying human expertise. Dr. Payer is the director for the anatomy, embryology, and imaging course and the director of the year 1 medical curriculum at the FSU College of Medicine. His primary interests are in developing continuous quality improvement, enhancing student learning, and increasing student satisfaction while also reducing operational costs. Dr. Payer is a member of the AAA Educational Affairs Committee

  • Tristan E. Johnson,

    Search for more papers by this author
    • Dr. Khalil is an assistant in research at the Learning Systems Institute (LSI) at Florida State University (FSU). He taught gross anatomy for several years and currently his research interests are in the areas of learning and instructional technology. He is interested in advancing medical education with innovative strategies and technologies. Dr. Paas is an associate professor of educational technology at Educational Technology Expertise Centre, Open University, The Netherlands. He has developed research expertise in cognition and instruction, instructional design for complex learning, cognitive load, measurement, adaptive education, and multimedia instruction. Dr. Johnson is the associate director for research at the LSI at FSU. He has designed, developed, researched, and evaluated many computer-based prototypes and programs related to science, engineering, and medicine. He is currently working on several research projects studying human expertise. Dr. Payer is the director for the anatomy, embryology, and imaging course and the director of the year 1 medical curriculum at the FSU College of Medicine. His primary interests are in developing continuous quality improvement, enhancing student learning, and increasing student satisfaction while also reducing operational costs. Dr. Payer is a member of the AAA Educational Affairs Committee

  • Andrew F. Payer

    Search for more papers by this author
    • Dr. Khalil is an assistant in research at the Learning Systems Institute (LSI) at Florida State University (FSU). He taught gross anatomy for several years and currently his research interests are in the areas of learning and instructional technology. He is interested in advancing medical education with innovative strategies and technologies. Dr. Paas is an associate professor of educational technology at Educational Technology Expertise Centre, Open University, The Netherlands. He has developed research expertise in cognition and instruction, instructional design for complex learning, cognitive load, measurement, adaptive education, and multimedia instruction. Dr. Johnson is the associate director for research at the LSI at FSU. He has designed, developed, researched, and evaluated many computer-based prototypes and programs related to science, engineering, and medicine. He is currently working on several research projects studying human expertise. Dr. Payer is the director for the anatomy, embryology, and imaging course and the director of the year 1 medical curriculum at the FSU College of Medicine. His primary interests are in developing continuous quality improvement, enhancing student learning, and increasing student satisfaction while also reducing operational costs. Dr. Payer is a member of the AAA Educational Affairs Committee


Abstract

With the increasing use of computers in the classroom and the advancement of information technology, a requirement to investigate and evaluate different strategies for the presentation of verbal information in interactive and dynamic visualizations has risen to a high level of importance. There is a need for research efforts that apply cognitive load theory (CLT), cognitive learning strategies, and established principles of multimedia design to conduct empirical research that will add to our knowledge of designing and developing dynamic visualizations for teaching and learning anatomy. The impact of improved teaching and learning of anatomical sciences and the development of a set of guiding principles to facilitate the design and development of effective dynamic visualizations represent a significant achievement for medical education with wide application. This theoretical paper presents the foundations of CLT, cognitive learning strategies, and principles of multimedia design to guide the needed research on dynamic visualizations. Anat Rec (Part B: New Anat) 286B:8–14, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

With increasing use of computers in the classrooms, subject specialist-friendly hypermedia authoring tools, and other advances in information technology, the benefits of integrating visualizations in anatomy teaching and learning have been extended from static visualizations to interactive and dynamic visualizations. This capability introduces the new problem of understanding, learning (and how best to support it) with dynamic visualizations (Ploetzner and Lowe, 2004). Very little principled or empirically established guidance is available with regard to the design and integration of visualizations in teaching and learning. Making visuals more dynamic and interactive will not always benefit learning and may confuse some learners in some situations (Sweller and Chandler, 1994). The design of effective interactive dynamic visualization should be sensitive to the limitations of cognitive architecture and learner expertise.

Dynamic visualizations and animations are often perceived to be synonymous, although animations are a subset of dynamic visualizations. Ainsworth and Van Labeke (2004) identified three classes of dynamic visualizations based on how time is represented: time-persistent, time-implicit, and time-singular representations. This classification expands the scope of the narrow prevailing ideas about dynamic visualizations from animations to a richer set of visualization possibilities. Moreover, according to learner-centered instructional design research, it is important to provide learners with interactivity and to encourage learners to assess their own learning (Mory, 2004). The process of structuring activities and providing feedback can be supported with appropriate labeling strategies that present verbal information in dynamic visualizations. Labeling strategies are not limited only to labeling anatomical structures in visualizations; the strategies can also be used to create interactive learning environments where learner can hide or reveal the labels, thus making the static visualizations dynamic. The interactivities provided by labeling strategies can range from simple hide-and-show labeling to labeling that is adaptive to learners' levels of content knowledge.

