Use of Cognitive Load Theory to Deploy Instructional Technology for Undergraduate Medical Education: a Scoping Review

Abstract

Although new instructional technologies have been widely adopted, cognitive load theory (CLT) is rarely used to inform deployment of those technologies. This scoping review of published literature examined CLT use in the design of teaching technologies for medical students. Three databases were queried, and thematic characteristics were extracted. Fourteen articles met the inclusion criteria. Themes extracted were: Subjects that are inherently visual were contexts for innovations, more than half of the interventions used the CLT modality principle, and CLT-based interventions had mostly positive outcomes. CLT is advantageous for medical education, but its full scope is rarely applied. A broader range of subject areas may benefit from CLT-based teaching.

Introduction

Cognitive load theory (CLT) was first described more than 30 years ago [1]. The theory is multi-faceted and posits that human learning is bounded by both the quantity of new information presented and the demands that the intrinsic and extrinsic qualities of that information make on the learner’s mental processing. CLT underlies a number of instructional methods which aim to organize and present content in ways that do not overwhelm the learner’s working memory (intrinsic load), that minimize nonessential information within the presentation (extraneous load), and that optimize the processes by which the intrinsic load is integrated into existing mental frameworks (germane load). In 2010, psychologist Richard Mayer proposed the cognitive theory of multimedia learning, which heavily relies upon CLT to further delineate the pathways by which learners process combinations of verbal and pictoral information [2, 3]. It is frequently referenced in papers exploring CLT and educational design. CLT was rapidly adopted in the field of psychology, but papers investigating its applications to medical education did not begin to emerge until the early 2000s [4, 5].

Over the last two decades, new teaching technologies have shifted the landscape of medical education. Examples of such technologies are ubiquitous and diverse. They include popular animation websites like Sketchy Micro™ and Sketchy Pharm™, three-dimensional interactive anatomy software like Complete Anatomy™, adjunctive learning programs like Boards & Beyond™ and Pathoma™™, and many more. Calls for a more deliberate incorporation of pedagogical research into the design of such instructional methods have begun to resonate [6]. Much of the current literature examining learning theory in the context of these innovative techniques has been focused on the use of multimedia learning tools as they relate to surgical simulation and resident training [7]. Markedly fewer studies have been performed that investigate the use of CLT and/or the cognitive theory of multimedia learning in the design of educational tools geared toward undergraduate medical students.

This imbalance appears more salient in the context of the “expertise reversal effect,” a phenomenon observed by psychologists in the CLT literature. The effect states that as learners grow in mastery, strategies aiming to optimize cognitive load at the novice level begin to lose their positive effects on learning or even make learning more difficult [8]. This pattern has been shown to apply to several aspects of CLT, including the worked example principle (providing learners with a full solution to study before asking them to demonstrate the skill or knowledge), split attention principle (integrating information in either space or time), and, importantly, the modality principle (adding an auditory component to visual material or vice versa) [9]. These studies suggest that the guiding principles of CLT should be most applicable at the undergraduate medical education (UME) level, where learners tend to be novices, and less effective if employed for more advanced levels of training, such as graduate medical education (GME), where the utilization of these strategies may contribute only minimally to learning or even inhibit it.

Thus, we set out to perform a scoping review of the literature examining the use of CLT to guide the design and implementation of digital learning tools, specifically in the context of UME. We aimed to examine the size and scope of this literature and sought to explore the following questions:

1. How has CLT been applied to learning technologies in the undergraduate medical curriculum? Are certain elements of the theory emphasized over others?

2. Have more innovations been reported in certain academic disciplines, either preclinical or clinical?

3. Has the reported use of CLT strategies been shown to improve learning outcomes?

Materials and Methods

Establishing Inclusion Criteria and Article Selection

Adopting the methods for a scoping review from Levac et al. [10], we began by narrowing the scope of the review to a specific niche of CLT in education. Our primary intent was to examine how the theory has been applied to technological interventions geared towards pre-professional medical students. Therefore, we excluded any studies that dealt with postgraduate or continuing medical education and those that examined other healthcare professions. Our goal was to evaluate CLT-based methods specifically, so papers that focused primarily on the use of other learning theories for the design of teaching technologies were excluded. Papers referencing the cognitive theory of multimedia learning were included, as CLT forms its ideological foundation. Finally, we adhered to a broad, but strictly enforced, definition of technology which excluded technological updates or modifications to established instructional methods such as Problem Based Learning, Team Based Learning, or the PowerPoint lecture. We included only original research or theoretical models and excluded reviews, commentary, and letters. The inclusion and exclusion criteria are summarized in Table 1.

