Cognitive load is the amount of working memory a task uses at a given moment. Because working memory can hold and juggle only a few new pieces of information at once, material that demands more than that — or is presented in a way that wastes capacity — is harder to understand and to learn. Cognitive load theory, developed by John Sweller from 1988, turns that limit into practical rules for teaching and design.
What cognitive load means
The theory starts from two parts of human memory. New information is handled first by working memory, which is limited in both capacity and duration. What we learn is stored in long-term memory, which has no known limit. The key asymmetry, in the words of the theory's own 2019 review, is that once information is stored in long-term memory the capacity and duration limits of working memory disappear when it is brought back to use.
That is why the same task can be heavy for one person and light for another. A beginner reading a chess position has to hold every piece separately; a strong player sees a few familiar patterns. The position has not changed — the load has.
The three types of cognitive load
| Type | Where it comes from | What to do about it |
|---|---|---|
| Intrinsic | The material itself: how many elements must be held in mind together to understand it | Cannot be removed, but can be sequenced — learn the parts first |
| Extraneous | The way it is presented: clutter, split sources, unneeded repetition, unclear instructions | Reduce it — this is where design helps most |
| Germane | The effort spent actually building reusable knowledge | Free up capacity for it |
The three-way split comes from the 1998 review by Sweller, Jeroen van Merriënboer and Fred Paas. Later work by Sweller recast it around element interactivity — how many elements must be processed simultaneously — and treated germane load less as a separate source to be added and more as the working memory devoted to the intrinsic material. For practical purposes the message is unchanged: keep what the material requires, cut what the presentation adds.
Where the theory came from
Sweller's 1988 paper in Cognitive Science asked why solving lots of problems is often a poor way to learn how to solve them. His answer: the usual strategy for an unfamiliar problem — working backwards from the goal and reducing the difference step by step — takes up so much processing capacity that little is left for learning the underlying pattern. Experts, he argued, differ from novices mainly in the stored patterns (schemas) they bring, and conventional problem solving is a slow way to acquire them.
What the research found
Several findings that teachers and designers now use daily came out of this work:
- Worked examples. In five algebra experiments, Sweller and Graham Cooper (1985) found that students who studied worked examples processed them much faster than students solving the same problems, then solved similar problems more quickly and with fewer errors — though the benefit was specific to problems with the same structure. Fred Paas (1992) found that studying worked or partly worked statistics problems led to better transfer, achieved with less effort, than practising conventional problems.
- Split attention. Paul Chandler and Sweller (1991) showed that when a diagram and the text explaining it must be mentally matched up, putting them together helps. In a three-month industrial training study, integrated instructions won throughout. Where the diagram could be understood on its own, integrating helped nothing.
- Redundancy. In the same series, adding seemingly useful but unnecessary explanation — such as a commentary on a self-explanatory diagram — could harm learning, even when it was well laid out.
- Expertise reversal. Slava Kalyuga and colleagues (2003) reviewed evidence that techniques which help beginners can lose their benefit, or even backfire, with more experienced learners, for whom the extra guidance is itself redundant.
Cognitive overload and how to reduce it
Cognitive overload is the everyday name for a task that exceeds working memory at that moment: you reread without taking it in, lose the step you were just told, or make errors you normally would not. It says something about the task and the moment, not about the person's ability — and it is not a diagnosis. The research above translates into a short list:
- Study a worked example before attempting a problem cold.
- Put labels on the diagram rather than in a separate key.
- Cut information that is interesting but not needed for the task.
- Break complex material into parts and learn the parts first.
- Reduce guidance as you improve — what helped at the start can get in the way later.
- Build background knowledge: what is already in long-term memory does not compete for working memory.
Knowing when you are overloaded, and choosing a strategy, is itself a skill — see metacognition. Keeping a goal in mind while resisting distraction is the related job of executive function.
How cognitive load is measured
Most studies ask learners to rate how much mental effort a task took, a method associated with Paas's early work. A 2003 review by Paas and colleagues argued that load ratings reveal things performance scores alone miss, and that combining the two gives a useful estimate of how efficient a teaching method is. Newer techniques have been tried, but self-rated effort remains the workhorse — and one of the points critics question.
Cognitive load and IQ
The limit cognitive load theory builds on — working memory capacity — is one of the abilities most closely linked to reasoning scores; the size of that link is covered on our working memory page. That is one reason reasoning tests rely on unfamiliar material: with no stored patterns to draw on, the task falls on working memory.
Cognitive load theory draws a different lesson. In a 2024 paper, Sweller argued that differences in knowledge held in long-term memory are a major source — possibly the only source — of the cognitive differences between learners that teaching can change. Working memory capacity is hard to raise; what you know is not. That is also why the evidence on raising IQ and the evidence on learning well point in different directions.
Criticisms and limits
Ton de Jong's 2010 critique credits the theory with advancing educational research and explaining a large set of experimental findings, but identifies conceptual, methodological and application-related problems — among them how the load types can be told apart in practice and how load is measured. The theory's authors summarised twenty years of further work in 2019; the practical findings above are its most used part, while the precise accounting of load types is where debate continues.
Trust and scope notes
This page explains published research on learning in general terms. It does not assess anyone's memory or attention, and it is not clinical or educational advice. Our IQ test measures reasoning; it does not report a cognitive-load or working-memory score.
IQ Revealed is not affiliated with, endorsed by or connected to the researchers, universities or journals named on this page. Their names are used only to identify the work being described.