In this article, we present the foundation of cognitive load theory, cognitive strategies (e.g., imagery strategy), and principles of multimedia learning for establishing guiding principles to create and utilize dynamic visualizations to support teaching and learning of anatomy. The article introduces new terminologies and concepts with their explanations and definitions. In Table 1, we summarize the most relevant terms and concepts. We recognize that this information might be new to most anatomist and anatomy teachers. In a follow-up article (Khalil et al., 2005), we present the implications of designing instructional materials based on cognitive load theory (CLT) and the needed research to advance the effectiveness and efficiency of dynamic visualizations.

Table 1. Definitions and characteristics of terms and concepts related to CLT
TermsDefinition and characteristics
Cognitive load— A multidimensional construct representing the load that performing a particular task imposes on the learner's cognitive system
Mental load— A load that is imposed by task (environment) demands
Mental effort— The amount of cognitive resources allocated to accommodate the task demands
Intrinsic cognitive load— Imposed by elements in the content information
 — Closely related to the complexity of subject matter
Extraneous cognitive load— Imposed by poorly designed instructional materials
 — Ineffective and does not contribute to the construction and automation of schema
Germane cognitive load— Created by effective instructional strategies
 — Foster the process of schema construction and automation
Sensory memory— Perceive incoming information
 — Include a separate partition for each of the senses
 — Store incoming information for few seconds
Working memory— Provide our consciousness
 — Attend to incoming information
 — Its capacity is very limited to about seven ± two chunks of information
 — Its duration is very limited
 — Possess separate processors for visual and auditory information
Long-term memory— Store knowledge permanently
 — Its capacity is unlimited
Schema— Organization of knowledge or mental data structure (data schema)
 — Procedures or ways of processing and organizing information (process schema)
Automation— An important process in the construction of schema
 — Occurs after extensive practice to process information with minimal working memory load

COGNITIVE LOAD THEORY

The information processing model of human cognitive architecture (Fig. 1) consists of three types of memories: sensory memory (SM), working memory (WM), and long-term memory (LTM). The SM perceives the incoming information from the environment and stores it for no more than few seconds. The WM provides our consciousness and possesses separate processors for visual and auditory information. It is very limited in capacity and duration. The LTM stores information permanently in unlimited capacity. From the external environment, learners receive input into the sensory memory, which elicits an orienting response that focuses the attention on the stimulus. Attention is fundamental in getting the stimuli into the information processing system, otherwise stimuli will be forgotten. Information in WM is passed to LTM following its rehearsal and encoding.

Figure 1.

Information processing model. Modified from Gage and Berliner (1998).

According to CLT (Sweller, 1988, 1999; Sweller and Chandler, 1991, 1994; Paas et al., 2003a), effective learning occurs if instructional conditions are aligned with human cognitive architecture. CLT is based on a view of cognition that involves a limited WM interacting with an unlimited-capacity LTM. Novel information must be processed through WM that is limited in capacity and duration in order for learning to occur (Miller, 1956; Simon, 1974). Therefore, instructional strategies that promote successful processing in WM will enhance learning and understanding.

In developing knowledge, information is stored in LTM in the form of schemas (information network). The processes or ways in which information is processed and organized are referred to as process schemas (West, 1981). The primary function of a schema is to provide a mechanism for knowledge organization and storage in LTM and to reduce WM load. When constructing schemas, information must be processed in WM before being stored in LTM. Automation is a process that may bypass WM and update existing schema after sufficient practice, i.e., ability to perform task without concentrating. It is an important factor in schema construction that frees WM capacity for the processing of additional information. CLT assumes that a WM of a limited capacity becomes very effective when dealing with familiar material that is previously stored in LTM. The efficiency of processing information in WM is the primary focus of CLT.

Cognitive load can be defined as a multidimensional construct representing the load that performing a particular task imposes on the learner's cognitive system (Paas and Van Merriënboer, 1994). It is a construct that represents both mental load and mental effort. Mental load refers to load imposed by task demands, whereas mental effort refers to the amount of cognitive resources available for the task demands. The load in working memory is mainly affected by the intrinsic nature (difficulty) of the instructional materials (intrinsic cognitive load) and the strategies for information presentation (extraneous or germane cognitive load). Thus, CLT differentiates between three types of cognitive load: intrinsic, extraneous, and germane (Paas et al., 2003b). The sum of intrinsic, extraneous, and germane loads is equal to the total cognitive load. The latter should not exceed the memory resources available, otherwise learning will not be effective (Paas et al., 2003b). Accordingly, the design and implementation of activities and representations should maximize germane load while minimizing extraneous load.

