Here is a theory with an unusually strong claim to your attention: it is Australian, it is nearly 40 years old, and it is still the single most useful idea in learning design.
Cognitive load theory came out of the University of New South Wales in 1988. It explains why the training everyone sat through last quarter didn’t stick, and it gives you specific, testable things to change. This guide covers what it is, the three load types and what has recently changed about them, the design effects worth knowing, and how we apply it across clinical, safety and compliance programs.
What is cognitive load theory?
Cognitiveload theory is an instructional design theory based on the limits of working memory. It holds that people can only process a few pieces of new information at once, so learning fails when instruction overloads that capacity. Good design reduces unnecessary load so attention goes to the material itself.
The practical version is blunter. Your learners aren’t slow. Your design is heavy.
Where cognitive load theory came from
John Sweller, an Australian educational psychologist, first set out cognitive load theory in a 1988 paper in Cognitive Science. He is now Emeritus Professor at the University of New South Wales and a Fellow of the Academy of Social Sciences in Australia, and cognitive load theory is one of the most cited theories in educational psychology.
The origin story is worth knowing because it explains the theory’s shape. Sweller was running problem-solving experiments and noticed something odd: people solved the puzzles easily but learned almost nothing from doing it. The explanation he built to account for that result became cognitive load theory.
That’s why the theory is sceptical of learning-by-struggling. Effort and learning aren’t the same thing, and a task can be hard without teaching anything.
Working memory and long-term memory
Cognitive load theory rests on one asymmetry in human cognition.
Working memory is tiny and brief.It is where conscious processing happens, and it holds only a few bites of new information at a time before it starts dropping things.
Long-term memory is effectively unlimited.It stores schemas, which are organised knowledge structures that let you treat many elements as a single chunk.
This is why an experienced electrician glances at a switchboard and sees one thing, while an apprentice sees forty. Same board, different schemas, radically different load.
All instruction is therefore a bottleneck problem. Everything new has to squeeze through working memory before it can reach long-term memory, and your design decides how much of that narrow capacity gets wasted.
Worth noting: stress, anxiety and what researchers call affective factors also consume working memory. Applying cognitive load theory to health and behaviour change programs, Baxter and colleagues highlight that people under financial or personal pressure arrive with less capacity available before your content does anything at all. That has direct implications for who your training is actually accessible to.
The three types of cognitive load
Almost every article on this topic lists three load types as though they add up neatly. The picture has changed, so read the third one carefully.
| Load type | What it is | What to do about it |
| Intrinsic | The inherent complexity of the material, relative to the learner’s existing knowledge | Optimise it. Sequence and chunk so complexity arrives in a manageable order. You cannot remove it without removing the learning. |
| Extraneous | Load created by how the material is presented rather than by the material itself | Cut it ruthlessly. This is the load your design is responsible for, and it is pure waste. |
| Germane | The working memory actually devoted to making sense of the material and building schemas | Nothing directly. See below. |
What changed about germane load
For years, germane load was treated as a third, separate load that designers should try to increase. Current thinking has moved on.
Sweller and colleagues found that when extraneous load was reduced, total cognitive load went down rather than germane load rising to fill the space. As the review by Baxter and colleagues puts it, germane load is no longer considered an independent source of load. It is better understood as the working memory resources available for the intrinsic load of the material once extraneous load has been stripped out.
This isn’t just semantics. You don’t “add germane load” by bolting a reflection activity onto the end and calling it a day. You create room for genuine sense-making by removing the junk. If you have inherited a design template that instructs you to increase germane load, it is working from a superseded version of the theory.
Cognitive load effects worth designing around
Cognitive load theory is unusual among learning theories in that it generated a long list of specific, experimentally tested effects. These are the ones we use most.
| Effect | What it means | What it looks like in practice |
| Worked example effect | Novices learn more from studying a completed example than from solving the problem themselves | Show the finished risk assessment before asking anyone to write one |
| Split-attention effect | Load rises when learners must mentally integrate two separated sources | Put labels on the diagram, not in a legend below it |
| Redundancy effect | Duplicating the same information in two forms hurts rather than helps | Do not narrate on-screen text word for word |
| Modality effect | Narration plus visuals beats on-screen text plus visuals, because it uses two channels | Voice over a diagram, not a paragraph beside it |
| Segmenting | Breaking content into learner-controlled chunks reduces overload | Short modules with a pause between, not a 45-minute run |
| Expertise reversal effect | Support that helps novices actively harms experts | Give experienced staff a different, shorter path |
| Guidance fading | Withdraw scaffolding as competence grows | Full worked example, then partial, then independent practice |
| Transient information effect | Video and audio disappear as they play, so complex content overloads | Give controls, chapters and a written companion for anything intricate |
Several of these map directly onto visual design in learning, and guidance fading is the mechanism behind scaffolding.