Table 1.

Database search criteria and reasons for exclusion among 95 articles initially found by Boolean search string, ultimately resulting in 14 included publications

Inclusion criteria (yielded 95 articles)
Included characteristics:	Excluded characteristics:
Undergraduate medical education	Postgraduate, continuing, or patient education
CLT theory explicitly referenced	Focus on other pedagogical learning theories
Technology-focused	Traditional teaching methods
Original research articles or theoretical models	Reviews, commentary, letters
Exclusion criteria (removed 81 articles)
Reason for exclusion:	Number of papers removed:
Duplicate article	12
Nontechnology-based instruction	12
Type of publication (review, letter, etc.)	9
Not medical student subjects	36
No (or minimal) CLT discussion	12

Having established these criteria, we conceived of a Boolean construct that optimized the search results for studies that would be relevant to our investigation. The final search string was:

(((cognitive load theory) OR (multimedia theory)) AND ((software) OR (e learning) OR (online) OR (interactive)) AND ((medical student) OR (medical education) OR (undergraduate medical) OR (preclinical)))

This phrase was run through three databases: PubMed, ERIC (Education Resources Information Center), and EBSCO (Academic search complete). Results were filtered for peer-reviewed articles. The search described above yielded 95 publications. All of the 9 results returned by the ERIC database were also represented in one of the two other queried data bases, leaving 86 results. The abstract of each was screened to check for duplicates and to exclude articles that clearly did not meet the criteria for inclusion. Of the remaining 86 articles, 83 were unique, and 3 duplicates were removed. Studies that were not UME-focused, did not involve technology, or did not represent original research were excluded. For a smaller number of papers, a thorough inspection of the full text was required to determine if the inclusion criteria were fulfilled. Papers that made mention of cognitive load theory or the cognitive theory of multimedia design yet failed to base the design or aims of their study on CLT required this additional level of scrutiny. Table 1 summarizes the exclusion criteria and the number of articles excluded by each criterion. Ultimately, 14 papers were chosen to be analyzed for this scoping review.

Charting Data and Reporting of Results

The 14 included articles were analyzed and tracked using Microsoft Excel. Articles were categorized by bibliographic characteristics and key themes. Papers were classified by type of study (either original research or original theoretical model/framework), year of publication, institution and region, subject area within the medical school curriculum, interventional modality, influence of CLT or multimedia theory on project design, and primary outcomes. The categories chosen reflected the nature of the questions posed in the study aims with the goal of reaching a comprehensive understanding of the scope and nature of the literature. Table 2 presents a list of the included articles along with relevant descriptive information.

Table 2.

Institution, subject matter, technological characteristics, and CLT incorporation among articles meeting criteria for inclusion

