Cognitive load theory and instructional consequences The human cognitive architecture on which cognitive load theory is based

assumes that most cognitive activities are driven by a large store of infor-mation held in long-term memory (the inforinfor-mation store principle). This knowledge acts as a central executive for working memory, directing the manner in which information is processed (the environmental organizing and linking principle). The ultimate goal of instruction is the alteration of long-term memory via the borrowing and reorganizing principle. In other words, for learning to occur, novel material must be organized and incorporated into long-term memory via a limited working memory. For instruction to be effective, it has to be designed in ways in which the limitations of the working memory are overcome. Many instructional materials and techniques may be ineffective because they ignore the lim-itations of human working memory and impose a heavy extraneous cog-nitive load. Cogcog-nitive load theory distinguishes between three kinds of cognitive load: intrinsic, germane and extraneous load.

Intrinsic cognitive load is related to task difficulty and is due to the complexity of the information that must be processed. It is determined by the levels of element interactivity inherent in the learning material.

If element interactivity is low, so is working-memory load. For example, in learning to translate nouns of a foreign language, each translation can be learned independently of every other translation; learning to translate the word potatodoes not depend on learning to translate the wordegg.

In contrast, some material consist of elements that interact in the sense that one element cannot be meaningfully learned without learning many other elements simultaneously. For example, in learning the English word order for the phrase inside the old building, it is not sufficient to attend

to individual words to decide thatthe inside old buildingis not appropri-ate. All the words in the phrase and the relations among them must be considered simultaneously because they interact. Element interactivity in this case is high, resulting in a high intrinsic cognitive load. Process-ing high-element-interactivity material imposes a high workProcess-ing-memory load. Because intrinsic load is inherent in the learning material, it cannot be manipulated without compromising understanding.

Germane load (Paas and van Merri¨enboer 1993) is the cognitive load caused by effortful learning due to attentional (working-memory) resources being directed to intrinsic cognitive load. For instance, giv-ing learners lots of examples to demonstrate a point increases germane cognitive load although it is likely to assist in learning.

Extraneous cognitive load is caused by inappropriate instructional designs that fail to take into consideration the limitation of working-memory resources that are necessary for learning. This cognitive load is important in multimedia learning because of the cognitive effort required to process different sources of information. Theoretically, extraneous cognitive load can be alleviated by overcoming the limitations of working memory in two ways. First, instructional procedures can alleviate extra-neous cognitive load by formatting instructional material in such a way as to minimize cognitive activities that are unnecessary to learning so that cognitive resources can be freed to concentrate on activities essential to dealing with intrinsic cognitive load. Second, working memory can be increased in capacity by taking the advantage that is offered by dual modality presentations.

Intrinsic, germane and extraneous cognitive load are additive. In any instructional situation, the aim is to reduce extraneous cognitive load imposed by an inappropriate presentation format. Reducing extraneous cognitive load frees working-memory capacity, permitting an increase in germane cognitive load which is necessary when dealing with complex material that imposes a high intrinsic cognitive load.

In the following sections, we focus on three cognitive load effects con-cerned with aspects of multimedia presentation of information: the split-attention effect, the redundancy effect and the modality effect.

4.5.1 The split-attention effect

Split attention occurs when learners are required to integrate multiple sources of information that are physically or temporally separate from each other where each source is essential for understanding the material.

The working-memory load imposed by the need to mentally integrate disparate sources of information may interfere with learning. The first

A 60°

40° C

D B

E

In the above Figure, find a value for Angle DBE.

Solution:

Angle ABC = 180° − Angle BAC − Angle BCA (Internal angles of a triangle sum to 180°)

= 180° − 60° − 40°

= 80°

Angle DBE = Angle ABC (Vertically opposite angles are equal)

= 80°

Figure 4.1 A conventional, split-attention geometry example

study on split attention was reported by Tarmizi and Sweller (1988), who used geometry worked examples in their attempt to replicate the effective-ness of worked examples in algebra learning (Cooper and Sweller1987;

Sweller and Cooper 1985) and in other mathematical-related domains (Zhu and Simon1987). Surprisingly, using worked examples did not pro-duce better learning outcomes than using conventional problem-solving strategies in their initial geometry experiment. They argued that the requirement to mentally integrate the two sources of information (dia-gram and textual solutions) due to the conventionally structured format of the worked examples (seefigure 4.1) must have imposed an increase in cognitive load that prevented cognitive resources being used for learning.