Intrinsic Cognitive Load

Intrinsic cognitive load is imposed by elements in the content information. It is closely related to the complexity of the subject matter. Instructional materials can be simple or complex, depending on elements and interactivities (e.g., the number of elements presented to a learner for simultaneous processing in WM). For example, when students learn new vocabulary, such as the names of muscles, nerves, bones, and blood vessels, the task might be difficult because of the huge number of terms involved. However, this type of learning does not impose a heavy cognitive load, because items can be learned in isolation without reference to other elements. On the other hand, other aspects of learning necessitate element interactivity. For instance, when learning about the triceps brachii muscle, the learner has to keep in mind that it has three “heads,” is found in the brachial region, its blood is supplied by branches of the brachial artery, and its nerves arise from the brachial plexus—this is heavy element interactivity. In this case, students must simultaneously manipulate several elements in working memory.

With the construction and automation of schema, element interactivities may act as a single element in working memory (i.e., interacting elements become an integrated whole to that learner). This effect explains why a large number of interacting elements for novices may be perceived as a single element by an expert. In the process of developing instructional materials, the levels of element interactivity can be determined by an analysis of the instructional materials along with an analysis of the knowledge levels of targeted learners.

Previous research has suggested a variety of instructional strategies that promote the construction and automation of schema to decrease intrinsic cognitive load. Scaffolding whole-task practice (Van Merriënboer et al., 2003) and two-phase isolated-interactivity instruction (Pollock et al., 2002) were seen as effective strategies. The whole-task approach focuses on the presentation of a sequence of tasks that proceed from relatively simple tasks supported by explanations to complex tasks with support faded as learners progress. In the two-phase isolated-interactivity strategy, the first phase focuses on presenting isolated elements that help novices construct partial schemas, while the second phase focuses on the explanation of the information elements and their interaction.

Extraneous Cognitive Load

Extraneous cognitive load can be inadvertently imposed by poorly designed instructional materials. It is ineffective and does not contribute to the construction and automation of schemas. In some instances, learners are engaged in cognitive activities superfluous to the intended learning objectives. For example, when students are presented with anatomical visuals accompanied by explanatory text that is spatially separated from the visual, students have to split their attention between the two sources of information. This split-attention process exerts extraneous cognitive load and can make learning inefficient or ineffective.

Strategies such as visual grouping (Rieber, 1993), segmentation (Zacks and Tversky, 2001), modality principle (Penny, 1989; Mousavi et al., 1995), contiguity principle (Moreno and Mayer, 1999), and signaling or cuing (Jeung et al., 1997) have been adopted to decrease extraneous cognitive load in dynamic visualization. In visual grouping, chunks of information are sequentially presented rather than shown all at once. The presentation of visual groups of information gives learners the opportunity to build more parsimonious mental representations. Regarding the modality of presentation, spoken verbal explanations that accompany visual learning materials are generally superior to written verbal explanations (Sweller et al., 1998; Moreno and Mayer, 1999). Contiguity principle strategies are used to overcome split-attention effects by the simultaneous presentation, for instance, of verbal explanation of visual materials (temporal contiguity), and presenting text materials that are juxtaposed to visual materials (spatial contiguity). Signaling or cuing strategies are used to attract learner's attention by a signaling device (e.g., color or pedagogical agent) to the relevant parts of the instructional materials.

Germane Cognitive Load

Germane load is created by effective instructional strategies that foster the process of schema construction and automation. To increase germane load, learners must engage in cognitive activities such as elaboration, abstraction, and drawing inferences. In increasing germane load, strategies such as imagining (Cooper et al., 2001), variability (Paas and van Merriënboer, 1994), subgoaling (Catrambone, 1998), and expectancy-driven (Renkl, 1997; Hegarty et al., 2003) instructional methods were sought to increase germane cognitive load.

The imagining strategy is effective for more advanced learners. It stimulates learners to imagine procedures and concepts in order to facilitate the automation of schemas. Through imagination, learners can rehearse and further automate an existing schema. In variability, learners are promoted to link new information to their prior knowledge in order to develop a general schema on the basis of a varied range of cases or variability of practices. In the subgoaling strategy, learners are prompted to group logical steps with a procedure into meaningful subgoals that encourage learners to self-explain the logic of these steps. Expectancy-driven strategies enable learners to actively process instructional materials by predicting the next step in a task.

