What is cognitive load theory?
Cognitive load theory began as a claim about teaching algebra. Its worked-example, split-attention and expertise findings came from classrooms and training rooms, and its categories are still disputed.
Answer to the question
What is cognitive load theory?
Cognitive load theory is an instructional design theory built on the limits of working memory. John Sweller set it out in 1988, arguing that new information must pass through a small working memory before it is stored. Experiments on worked examples, split-source diagrams and learner expertise followed, and its categories of load are still debated.
Key takeaways
- Sweller (1988) argued that means-ends analysis, a common problem-solving strategy, absorbs so much working memory that little is left for acquiring schemas.
- Sweller and Cooper (1985) reported five algebra experiments: worked examples took less time to process and led to faster, more accurate solving, but only on problems with the same structure.
- Chandler and Sweller (1991) reported six experiments favouring integrated over split-source instruction, the pattern now called the split-attention effect.
- Kalyuga and colleagues (2003) reviewed evidence that methods which help novices can lose their value, or hurt, for experienced learners.
- de Jong (2010) credited the theory with explaining many findings and set out conceptual, methodological and practical problems with it.
Cognitive load theory is an account of how instruction should be designed around the limits of working memory. It began as a claim about teaching algebra, well before anyone applied it to screens or checkouts. The Australian educational psychologist John Sweller set out its first full statement in a paper in the journal Cognitive Science. The central idea is that new information has to pass through a working memory of small capacity and short duration before it reaches long-term memory, and that the way material is presented can use up that capacity before any learning has happened. Four decades of experiments have produced named effects, shifting terminology and a sustained critique from inside the field.
A Bottleneck and a Store
Cognitive load theory rests on two memory systems that behave differently. Working memory holds what a person is thinking about at this moment, and it holds little. George Miller's 1956 paper in Psychological Review placed the limit near seven items, and Nelson Cowan's 2001 reconsideration in Behavioral and Brain Sciences argued for a lower figure of about four chunks. Long-term memory has no comparable ceiling. Working memory also figures in listening, a link followed in Why You Hear Your Name in a Crowd.
A 2019 review by Sweller and two co-authors states the claim of cognitive load theory directly. Novel information passes through a working memory limited in capacity and duration before it is stored, and once it has been stored it is no longer subject to those limits. Instruction, on this view, is a problem of getting material across the bottleneck without wasting the space.
Nicholas Carr carried the theory to a general readership in The Shallows, where Miller's capacity limit and Sweller's load theory support his argument about reading on the web. The step from classroom materials to hyperlinks is Carr's extension of cognitive load theory, and it is not a result of the worked-example and split-attention experiments described below, which used algebra, electrical engineering and biology materials. What Cognitive Load Costs at Checkout follows the theory into shopping and consent screens, and Does Screen Reading Cost Comprehension takes up the reading studies.
Can Examples Replace Solving?
Sweller's 1988 paper made a pointed argument about the most familiar way of teaching technical subjects, which is to set problems and let students solve them. Within cognitive load theory this is the founding complaint. Sweller argued that expertise in problem solving depends mainly on knowledge organised as schemas, and that conventional problem solving is a poor way to acquire them.
His explanation was a mismatch between the two activities. Means-ends analysis, a common strategy in which the solver keeps comparing the current state with the goal and searching for a step that narrows the difference, absorbs so much of the solver's limited capacity that little remains for learning how the problem is built. Sweller reports a computational model and experiments in support, and says they bear out the argument. The implication for teaching is that practice at solving, the standard method, may build schemas less efficiently than other uses of the same time.
The alternative that followed within cognitive load theory was the worked example, a solved problem the student studies instead of attempting. In 1985 Sweller and Graham Cooper reported five experiments on algebra in the journal Cognition and Instruction. The first compared Year 9 pupils, Year 11 pupils and university mathematics students. The more experienced groups recalled equations better and were better at telling apart equations that look alike but call for different solution methods, which, the authors say, points to experience building richer mental representations.
The other four experiments asked whether studying worked examples could build that knowledge. The authors report that the examples took less time to process and that students who then met similar problems solved them faster, with fewer mathematical errors. The gains were confined to problems with the same structure as the examples. Sweller and Cooper concluded that novices store general algebra rules as a limited set of problem-specific schemas, and that turning these into broader rules probably takes sustained practice across a wider range of problem types.
Layouts That Slow Experts Down
A page can add load without adding content, and Paul Chandler and Sweller tested that in 1991 in Cognition and Instruction. They compared split-source instruction, in which related material such as a diagram and its explanatory text is placed apart, with integrated instruction, in which the two are combined. Their argument was that split sources force the learner to integrate the material mentally before learning can begin, which is a load that serves no purpose.
The authors report six experiments using electrical engineering and biology materials. They describe the first as running over several months in an industrial training setting, with materials that could not be understood without mental integration, and say the results favoured the integrated version. The pattern became known as the split-attention effect. The design lesson the authors drew is to bring related material together, so a label sits where the reader is looking and not in a legend elsewhere.