Author and date	Institution	Topic area	Interventional modality	Aspects of CLT discussed
Allen et al. (2015) [23]	Schulich School of Medicine, Ontario, Canada	Anatomy	Interactive 3D model of the eye	Modality principle, extraneous load reduction, spatial contiguity, personalization effect
Andersen et al. (2016) [11]	Rigshospitalet, Copenhagen, Denmark	Surgical simulation	VR surgical simulator	Validation of cognitive load estimation techniques
de Araujo et al. (2016) [12]	State University of Campinas, São Paulo, Brazil	Emergency medicine	Virtual rounds	Modality principle, repetition of visual scheme
Drees (2020) et al. [13]	Goethe University, Frankfurt, Germany	Histology	Interactive e-learning software	Modality principle, repetition of visual scheme
Haji (2015) et al. [14]	University of Toronto, Toronto, Canada	Surgical knot tying	Computer-based video instruction	Validation of cognitive load estimation techniques
Holland (2015) et al. [15]	Royal College Surgeons, Dublin, Ireland	Histology	Online multimedia exam	Modality principle
Holzinger (2009) et al. [16]	Medical University of Graz, Graz, Austria	Hemodynamics	E-learning physiology simulator	Multimedia theory, dynamic imagery
Khalil (2005b) et al. [5]	Florida State University, Tallahassee, Florida	Anatomy	Computer-based learning strategies	Theoretical model, thorough discussion of CLT
Khalil (2008) et al. [17]	Tuskeegee University College of Veterinary Medicine, Nursing and Allied Health, Tuskegee, Alabama	Anatomy	Digital enhancement of anatomical slices	Worked example theory
Makransky (2020) et al. [18]	University of Copenhagen, Copenhagen, Denmark	Medical genetics	VR laboratory simulator	Cognitive load feedback theory
Pickering (2015) [19]	University of Leeds, Leeds, United Kingdom	Anatomy	Online anatomical drawing screencasts	Modality principle
Sagoo (2020) et al. [20]	King's College, London, UK	Anatomy and radiology	Online multimedia exam	Modality principle
Stevens and Palacio-Cayetano (2003) [22]	UCLA, Los Angeles, California	Online problem solving	Techniques for online problem solving	Theoretical model, thorough discussion of CLT
Yue et al. (2013) [21]	UCLA, Los Angeles, California	Medical animation	E-learning animations	Observational study assessing modality principle, minimization of extraneous load, others

Results

Type of Study

Of the papers chosen for analysis, 11 were original articles [11–21], while two were theoretical frameworks [5, 22] intended to guide the design of future e-learning tools. One study described how an e-learning platform was developed but did not present data validating its use [23].

Year of Study

The year of publication for the chosen studies was skewed towards the last several years as shown in Table 1. This was unsurprising given the increasing emphasis on the use of technology in medical education and the relatively recent emergence of papers connecting CLT to medical teaching [2]. Accordingly, the two papers that described theoretical frameworks for the inclusion of CLT in e-learning design represented the two earliest included publications in 2003 and 2005 [5, 22].

Institution

The articles chosen did not demonstrate an appreciable regional bias, though two articles came from the same institution [21, 22]. While the USA contributed four articles to the list from three different institutions [5, 17, 21, 22], several other regions were well-represented. King’s College and the University of Leeds, institutions in the UK, each contributed one article [19, 20]. A total of two were from Denmark; University of Copenhagen and Rigshospitalet in Copenhagen appeared once each [11, 18]. The rest of the list was comprised by articles from various institutions in Europe or the Americas [12, 16].

Medical School Subject Matter/Topic Area

Consistent themes arose when categorizing the articles by course, module, or subject matter (Table 2). From a preliminary search of the CLT and educational technology literature, it appeared that at the resident level, many studied interventions were in the fields of surgical instruction and surgical anatomy [3, 24–27]. Our study confirmed that this trend was present at the undergraduate level. Medical student anatomy courses were the subject of five of the fourteen chosen articles [5, 17, 19, 20, 23]. Two articles dealt with surgical skills, specifically knot tying and procedural simulation [11, 14] and two were related to histology [13, 15]. The remaining subjects were represented once each and comprised emergency medicine, medical animation, medical genetics, hemodynamics, and online problem solving.

Nature of Technological Intervention Studied or Proposed

The interventions described by studies in our review were highly varied (Table 2). Five of the studies examined online software that consisted of either three-dimensional representations of interactive anatomical structures or two-dimensional simulations of physiological concepts [13, 16–18, 23]. Two papers described online multimedia exams that included questions with and without added images [15, 20]. The remaining articles described virtual bedside rounds, educational animations, virtual reality simulation, anatomical drawing screencasts, and computer-based video instruction.