IMAGERY STRATEGY

Imagery strategy is based on the learner's ability to form mental images or representations of things or events to help in remembering, understanding, and comprehending information and in drawing inferences. Functions that relate to verbal and visual coding abilities are separated in the two hemispheres of the brain. Speech is organized chiefly in the left hemisphere and nonverbal imagery in the right hemisphere. The storage capacity for pictorial information seems to exceed the very large storage capacity for verbal materials. The imagery system works better for processing concrete and spatial information and the verbal system works better for processing abstract and sequential information. Better information retrieval occurs when people can attach verbal material to pictorial images and pictorial images to verbal labels, since the two systems are richly interconnected (Gage and Berliner, 1998).

Imagery and Learner Characteristics

There are two important factors that need to be considered before utilizing imagery strategy. Given the fact that imagery strategies are better for concrete information and benefit more visually oriented learners, content analysis and learner analysis are important during the design process. Low-experience, high-spatial-ability learners are the most likely to benefit from the strategy that coordinates verbal and visual presentation of scientific explanation (Mayer and Sims, 1994). Introducing visuals to the learners would help them develop mental representations that can be manipulated afterward to draw inferences (Gyselinck and Tardieu, 1999). Moreover, learner level of expertise has been shown to be an important factor that affects instructional strategies. That is, effective instructional strategies for novices become ineffective as learners develop expertise in a specific domain, i.e., expertise reversal effect (Kalyuga et al., 2003).

Text-Picture Combinations

Since imagery is a multipurpose strategy (West et al., 1991), it helps in delivering varying forms of information when combined with other instructional strategies. That is, it can accommodate a diversity of learning styles. The study of Gyselinck et al. (2000) on the integration of verbal and pictorial information within the instructional multimedia system indicated that pictorial information enhances the learning process. Their results showed the facilitative effects of illustrations in improving the comprehension of text and hence deeper understanding. Pictures help learners build connections between the sentences of the text and the picture for better recall, better accuracy, and shorter response times (Gyselinck and Tardieu, 1999). Visual aids (illustrations, diagrams, pictures, and models) can be presented to learners to help them develop mental images in order to understand spatial relationship. For example, to study the structure of the heart, an instructor can present a model, pictures, and diagrams of the heart labeled with the correct information. These visual aids make the structural relationship more transparent and they make spatial relations explicit (Gyselinck and Tardieu, 1999).

Considering the recall of information, pictures have a superior effect (Paivio, 1971; Berry et al., 1997). The picture's superiority in explicit memory tasks is due to its stronger associative perceptual information than that for words (Kinjo and Snodgrass, 2000). Pictures enable the extraction and retention of information that readers do not encode effectively (McDaniel and Waddil, 1994). Pictures highlighting details effectively increased the recall of those details, and picture-depicting relationships effectively increased recall of that relational information (Waddil and McDaniel, 1992).

Using text and picture as complementary media was shown to be a more effective approach in increasing knowledge acquisition. Both sources of information are processed separately in order to acquire the entire meaning of their combination (Molitor et al., 1989). The complementary effects were seen as each medium can provide the context necessary to understand the other medium; each medium can help organize the other by providing the same information in different formats; each medium can guide the reader in the processing of the other medium. However, the information processing of the text-picture combination was shown to be cognitively demanding, since the readers must split their cognitive capacity between the two media (Hegarty et al., 1990). That is, split attention occurs if the picture is spatially separated from its explanatory text. The split-attention effect (Kalyuga et al., 1998; Mayer and Moreno, 1998) occurs as learners process simultaneously two or more sources of information in order to generate meaning from the instructional materials. The process of searching, matching, and integrating sources of information mentally will impose working memory with extraneous cognitive load that negatively affects learning.

The experiments by Betrancourt and Bisseret (1998) reinforced the hypothesis that materials where text and picture are integrated improve learning. They examined the effect of the spatial display of the text-picture information on the user's cognitive processes. Three different displays on the computer screen were compared: A split display (text and picture information displayed in separate areas on the screen); an integrated display (text information close to the part of the picture to which it refers); and a pop-up display (text information integrated in pop-up fields that appeared only via the user's action). The results showed that the integrated display, especially the pop-up display, led to higher performances for an equal or lower learning time.