Cognitive load theory in education and in the workplace
Cognitive load theory took hold in schools first, and Australia is where much of the translation work happened. The NSW Department of Education’s Centre for Education Statistics and Evaluation published a literature review on cognitive load theory and a companion practice guide with classroom examples. Sweller himself points to those two guides as the useful entry point.
Workplace learning has been slower to catch up, which is odd, because the workplace case is arguably stronger.
| Classroom | Workplace | |
| Learner attention | Timetabled, protected | Squeezed between meetings and a live inbox |
| Prior knowledge | Reasonably uniform cohort | Wildly variable, from graduate to 30-year veteran |
| Environment | Designed for learning | Noisy plant, moving vehicle, ward mid-shift |
| Consequence of overload | A poor test result | A missed hazard or a clinical error |
| Who writes the content | A trained teacher | A subject matter expert with no design training |
Every one of those differences raises extraneous load. That last row is the biggest problem in corporate L&D, and it has a name.
Expert blind spot: the biggest source of load in workplace training
Researchers call it expert blindness or the expert blind spot: once you know something well, you lose the ability to see what a novice finds confusing. Your knowledge is chunked into schemas, so the complexity has become invisible to you.
This is why subject matter experts produce training that overwhelms people. It is not carelessness. They genuinely cannot perceive the load, because for them it is not there.
Three things we do about it:
- Never let content go straight from SME to build.A designer sits in between whose job includes asking the questions the expert has stopped being able to see.
- Test with actual novices, not reviewers.Executives and SMEs catch typos. A person from the target audience catches the overload.
- Co-design with the audience where you can.On the Baby Communication project with The Kids Research Institute Australia, we co-designed withearly childhood educators rather than designing for them. That is a direct structural defence against expert blind spot.
How cognitive load theory shows up in our work
Some real examples where load management drove the design decision.
Giving experts a different path: St John of God Health Care
The expertise reversal effect says instructional support that helps a novice can actively slow an expert. That is not a theoretical concern in clinical education, where the same Advanced Life Support content has to work for a new graduate and a 20-year intensivist.
St John of God Health Care told us their existing training worked but carried too much repetitive theory for experienced clinicians. So the design deliberately splits the path: structured foundational learning for those who need it, streamlined and relevant refreshers for those who do not. The online component moves through four stages, Recognise, Respond, Review and Reflect, before a team-based simulation workshop using rapid cycle deliberate practice.
The modular structure also works around shift patterns, which matters because a nurse studying between shifts has less working memory available than the same nurse on a training day.
St John of God Health CareA project replacing externally delivered training with an internal blended program, now running across multiple hospitals from small regional sites to large tertiary centres. The design brief was explicitly about confidence under pressure, not just technical competency. Read the St John of God Health Care case study.
Cutting redundancy: EnergyAustralia
Two separate safety modules covering overlapping ground is a redundancy problem wearing a compliance costume. Learners pay the load cost twice and remember less than if they had done one.
We merged them into a single cohesive 30-minute eLearning covering both physical and psychosocial hazards. Less content, better retention, and one thing to maintain instead of two. Read the EnergyAustralia case study.
Segmenting across time: Sentis
Sentis needed safety leadership development for a dispersed workforce. We built a 12-week learning journey on a weekly rhythm rather than a compressed multi-day event.
That is segmenting applied at program scale. Each week carries one theme, with self-directed material, a live session and a community hub between touchpoints. Nobody has to hold twelve topics in working memory at once.
Designing for depleted capacity: sensitive content
When the subject matter is trauma, family violence or consent, learners arrive with working memory already partly consumed by the emotional weight of the topic. Loading a dense module on top of that is a design failure, not a learner failure.
Across Mitigating Vicarious Trauma, Consent Labs, The Shark Cage and Challenge DV, the shared design response is learner-controlled pacing, short segments, and framing before content.
Mitigating Vicarious TraumaFrontline community services workers carry the weight of other people’s trauma. We designed short, self-paced learning that let people control their own exposure while still building practical strategies. Read the Mitigating Vicarious Trauma case study.