When diagrams and the statements required to understand the dia-grams are physically separate, as normally occurs in conventionally struc-tured geometry worked examples, working-memory resources must be expended to mentally integrate the two sources of information, reducing

A 60°

40°

80°

1

2

180° − 60° − 40° = 80°

B

E

D C

Figure 4.2 A physically integrated geometry example

the effectiveness of worked examples. In subsequent experiments, when the format of presenting geometry worked examples was altered to an integrated version in which statements were physically integrated with the diagram (seefigure 4.2), studying worked examples proved more effective than a conventional problem-solving strategy. The integrated format alle-viated the extraneous cognitive load imposed by the requirement to split attention between diagrams and text, freeing working-memory resources to attend to processes that facilitate learning.

Tarmizi and Sweller’s (1988) findings have led to a number of studies that have extended the split-attention effect to the learning of coordinate geometry (Swelleret al.1990), physics (Ward and Sweller1990), mate-rial designed for training of electrical apprentices (Sweller and Chandler 1991) and learning in a computer environment (Sweller and Chandler 1994; Chandler and Sweller 1996). Similarly, Mayer and his research associates have demonstrated that split attention could also occur with temporal separation, thus leading to unnecessary extraneous cognitive load (Mayer and Anderson1991,1992; Mayer and Sims1994; Moreno and Mayer1999). In the area of language learning, Yeung, Jin and Sweller (1998) found that the integrated format, which combined explanatory notes with reading passages, was helpful in reducing the cognitive load related to vocabulary search and facilitating the process of reading com-prehension for young native speakers as well as inexperienced English as a second language (ESL) learners. More recently, Hung (2007) found that

reducing split attention facilitated learning by undergraduate geography students studying ESL.

The split-attention effect has implications for instructional design in a multimedia context where there will inevitably be at least two sources of information. Research suggests that multiple sources of information need to be integrated into an optimal format to minimize extraneous cognitive load. However, those sources of information must be complementary to each other rather than mere repetition of the same information (seenext sectionon the redundancy effect). In addition, as pointed out by Sweller (1994), an integrated format is only effective when the learning material has high-element interactivity (i.e., the elements in the learning content must be simultaneously processed because they interact). Under this con-dition, the intrinsic cognitive load is relatively high and the employment of an integrated method can be helpful. On the other hand, if the learn-ing material has low-element interactivity (i.e., elements in the learnlearn-ing content have little interaction and thus can be processed individually), there is a low intrinsic cognitive load. In this case, using an integrated format is unlikely to lead to a noticeable impact on learning outcomes. A further factor to consider when integrating different sources of informa-tion is learner characteristics that interact with material characteristics.

Material that is not intelligible in isolation and is high in element inter-activity for low-knowledge learners may be intelligible in isolation and low in element interactivity for learners with more knowledge. For high-knowledge individuals, physical integration may be harmful because of the redundancy effect (Kalyuga, Chandler and Sweller1998; Yeung, Jin and Sweller1998), discussed next.

4.5.2 The redundancy effect

The redundancy effect occurs when additional information presented to learners results in negative rather than positive effects on learning. In mul-timedia learning, it is not unusual for an instructor or a web-based course designer to consider providing the same information in different formats or presenting additional detailed information for the topic to enhance learning. Intuitively, it is easy to assume that such additional information will not produce negative learning outcomes. However, cognitive load theory suggests the possibility of negative effects due to redundancy.