Colors as Imagery Codes

Color has been used in instructional visuals to increase audience attentions, to provide basis for discrimination, and to make illustrations attractive and emotionally appealing (Dwyer and Lamberski, 1983). Previous research showed that concreteness (e.g., existing in reality, perceptible by senses) and vividness (e.g., clarity, distinctness, strength, colorfulness) are powerful additives for images that have positive effects on learning (West et al., 1991). Learning with graphics is seen as a learner problem-solving activity in order to reconstruct the designer decisions and intentions by adequate interpretation of the visual codes (Weidenmann, 1994). Graphic codes are classified into two classes, directing codes and depicting codes. Weidenmann (1994) defined directing codes as codes used by the graphic designer to direct the picture processing of complex visual arguments in order for the learner to perceive and process them sufficiently. The depicting codes were defined as devices to construct surface structure, which is processed in analogy to real-world perception.

In a review of research by Dwyer and Lamberski (1983) on the effect of the use of color in the teaching-learning process, color was seen as a viable instructional variable for learning specific types of tasks. However, the relative effectiveness of color as an aid to improve students' achievement was inconclusive. Instructional variables such as color and degree of realism may not act as a significant aid in learning; however, they may direct learner attention (Anglin et al., 2004). Color pictures or illustrations are recalled better than black-and-white pictures and line drawings.

Worley and Moore (2001) examined how cognitive style in terms of field dependence (learner cannot separate or restructure the objects presented from the surrounding field) and field independence is influenced by the use of color in visual imagery. They evaluated the effect of image characteristics (e.g., black and white, full color, or highlight color) on recognition memory and recall for field-dependent and field-independent learners. Their results showed that there is no difference between utilizing either realistic or contrived highlight color as a cue for improving recall memory for images versus the use of black-and-white or realistic images.

Computer-Based Imagery

Dynamic digital visualizations can be considered as a type of computer-based imagery that represent changes in time or space and allow learner interactivity. A review of studies by Tversky et al. (2002) showed no advantages of utilizing animated graphics over static graphics. In comparing two types of computer presentation methods of anatomical information (i.e., multiple views versus key views), Garg et al. (1999) concluded that the self-rotating multiple-view presentation (e.g., views at 10° intervals throughout a 360° sphere) had no overall instructional advantage over a simple presentation of key views (e.g., anterior, posterior, lateral, superior, and inferior views). However, when spatial ability was controlled, the key view presentation showed an overall advantage over the multiple-view presentation. Students of high spatial ability can learn with either representation format. However, students with low spatial ability have better success with key-points formats. In conclusion, they suggested that certain computer models that involve moving the object to be learned through multiple perspectives may be less effective for some, and they disadvantage students of low spatial ability. Their study assessing the effectiveness of student-controlled multiple views (model rotated horizontally by 10°) on the spatial learning process indicated that multiple views of an object improved the mental representation if compared with two key views (Garg et al., 2001). Student control of the multiple views is the important factor that helps them remember rather than the presentation of all possible key views.

PRINCIPLES OF MULTIMEDIA LEARNING

Principles of multimedia design and learning (Mayer, 2001; Mayer and Moreno, 2002, 2003) are drawn from dual-coding theory (Paivio, 1986; Clark and Paivio, 1991), cognitive load theory (Chandler and Sweller, 1991; Baddeley, 1992; Sweller, 1999), and constructivist learning theory (Mayer and Sims, 1994). These principles are based on three assumptions: dual channel, limited capacity, and active learning assumptions. In the dual-channel assumption, visual and verbal information are processed by two different information-processing systems. The limited-capacity assumption is based on the fact that information processing is very limited within each of the information processing channel. In the active learning assumption, meaningful learning occurs as learners mentally organize and integrate visual and verbal information when they engaged in active learning processes. Mayer and Moreno (2003) used cognitively based principles to offer suggestions to overcome cognitive overload in multimedia design. They concluded that effective instructional design should be sensitive to learners' cognitive load, which is itself dependent on the learner's prior experience and level of relevant knowledge.

In this article, we have presented aspects of cognitive theories and internal cognitive processes that lead to the development of knowledge. The fundamental of cognitive load theory focuses on the role of working memory in the learning process. Learning requires the active engagement of working memory in the comprehension and encoding of instructional materials into long-term memory. Due to the fact that working memory is very limited in its capacity and duration, special attention should be paid to avoid cognitive overload; otherwise, learning will be ineffective.

We provided brief examples of strategies thought to be effective in reducing intrinsic and extraneous cognitive loads and increasing germane (effective) load. In Khalil et al. (2005), we present the implications of the cognitive load theory on anatomy education.

Ancillary