Theoretical Basis of Intervention and Outcomes

The extent to which CLT and multimedia theory was used to guide technological design was variable among the chosen articles, as illustrated in Table 2. Many of the studies cited the modality principle, i.e., the inclusion of elements directed at more than one of the five senses, as at least part of their CLT influence. Some of the more basic interpretations of this concept were described by Sagoo et al. [20] and Holland et al. [15], in which the intervention consisted of adding images to online exams. The extent of CLT theory incorporated in these experiments was the addition of descriptive images alongside text in MCQ exams to assess whether or not added visual elements reduced extraneous load and led to better performance. Sagoo et al. saw significantly improved performance when images were added while Holland et al. observed no difference between groups.

Pickering [19] employed a similar strategy when introducing anatomical drawing screencasts to traditional lectures, though the added modality introduced was auditory narration rather than descriptive text. Pickering reported increased student satisfaction with the course but made no quantitative claims about changes in learning outcomes. Drees et al. [13] also described the use of a combination of auditory and visual splitting in an online histology course, but they incorporated an additional layer of multimedia theory which states that repetition of visual schema is essential for the construction of long-term memory. The authors saw measurable improvements in student reception of the course as well as knowledge acquisition when compared to a traditional learning method. de Araujo Guerra Grangeia et al. [12] heavily incorporated this principle of repeated visual schema in the design of virtual emergency medicine rounds and saw significant improvement in student satisfaction and final grades in the emergency medicine clerkship.

Other articles made more extensive reference to CLT or integrated a greater number of CLT principles into the interventions they examined. For example, Yue et al. [21] considered a number of different aspects of CLT in their evaluation of medical animations and ultimately found the use of CLT to be lacking in a sample of 430 online animations. Areas they targeted for improvement included the use of auditory input (demonstrated in less than 20% of animations studied), pre-defining key terms (< 10%), a high number of extraneous processing elements found within the animations (> 2/3 of animations), and a large amount of redundant text throughout. Holzinger et al. [16] explored static versus dynamic imagery in a two-dimensional simulation of hemodynamics and reported mixed results among different physiological topics when incorporating dynamic models. Khalil et al. [17] hypothesized that intrinsic load could be reduced by superimposing radiologic images on anatomical cross sections to provide students with a visual “worked example,” which is a well-established CLT strategy [8]. Khalil and colleagues reported no clear positive effects, possibly due to an increase in overall cognitive load when superimposing images. Makransky et al. [18] referenced the explanatory feedback principle of multimedia theory when designing a virtual laboratory simulator for a medical genetics course. The group found a significant increase in students’ positive perception of the learning material as well as positive effects on student motivation and knowledge transfer.

Allen et al. [23] made explicit mention to elements of cognitive load proposed by Mayer in 2003 and 2005 [28, 29] in the design of a 3D interactive model of the oculomotor system. The authors reference the modality effect, the reduction of extraneous audio and visual effects to accommodate limited working memory, the use of hover-over labels that adhered to the principle of spatial contiguity, and the use of a “welcoming voice” in its narration, a term Mayer referred to as the “personalization effect” [28]. There were no outcomes reported in this descriptive piece that served to describe the steps taken when designing the model.

Two studies, Andersen et al. [11] and Haji et al. [14] attempted to validate methods of measuring cognitive load during technology-enhanced learning tasks (surgical knot-tying video simulations and ENT surgery virtual reality) rather than attempt to study the effects of load reduction. They reported promising evidence for using secondary task performance and mental effort rating as surrogate measurements of cognitive load. Khalil et al. [5] and Stevens and Palacio-Cayetano [22] both utilized a number of aspects of CLT to construct theoretical frameworks for the design of future educational projects; however, these papers did not report outcomes.

Themes from the Scoping Review

The available literature related to the use of CLT and multimedia theory to guide the design of technological interventions in undergraduate medical education is relatively small. There is a much greater body of evidence on this topic in the context of resident physician education, especially in the fields of surgery and radiology. However, in light of the aforementioned “expertise reversal effect,” the application of these GME studies to UME is tenuous. The remainder of the literature is likewise inapplicable, as it is predominantly composed of studies geared towards continuing medical education, the use of CLT in other professions (e.g., nursing, veterinary medicine), patient education, college undergraduate education, the use of CLT strategies to enhance traditional learning modalities, or the incorporation of alternative pedagogical learning theories. Nonetheless, common threads were extracted from the number of articles included in this study. We will present those themes by revisiting our original aims.

