This post aims to consider cognitive load theory and what considerations should be drawn from it in the design of electronic instructional materials. Sweller (2008) discusses several strategies for harnessing the principles of CLT in e-learning design. Several of these strategies are described by Clark and Mayer (2008), so overlap between both are discussed in tandem below. Mayer’s multimedia learning model (Mayer 2005) is used here as the underlying framework for the principles discussed. Before these are discussed, there is a brief explanation of what CLT is, along with the processes involved in learning new information.
What is Cognitive Load Theory?
Cognitive load theory (CLT) is model for instructional design based on knowledge of how we acquire, process and retain new information. It proposes that a successful use of the model will result in more effectual learning, and the retaining of information in the long term memory, which can be recalled when required in a given context. The theory distinguishes three types of cognitive load (Sweller 2008, Ayres and Paas, 2009):
- Intrinsic load is caused by the complexity of the material. This depends on the level of expertise of the learner – in other words it depends on the learner’s understanding of the subject.
- Extraneous load depends on the quality or nature of the instructional materials. Poor materials or those that require a large amount of working memory to process will increase the load and leave little capacity for learning.
- Germane load is the mental effort required for learning. Because of the limited capacity of the working memory, germane load (the extent of learning) is dependent on the extent of the extraneous load, and also on the material and expertise of the learner – the intrinsic load. An expert on a topic is able to draw from prior knowledge, and release working memory capacity for germane load processing.
The mechanism of information processing was summarised succinctly by Mayer for the purposes of multimedia learning. This is similar in many respects to the information processing model familiar to many chemists through the work of Alex Johnstone (Johnstone, Sleet and Vianna 1994, Johnstone 1997). Mayer’s model is shown in the figure below (Clarke and Mayer 2008).
Information is presented to users in the form of words and pictures (there are other channels too, but these are the most pertinent to e-learning). The user senses these (what Johnstone refers to as a perception filter) and some of this is processed in the working memory, which can hold and process just some information at any time (this can be quantified using the M-capacity test). If this material can be related to existing prior knowledge, it is integrated with it, and effective learning occurs – the new experiences/information are stored in the long-term memory.
Considerations for Presentation of Information
Learning materials that provide two sources of mutually dependent information (e.g. audio and visual) will require the learner to process both channels and integrate them, requiring working memory. Design of the materials should therefore ensure that as, for example, a reference to the diagram is being verbalised, the associated diagram reference is clear for the viewer to see. The alternative is that learners require working memory to process the diagram to find the reference being verbalised. Clark and Mayer call this the contiguity principle, and provide two strategies for considering it in practice, namely to place printed words near the corresponding graphics (including, for example, feedback on the same screen as the question and integration of text legends) and to synchronize spoken words with the corresponding graphics.
Because the working memory has “channels”, the most significant being the visual/pictorial and auditory/verbal information channels, consideration of the nature or mode of information can be beneficial. In the split-attention effect, above, it was argued that they different modes must be integrated effectively to ensure that working memory was not overloaded. This can be teased out a little further. If learning material contains a diagram and explanation, (mutually dependent), the explanation can be in text or audio form. Presenting the explanation in text form means that learners’ visual/pictorial channel will be overloaded more quickly, as they must process the diagram and read the text. If the text is presented as audio, both channels are being used effectively. Clark and Mayer also discuss the modality principle, advising that words should be presented as audio and not on screen. However, they limit it to situations where there are mutually dependent visual/auditory information being presented (see below). Additionally, they argue that there are occasions where text is necessary – for example a mathematical formula or directions for an exercise, that learners may need to reread and process.
The split-attention and modality effects considered mutually dependent information. If there is multiple representations of the same material, each self-sufficient, or if there is material of no direct use to learning, it can be considered redundant. The time required to attend to unnecessary information or process multiple versions of the same information means that the working memory capacity is reduced. Clark and Mayer also discuss the redundancy effect, especially recommending that on-screen text is not used in conjunction with narration, except in situations where there are no diagrams, or the learner has enough time to process pictures and text, or the learner may have difficulty processing the speech.
Consideration for Design of Interactions
1. Worked Examples
Worked examples have been shown to reduce cognitive load. The reason is that students who were exposed to worked examples and who then were required to solve problems did not need to spend extraneous load on the process of solving the examples, and could concentrate on the problem itself instead. Clark and Mayer agree, and discuss five strategies for incorporating worked examples into e-learning instruction, including fading, below. (Crippen and Brooks (2009) have previously discussed the case for worked examples in chemistry.)
While the case for worked examples is strong, the situation becomes problematic when learners who are already expert engage with the material. In this scenario, their learning may be at best the same as solving problems without worked examples and at worst hampered by the presence of worked examples. The nature of delivery of material (considered for example in the split-attention and modality sections) can also differ for experts, as some material may become redundant. A potential solution offered by Sweller is to present learners with a partially completed problem and asked to indicate the next step required. The response was then used to direct the further instruction pathways.
Fading is related to worked examples, and involves a progressive reduction in the information presented in worked examples, so that learners are initially provided with many details on how to process a worked example, with the amount of guidance (scaffolding) reduced as more examples are provided. Clark and Mayer discuss this in some detail, and highlight it as a potential remedy for the expertise-reversal effect. For a three step problem, they propose that in the first worked example, all three steps are shown, and in each subsequent example, one step is left to the learner until they are required to complete an entire problem. They do acknowledge though that there is not yet sufficient evidence for how fast fading should proceed. Clarke and Mayer state that some students may not engage with the worked out components of a faded example, and propose that a worked out step of a faded example could be coupled with request requiring learners to state a reason/principle why a particular step was used. This is aimed to ensure learners are interacting with material that may otherwise be passive.
Having considered these principles, the next task is to implement them into a design framework. This will be discussed in a subsequent post.
Ayres, P. and Paas, F. (2009) Interdisciplinary perspectives inspiring a new generation of cognitive load research, Educational Psychology Review, 21, 1-9.
Clarke, R. C. and Mayer, R. E. (2008) E-Learning and the science of instruction, Pfeiffer (Wiley): San Francisco, 2nd Ed.
Crippen, K. C. and Brooks, D. W. (2009) Applying cognitive theory to chemistry instruction: the case for worked examples, Chemistry Education Research and Practice, 10, 35 – 41.
Johnstone, A. H., Sleet, R. J. and Vianna, J. F. (1994) An information processing model of learning: Its application to an undergraduate laboratory course in chemistry, Studies in Higher Education, 19(1), 77-87.
Johnstone, A. H. (1997), ‘…And some fell on good ground’, University Chemical Education,1, 8-13.
Mayer, R. E. (2005) Cognitive theory of multimedia learning, in Cambridge handbook of multimedia learning, R. E. Mayer, Cambridge University Press: Cambridge.
Sweller, J. (2008) Routledge: Human Cognitive Architechture, in Handbook of research on educational communications and technology, Spector, J. M., Merrill, M. D., van Merrienboer, J. and Driscoll, M. P., New York, 3rd Ed.