Cognitive Load Measurement and Application

A Theoretical Framework for Meaningful Research and Practice

Robert Z. Zheng, Robert Z. Zheng

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781315296234
• Edition: 1
• Topic: Education
• Subtopic: Educational Psychology

Cognitive Load Measurement and Application provides up-to-date research and theory on the functional role of cognitive load measurement and its application in multimedia and visual learning. Grounded in a sound theoretical framework, this edited volume introduces methodologies and strategies that effect high-quality cognitive load measurement in learning. Case studies are provided to aid readers in comprehension and application within various learning situations, and the book concludes with a review of the possible future directions of the discipline.

PART I

Theoretical Perspectives on Cognitive Load Measurement

1

THE ROLE OF INDEPENDENT MEASURES OF LOAD IN COGNITIVE LOAD THEORY

John Sweller

Cognitive load theory (Sweller, Ayres, & Kalyuga, 2011) was designed as an instructional theory based on our knowledge of human cognitive architecture. It has constantly developed as new data and new theoretical constructs became available. For most, but not all, of its history, cognitive load theorists have attempted to find techniques that could be used to measure load rather than simply relying on it as a theoretical construct. The primary aim of the theory was the development of new instructional procedures, but that aim could be assisted by accurate, independent indicators of cognitive load. Since the load on working memory is central to the theory, independent measures of that load are important.

In this chapter, I will consider some of the issues associated with the measurement of cognitive load. I will begin with a brief history of work by cognitive load theorists to measure the construct.

A Brief History of Attempts to Measure Cognitive Load

In some respects, attempts to measure cognitive load during instruction have been unusual. In early efforts to use cognitive load as an explanatory variable for the effectiveness and efficiency of instructional design, directly and independently measuring cognitive load was either nonexistent or, at best, peripheral. The aim of the theory was not to measure cognitive load but rather to provide effective instructional procedures based on human cognitive architecture. To achieve this aim, aspects of human cognition that might potentially be relevant to instruction were identified. Possible candidates for instructional relevance were used to generate experimental hypotheses that could be tested using randomized, controlled trials. If the experimental results were in accord with the theory, it was assumed that the results were due to the variables that were the subject of the hypotheses. Direct, independent measures of cognitive load were rarely used.

Nevertheless, as early as Owen and Sweller (1985) and Sweller (1988), attempts were made to determine differences in cognitive load due to different instructional procedures, albeit that the measures were somewhat indirect. For example, Owen and Sweller (1985) used differences in the number of mathematical errors during problem solving as a proxy for differences in cognitive load, while Sweller (1988) constructed computational models and counted numerical differences in the way the models functioned under different instructional conditions to indicate differences in load. In an experiment, he also used a secondary task as a more direct, independent measure. As can be seen in this volume, secondary tasks provide an important source of cognitive load measurement.

For several years following this period, most of the instructional effects generated by cognitive load theory used cognitive load as a theoretical construct to generate instructional hypotheses and to explain empirical results without attempting to independently verify that differences in instructional effectiveness and efficiency really were due to differences in load. That situation changed substantially with the publications of Paas (1992) and Paas and van Merriënboer (1994). Their introduction of a subjective rating scale to determine cognitive load was influential and heavily adopted. Today, various rating scales are the most commonly used independent measures of cognitive load. In addition, alternatives such as secondary tasks have continued to be studied (Brunken, Plass, & Leutner, 2004; Brunken, Steinbacher, Plass, & Leutner, 2002) and indeed, continue to be developed to this day (Park & Brunken, 2014).

As well as rating scales and secondary tasks, there have been many attempts to use various physiological measures, beginning with Paas and van Merriënboer (1994), although these attempts have met with limited success. Physiological measures may be able to distinguish between very large differences in cognitive load such as differences between resting and concentrating, but they may be insufficiently sensitive to detect differences between instructional procedures. Attempts to find a sufficiently sensitive measure are ongoing.

Categories of Cognitive Load

A major issue associated with independent measures of cognitive load concerns attempts to distinguish between categories of cognitive load. Initially, cognitive load was treated as a single, undifferentiated concept, and so a single measure of load was all that was required. Subsequently, it became clear that a distinction needed to be made between intrinsic and extraneous cognitive load, where extraneous cognitive load could largely be equated with the original, single concept due to instructional factors that can be altered, while intrinsic cognitive load was defined as being due to the basic nature of the information being processed (Sweller, 1994; Sweller & Chandler, 1994). Later (Sweller, van Merriënboer, & Paas, 1998), germane cognitive load was added to account for the effort needed to learn.

Sweller (2010) explained all forms of cognitive load in terms of element interactivity. Some elements of information can be processed in isolation without reference to other elements and so impose a low working memory load. Other elements interact, and interacting elements must be processed together, resulting in an increase in working memory load. If interacting elements are intrinsic to the information being processed and so are unavoidable, then an intrinsic cognitive load is imposed. If multiple elements interact because of an instructional design choice that can be changed, then an extraneous cognitive load is imposed. Germane cognitive load can be defined in terms of the working memory resources devoted to dealing with intrinsic cognitive load. The more resources that deal with intrinsic cognitive load and the less devoted to dealing with extraneous cognitive load, the higher the germane load. Under this definition, germane cognitive load is dependent on the relation between intrinsic and extraneous load and is not an independent source of load. In effect, it summarizes the relation between intrinsic and extraneous load.