If the effects depended only on layout, a good layout would be good for everyone. A 2003 paper in Educational Psychologist by Slava Kalyuga and three co-authors, Chandler and Sweller among them, reviews evidence that it is not. Techniques that reduce working memory load help learners build and automate schemas, but they work differently depending on expertise. Methods that help novices can lose their value, and can even hurt, when used with more experienced learners. The authors named this the expertise reversal effect.
One plausible mechanism is that guidance an expert no longer needs becomes one more thing to process alongside what the expert already knows, though the authors' claim concerns the pattern and not that account. Either way, the finding changes what cognitive load theory can promise. No format is good in itself, only good for a learner at a particular stage.
Taken together, the four studies show that none of these effects stands alone. Each holds for a particular kind of material and a particular kind of learner.
Who Disputes the Load Categories?
Cognitive load theory sorts the demands on working memory into kinds. Intrinsic load comes from the element interactivity inherent in a task, meaning how many elements must be held and related at once. In a 2010 paper in Educational Psychology Review, Sweller argued that element interactivity also underlies extraneous load, the load imposed by how instruction is designed, and that whether interactivity counts as extraneous depends on whether the task needs it or the design produced it. Germane load is defined in that paper in terms of intrinsic load.
| Kind of load | How the 2010 paper describes it |
|---|---|
| Intrinsic | Comes from the element interactivity inherent in a task |
| Extraneous | Imposed by how instruction is designed, and tied to whether the task needs the interactivity |
| Germane | Defined in terms of intrinsic load |
The same year, Ton de Jong published a review of the theory in Instructional Science. He credited the theory with explaining many experimental findings and then set out conceptual, methodological and practical problems, closing with a research agenda. The terms of load are still being argued over, so a reader who meets intrinsic, extraneous or germane load in an older source, or in a design guide that borrows them, should check how that source defines each one before relying on it.
Common questions
What is cognitive load theory?
Cognitive load theory is an instructional design theory built on the limits of working memory. John Sweller set it out in 1988. It holds that new information must pass through a small working memory before it is stored, so material should be presented in ways that do not waste that capacity.
Who developed cognitive load theory?
John Sweller introduced it in a 1988 paper in Cognitive Science. Later work with Graham Cooper on worked examples, with Paul Chandler on split-source instruction and with Slava Kalyuga and others on expertise extended it, and the theory has been revised since.
What is the split-attention effect?
It is the finding that learning suffers when related material, such as a diagram and its text, is presented apart so that the learner must combine them mentally. Chandler and Sweller reported it in 1991, and integrated formats favoured learners in their first experiment.
Sources
Carr, N. (2010). The Shallows. Seed book; vault hub Books/The Shallows.
Sweller, J. (1988). “Cognitive load during problem solving: Effects on learning.” Cognitive Science, 12(2), 257-285. doi:10.1016/0364-0213(88)90023-7 https://doi.org/10.1016/0364-0213(88)90023-7
Sweller, J., & Cooper, G. A. (1985). “The use of worked examples as a substitute for problem solving in learning algebra.” Cognition and Instruction, 2(1), 59-89. doi:10.1207/s1532690xci0201_3 https://doi.org/10.1207/s1532690xci0201_3
Chandler, P., & Sweller, J. (1991). “Cognitive load theory and the format of instruction.” Cognition and Instruction, 8(4), 293-332. doi:10.1207/s1532690xci0804_2 https://doi.org/10.1207/s1532690xci0804_2
Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). “The expertise reversal effect.” Educational Psychologist, 38(1), 23-31. doi:10.1207/S15326985EP3801_4 https://doi.org/10.1207/S15326985EP3801_4
Sweller, J. (2010). “Element interactivity and intrinsic, extraneous, and germane cognitive load.” Educational Psychology Review, 22(2), 123-138. doi:10.1007/s10648-010-9128-5 https://doi.org/10.1007/s10648-010-9128-5
de Jong, T. (2010). “Cognitive load theory, educational research, and instructional design: some food for thought.” Instructional Science, 38(2), 105-134. doi:10.1007/s11251-009-9110-0 https://doi.org/10.1007/s11251-009-9110-0
Sweller, J., van Merrienboer, J. J. G., & Paas, F. (2019). “Cognitive architecture and instructional design: 20 years later.” Educational Psychology Review, 31(2), 261-292. doi:10.1007/s10648-019-09465-5 https://doi.org/10.1007/s10648-019-09465-5
Miller, G. A. (1956). “The magical number seven, plus or minus two: Some limits on our capacity for processing information.” Psychological Review, 63(2), 81-97. doi:10.1037/h0043158 https://doi.org/10.1037/h0043158
Cowan, N. (2001). “The magical number 4 in short-term memory: A reconsideration of mental storage capacity.” Behavioral and Brain Sciences, 24(1), 87-114. doi:10.1017/s0140525x01003922 https://doi.org/10.1017/s0140525x01003922