Offloading to the environment: Acknowledge This! and VACCHO
The cheapest way to reduce load is to stop requiring memory at all.
For Acknowledge This! the deliverables included a workbook, a quick reference guide and an Acknowledgement of Country template alongside the eLearn. Nobody has to memorise what they can look up at the moment they need it.
For VACCHO and the Balit Durn Durn Centre we built a complete eLearning design system with branded templates and a practical guide their team owns. Consistency is a load reduction strategy: when every module behaves the same way, learners stop spending capacity working out the interface.
How to reduce cognitive load in your own programs
A practical checklist, in the order we work through it.
- Cut content before you design.The most effective load reduction is deleting what the learner does not need to do their job.
- Sequence from simple to complex.Prerequisite knowledge first, always.
- Segment and give control.Short chunks, visible progress, a pause between.
- Remove decorative visuals.A stock photo of a handshake is pure extraneous load.
- Integrate text with graphics.Labels on the image, not in a separate key.
- Do not read the screen aloud.Narrate over visuals instead.
- Lead with a worked example.Show it done before asking for it done.
- Fade the scaffolding.Full support, partial, then independent.
- Build a different path for experts.Or at least let them skip.
- Push reference material out of the head. Job aids and quick reference guides beat memorisation for anything lookup-able.
- Write plainly.Unfamiliar language raises intrinsic load before the concept does.
- Check accessibility.The Web Content Accessibility Guidelines overlap substantially with load reduction, particularly on clear structure, predictable navigation and readable content.
- Test on the real device, in the real environment.A module that works on a monitor may not work on a phone in a noisy depot.
Point one is where most of the win is. In compliance and safety training especially, “we have to cover it” is usually a claim about the document, not about the duty. Safe Work Australia’s guidance on training and supporting your workers is explicit that information and training must be understandable to the person receiving it. Overloaded training is not a safer legal position than concise training. It is a worse one.
Where cognitive load theory gets misused
We use this theory constantly. We do not treat it as beyond question.
- Measurement is genuinely contested.Separating intrinsic from extraneous load empirically is hard, and researchers including de Jong and Schnotz have criticised the theory’s conceptual clarity and measurement rigour. Sweller himself has written about the difficulty of measuring cognitive load independently of task performance. Treat load as a design lens, not a number you can report.
- “Reduce cognitive load” is not the same as “make it easy.”Learning requires effort. The goal is to remove wasted effort, not all effort. A design so frictionless that nobody has to think produces nothing.
- It is often used to justify cutting practice.Practice raises load in the short term and builds schemas in the long term. Cutting the hard part because it feels heavy is a misreading.
- One-size prescriptions ignore expertise.Nearly every cognitive load recommendation reverses at high expertise. Always ask who the learner is before applying a rule.
- It does not tell you what to teach.Cognitive load theory optimises how instruction is delivered. Deciding what needs to change starts with a learning needs analysis, and pitching the level is where Bloom’s taxonomy does the work.
There is also a live tension worth naming. Cognitive load theory’s founders have argued forcefully against minimal-guidance approaches such as pure discovery learning for novices. That does not invalidate problem-based learning, which we use often and rate highly. It sharpens when to use it: problem-based approaches work when learners have enough prior knowledge to handle the load, and with real guidance built in. Handing a novice an unstructured problem and hoping is the thing the research argues against.
FAQs
Frequently asked questions about cognitive load theory
Design for the brain you are actually talking to
Cognitive load theory is not a nice-to-have piece of theory. It’s the reason a 90-slide deck flops and a tight 20-minute module lands. And it hands you specific things to fix, not a vague pep talk to “be clearer.”
If your training is not landing and you are not sure why, that diagnosis is where learning strategy starts. If you know what needs building and want it designed around how people actually process information, our learning design and eLearning teams do this for a living. Send us the content nobody can get through and we will tell you what is carrying the load.
References
- Centre for Education Statistics and Evaluation, NSW Department of Education, Cognitive load theory: research that teachers really need to understand
- Centre for Education Statistics and Evaluation, NSW Department of Education, Cognitive load theory in practice: examples for the classroom
- Baxter, K. A., Sachdeva, N. and Baker, S. (2025), The application of cognitive load theory to the design of health and behavior change programs, Health Education & Behavior, National Library of Medicine
- Sweller, J. (2018), Measuring cognitive load, Perspectives on Medical Education, National Library of Medicine
- Safe Work Australia, Training and supporting your workers to stay safe at work
- W3C Web Accessibility Initiative, WCAG 2 Overview