Because working memory is extremely limited in terms of its capacity and duration, when the learners are exposed to material containing the same information in multiple forms or some unnecessary elaborations, their cognitive functioning can be inhibited and their learning process may be negatively affected. For instance, to teach pupils to learn a noun,

say, the name of an animal (horse), a teacher can write the word on a flashcard or on the whiteboard and read it (presenting essential informa-tion via both visual and audio channels). One may further assume that it is a good idea to show pupils a flashcard containing both the word and picture and at the same time provide the pronunciation. Such a type of instruction sounds more ‘interesting’ and more ‘informative’, and can be easily accomplished using modern information technology. However, controlled experiments designed to test the effect of adding pictures to words when learning to read have indicated that reading outcomes are superior using written words plus sounds rather than presenting written words, pictures and sounds simultaneously (Miller1937; Solman, Singh and Kehoe1992;Torcasio and Sweller 2010). This phenomenon can be explained by assuming that adding pictures to the presentation of written words and sounds is redundant, and that redundancy can have a nega-tive impact on the effecnega-tive use of limited working memory to process and transfer information to long-term memory. Redundancy imposes an extraneous cognitive load.

In a multimedia learning context, the redundancy effect has been demonstrated in a number of studies (Mayer2001; Sweller2005b). In one study, a series of experiments were carried out to test the redundancy effect in a computer course (Sweller and Chandler1994; Chandler and Sweller1996). The learners were divided into two groups: participants in one group worked on a computer with the assistance of a computer manual that combined text with diagrams; participants in the other group simply learned using the computer manual either on a screen or in hard-copy but did not actually work on a computer. Because the act of working on the computer was largely irrelevant to the real task of understanding the program, participants in the first group who had to work on a com-puter could not use working memory efficiently to transfer the knowl-edge of programming into their long-term memory. In other words, the computer work was redundant and occupied working-memory resources that otherwise could be used by learners to assimilate the appropriate information. In another series of experiments on the use of computer manuals, researchers reported that computer manuals with minimized explanatory text were more effective and user-friendly than conventional manuals (Carroll1990; Carrollet al.1987). Similarly, researchers have found that a reduced or summarized text was superior to a full text (in which some parts may be redundant) in terms of learning outcomes such as lower error rates, higher retention rates and shorter acquisition times (Mayeret al.1996; Reder and Anderson1980,1982).

In research on learning English as a foreign language (EFL), Hirai (1999) noticed that, for less proficient Japanese EFL learners, their

listening rate was far behind their reading rate. Diao and Sweller (2007) and Diao, Chandler and Sweller (2007) thus proposed that, if novice EFL learners were exposed to both auditory and visual information for reading comprehension, such an audio-visual presentation might result in a redundancy effect. They tested this hypothesis and found that the Chinese ESL learners who were exposed to simultaneous presentations of spoken and written text had a higher mental load and produced lower test scores in both word decoding and reading comprehension in com-parison with those who were given written information only. In a similar vein, Kalyuga, Chandler and Sweller (2004) found that the simultaneous presentation of written and spoken text during a presentation imposes an extraneous cognitive load. Learning is enhanced by presenting in one modality only.

As is the case with the split-attention effect, the redundancy effect has been demonstrated in a variety of contexts. It is not just diagrams and redundant text that can be used to demonstrate the redundancy effect.

While diagrams are frequently more intelligible than the equivalent text, there are instances where any one of diagrams, the presence of equipment or auditory information have been found to be redundant (see Sweller 2005b for experimental evidence). In other words, what is redundant depends on what is being taught. The redundancy effect provides a sim-ple guideline for instructional design in practice: eliminate any redundant material in whatever form presented to learners and any redundant activ-ity that instruction may encourage learners to engage in. However, this guideline alone does not indicate exactly what material may or may not be redundant. This guiding principle needs to be considered in conjunction with cognitive load theory. The theory can be used to provide guidance concerning the conditions that determine redundancy and hence what material is likely to be redundant. For instance, in deciding whether text should be added to a diagram, the instructional designer needs to con-sider several factors. Is the diagram intelligible on its own? If so, the text may be redundant. Does the text provide essential information? If so, it is not likely to be redundant and should be retained. Is there a high level of element interactivity within the text, that is, to understand one element, must one consider many other elements at the same time? If so, as far as possible, diagrams should not be presented with the text to avoid the risk of overloading working memory. Another factor to consider is learner expertise. Whether information is high in element interactivity and whether it is intelligible on its own depends largely on the learner.