There are some advantages to this formulation. Treating germane cognitive load as an independent source of load rather than as an adjunct to intrinsic load ran the risk of leading to logical conundrums. For example, the aim of instruction should be to replace extraneous load that interfered with learning with germane load that facilitated learning, but if the only effect of substituting poorer instructional designs by better ones is to substitute extraneous load by an equivalent germane load, then the overall load should not vary. In fact of course, a very large number of studies indicated that better instructional design lowered overall load. This problem disappears if, instead of being an independent source, germane cognitive load refers to working memory resources that are devoted to intrinsic rather than extraneous factors. A decrease in extraneous cognitive load with intrinsic cognitive load constant reduces overall load in accord with the empirical results. In addition, a decrease in extraneous cognitive load results in fewer working memory resources being devoted to extraneous factors and more resources being available to deal with intrinsic factors, resulting in enhanced learning due to germane cognitive load. By defining germane cognitive load in terms of working memory resources devoted to intrinsic factors with a consequent reduction of resources devoted to extraneous factors, we allow a reduction in extraneous load, an increase in germane load, and a decrease in overall load.

The reason that intrinsic cognitive load was originally added to extraneous load needs to be noted. The requirement for the concept of intrinsic cognitive load was suggested by experimental results. A series of experiments (Sweller & Chandler, 1994; Tindall-Ford, Chandler, & Sweller, 1997) indicated a failure to obtain cognitive load effects using some categories of information. Specifically, cognitive load effects only seemed to be obtainable using high element interactivity information. Using low element interactivity information, the instructional procedures used appeared not to matter. It was concluded that unless the information was intrinsically complex and so difficult to understand, cognitive load effects would not be obtained. Intrinsic and extraneous cognitive load are additive. Unless intrinsic cognitive load is high, the addition of extraneous cognitive load may not exceed working memory limits, and so instructional design considerations due to cognitive load may not apply.

Implications of Categories of Cognitive Load for Measurement

The distinction between categories of cognitive load had implications for the measurement of the construct. Given the existence of theoretical categories of cognitive load, there has been considerable interest in obtaining independent measures of those categories. For example, instead of measuring overall cognitive load consisting of extraneous and intrinsic load, we might attempt to measure intrinsic and extraneous load independently.

For many years, I argued at various meetings that it might be impossible to find independent measures of categories of load. (I do not recall indicating this view in writing but may have done so.) Secondary tasks are hardly likely to be able to make the necessary distinctions, and physiological measures have difficulty distinguishing between the overall load of different instructional procedures, let alone fractionating the overall load into categories. Theoretically, subjective ratings of cognitive load might be able to distinguish between differently worded questions concerning load, but I argued that, in order to work, they required learners to know themselves why they found a task difficult. To me, it seemed unlikely that learners who, by definition, did not have knowledge of an area could inform us that they felt a learning task was imposing a heavy cognitive load because it was intrinsically difficult or because the instructional procedures used were deficient. Instead, I argued that the only way to determine whether the cognitive load imposed belonged to the extraneous or intrinsic category was to run randomized, controlled experiments in which, for example, intrinsic cognitive load was kept constant and extraneous load was varied or vice versa. Any learning differences thus had to be due to the factor that was changed but not the factor that remained constant. Of course, the view that distinguishing between categories of cognitive load required randomized controlled trials rather than psychometric procedures only could be maintained in the absence of empirical evidence that learners actually could distinguish between categories of cognitive load.

While it is still preliminary, it is distinctly possible that the required evidence may be available. Leppink, Paas, van Gog, van der Vleuten, and van Merriënboer (2014) asked learners a series of questions similar to “The topics covered in the lecture were very complex” to determine intrinsic cognitive load as opposed to questions such as “The instructions and explanations during the lecture were very unclear” to determine extraneous cognitive load. They determined that these two question categories reflected different factors and that the scales could be used to differentiate between intrinsic and extraneous cognitive load. These results are promising and suggest that, at least under some circumstances, it is possible for learners to distinguish task complexity sourced from intrinsic and extraneous factors.

Of course, considerably more work needs to be done. It still remains to be seen whether learners can determine whether the source of their difficulty is due to intrinsic or extraneous factors if they are unfamiliar with the information being processed and unfamiliar with the techniques that cognitive load theory has devised to make instructional material less demanding. They may not be in a position to consider the possibility of presenting worked examples, reducing split-attention, eliminating redundancy, using dual-mode presentations, reducing the use of transient information, altering instructional procedures depending on levels of expertise, or any of the many other instructional procedures based on cognitive load theory designed to reduce extraneous cognitive load. In other words, learners may not feel that instructions are unclear if they have no concept of, for example, split-attention. Nevertheless, while it is too early to come to a definitive conclusion, the currently available Leppink et al. (2014) data provide optimism for the results of further research.

Conclusions

Independent measures of cognitive load have been a valuable addition to cognitive load theory and an active, ongoing area of research interest for many years. Because of its effectiveness and ease of use, the Paas (1992) scale has been the ...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Contents
• Foreword by Paul Chandler
• About this Book
• Acknowledgements
• PART I: Theoretical Perspectives on Cognitive Load Measurement
• PART II: Methodology in Cognitive Load Measurement and Application
• PART III: Practices in Cognitive Load Measurement
• Concluding Remarks
• About the Authors
• Index