1. How has CLT been applied to learning technologies in the undergraduate medical curriculum? Has there been a greater emphasis on certain elements of the model over others?

Many of the papers based at least part of their designs on the principle of multiple modalities and the concept of reducing extraneous load through the addition of auditory and visual content. Sagoo et al. [20] and Holland et al. [15] added images to MCQ tests, an intervention that does not adhere to the definition of multimodality described in the mainstream CLT literature [8]. There have been no demonstrated benefits to combining written words and visual stimuli as both of these design components represent visual modalities. Indeed, this was an area Yue et al. identified as a target for improvement in animated educational content [21]. With the exception of Allen et al. [23], no authors made significant attempts to implement methods of reducing intrinsic load and no papers explored methods aimed at optimizing germane load. Some of these studies had the potential to be approached with these concepts in mind, but the authors did not remark upon them.

• 2.Has there been greater representation of such innovations in specific preclinical or clinical subject areas?

There appears to be a strong focus on anatomy when designing CLT-based learning tools. This is in line with the broader body of literature on CLT and technology. It is possible this is related to a focus on the multimedia and multimodality elements of technological interventions since anatomy and radiology are inherently visual subjects. This assumption is supported by the fact that the second and third most common topic areas in our review described the study of histological images and surgical simulation respectively.

• 3.Have these strategies been shown to improve learning outcomes?

The literature scrutinized in our review showed a trend towards positive outcomes when attempts were made to incorporate CLT into learning-tool design. This consisted of either increased student satisfaction with the optimized tool, increased student performance on assessment following the use of the tool, or both. Two papers reported no difference when incorporating such methods, though it is worth noting that in one of those studies the authors did not adhere to evidence-based CLT in the design of their intervention [15]. One paper reported mixed results when applying technological learning methods to different learning topics under the umbrella of hemodynamics [16], suggesting that effectiveness of CLT methods may be domain-dependent.

Discussion

In general, these papers described studies in which learning tools were developed with CLT and/or multimedia learning theory in mind and were then compared to more traditional instruction methods. None of the papers we examined attempted to compare the efficacy of technology-based learning tools designed with principles of CLT against similar tools designed without elements of CLT. Such a comparison would be necessary to verify that any gains in learning outcomes could be attributed to the use of CLT rather than simply the employment of a new technology. Although we expect that introducing instructional technology that lacks influence from CLT would have a reduced impact on learning, empirical verification of this presumption is needed. This would seem to be an important contribution to the field of medical education, and to the field of CLT in general, as a proof-of-concept for theoretical models like those described by Khalil et al. [5] and Stevens and Palacio-Cayetano [22]. In addition, there is a great need to broaden the scope of these studies to include more comprehensive incorporation of the multiple elements of CLT, as the theory encompasses far more than the modality effect. Studies from other fields have validated attempts to regulate intrinsic load, reduce extraneous load, and optimize germane load when designing educational platforms [8].

This scoping review had several limitations. We chose to narrow the scope of our review to papers that were geared towards undergraduate medical students and excluded studies that examined nursing and veterinary students. The implications drawn from those studies may be generalizable to undergraduate medical education as well, since all of these groups comprise novice, pre-professional learners in the health professions. Our method of constructing the Boolean phrase was based on trial and error. The phrase was tweaked after each search until the databases returned the greatest number of relevant studies. Applicable and valuable studies may have been missed by the final phrasing of the construct but could feasibly be captured with a revised set of search parameters.

Conclusion

We performed a scoping review of the literature describing the deliberate incorporation of cognitive load theory and multimedia learning theory in the design and implementation of technological teaching interventions in undergraduate medical education. The results showed that this body of literature is small but has seen increased growth in the latter half of the last decade. There has been a disproportionate focus on anatomy and on a narrow interpretation of CLT, namely an overemphasis on the modality principle. There is ample room for further studies that attempt to expand on the use of CLT as a strategy for tool development.

Declarations

Ethical Approval

Not applicable

Conflict of Interest

The authors declare no competing interests.