Information that is intelligible for more expert learners may not make sense to novices who require additional explanatory material. In short, whether or not additional material is redundant can be determined by

considering the cognitive load implications of that material in the context of learner expertise.

4.5.3 The modality effect

It was noted earlier that the limitations of working memory can be overcome either by formatting instructional materials in a manner that minimizes extraneous cognitive load so that cognitive resources can be released to attend to processes useful for learning or by expanding effec-tive working-memory capacity. While both the split-attention effect and the redundancy effect fall into the first category of minimizing extrane-ous cognitive load, the modality effect falls into the latter category of expanding working-memory capacity. The modality effect occurs when information presented in a mixed mode (partly visual and partly audi-tory) is more effective than when the same information is presented in a single mode (either visually or in auditory form alone). Mousavi, Low and Sweller (1995) tested this hypothesis in educational settings, in which geometry problems and related instructions were used. There were two data presentation conditions: audio-visual and visual-visual. In the audio-visual presentation, diagrams were given as visual information and the related text was provided as audio input, whereas in the visual-visual presentation, both diagrams and associated text were in a visual-visual format. The data obtained from this series of experiments demonstrated that learners in the audio-visual group performed much better than did those in the visual-visual group.

From a cognitive load theory perspective, the modality effect can be explained by assuming the memory load due to a picture with written text presentation induces a high load in the visual working-memory sys-tem because both sources of information are processed in this syssys-tem.

In contrast, the diagram and narration version induces a lower load in visual working memory because auditory and visual information are each processed in their respective systems. Therefore, the total load induced by this version is spread between the visual and the auditory components in the working-memory system. In other words, the use of audio and visual information may not overload working memory if its capacity is effectively expanded by using a dual-mode presentation as opposed to a single-mode presentation.

This basic modality effect was confirmed in a number of subse-quent studies. For instance, Tindall-Ford, Chandler and Sweller (1997) reported increased effective working memory and improved learning outcomes under audio-visual conditions in comparison with visual-visual conditions in electrical engineering courses in which the learning

material was high in element interactivity. Adopting the scale recom-mended by Paas and van Merri¨enboer (1993), Tindall-Ford and col-leagues found that the cognitive load was lower under audio-visual con-ditions than visual-visual concon-ditions for learning such material. Applying the modality principle, Jeung, Chandler and Sweller (1997) reported improved learning outcomes by using visual indicators to highlight the most complex parts of information in the spoken text. In an industrial training course, beginners’ learning experience was enhanced by dual-mode presentations delivered by the instructor (Kalyuga, Chandler and Sweller2000).

The modality effect is especially important in the context of multime-dia learning because the instructional medium involves different presen-tation modes and sensory modalities. Multimedia instruction is becom-ing increasbecom-ingly popular and findbecom-ings associated with the modality effect that can be interpreted within a cognitive load framework can provide a coherent theoretical base for multimedia investigations and applications.

Using web-based or computer-aided instructional design, Mayer and his colleagues (Mayer and Moreno1998; Moreno and Mayer1999; Moreno et al.2001) tested the modality effect in a number of courses. In gen-eral, they found that students learned more when scientific explanations were given as pictures plus narration (or spoken text) than under the condition of pictures together with on-screen text. According to Mayer’s (2005) interpretation, when learners are dealing with pictures and related on-screen text, their visual channel may become overloaded while their auditory channel is unused. When words are narrated or the spoken text is provided, the learners can use their auditory channel to process such information, and the visual channel will deal with the pictures only.

The redistribution of information flow can lead to enhanced multimedia learning. When the information contained in a picture is too complex, simultaneous presentation of corresponding auditory information may still be beyond the capacity of working memory. In this case, a

The redistribution of information flow can lead to enhanced multimedia learning. When the information contained in a picture is too complex, simultaneous presentation of corresponding auditory information may still be beyond the capacity of working memory. In this case, a