Cognitive Load Theory
Cognitive Load Theory
What is cognitive load theory?
- Working memory is very limited
- Too much at once breaks learning
- Manage load by chunking and sequencing
Accepting cognitive load theory changes the question instructional designers ask. It shifts from 'How do we fit all of this in?' to 'What must we cut for the rest to stick?' In practice, this shift is painful. You have to say you will use only 40 of the 120 slides the business units sent, and persuade them that for the 80 left out, 'that belongs in a manual, not in training.' The logic that cutting is what makes the rest stick is where design starts.
Sweller split cognitive load into three types: intrinsic load (the difficulty of the content itself), extraneous load (created by the format of the materials), and germane load (spent on understanding and structuring). The load design can directly influence is mainly extraneous load. If a diagram sits at the top of the screen and its explanation at the bottom, learners spend capacity moving their eyes back and forth (the split-attention effect). And when an instructor reads on-screen sentences aloud verbatim, the burden actually grows because the same information is processed through two channels (the redundancy effect). All of these prescriptions rest on the premise that working memory holds only about four chunks at a time.
A common point of confusion is that load management is not the same as 'making it easy.' What should be reduced is the clutter the format creates, not the effort of thinking itself, so if you remove even the material worth chewing on, the basis for remembering disappears with it. Nor does the same material put the same load on everyone. Step-by-step examples that help beginners become an obstacle for experts, making them walk a familiar road again (the expertise reversal effect). Until the audience’s prior knowledge level is set, there is no such thing as 'the right amount.'
The most common misconception in the field is 'chunking cuts load.' Teams slice an 8-hour course into 32 fifteen-minute pieces and call a higher completion rate a win, but learners now bear a new burden of stitching the pieces together. Microlearning without sequence and connections is not reduced load but load passed on to the learner. Swapping sentences for icons to cut text, then facing a flood of 'What does this picture mean?' questions, is the same mistake. Unless you check what you left in the space you cleared, a lighter course is just an empty one.
Jongmu KimSpecialist in corporate talent development and job skills
John KimJohn Kim, CEO of Nanum Consulting and author of B2B Marketing DesignProposed in 1988 by educational psychologist John Sweller, the theory became a scientific foundation for learning design.
Consider a hypothetical case. Company A, a medical device manufacturer with 320 employees, ran product training for new hires and internal transfers as a single 6-hour course. It crammed 12 product lines into 148 slides. The average score right after the course was 61 out of 100, and in a re-test two weeks later it fell to 44. The training team’s diagnosis was not 'we ran out of time' but 'we gave it all at once.'
- Kim, a training manager, pinned 148 slides to the meeting room wall and asked the sales team, 'Which products do new hires actually handle in their first three months?' The answer: 4 of 12 products account for 78% of transactions.
- The other 8 products were removed from the course and moved to searchable product cards on the internal portal. Team Lead Park of product planning pushed back with 'Why cut our flagship?', but it was settled with 'We are cutting it from training, not from the materials.'
- On the 30-odd slides where the instructor had read the on-screen sentences verbatim, the text was deleted and only diagrams and photos of real products remained. When Instructor Lee said, 'With nothing to read, I feel anxious,' the script was provided separately in the speaker notes.
- The 6 hours were split into three 90-minute sessions. Each opened with a 5-minute recap of the previous session’s four key points using real samples, so learners would not have to connect the fragments on their own.
- Three months after the overhaul, a comparison of two cohorts showed the two-week re-test average rose from 44 to 71, and internal inquiries from the sales floor about product specs fell from 62 to 27 per month.