Psychology World — home
Blog

Cognitive load theory: why working memory gives up, and how to study within its limits

Cognitive load theory says working memory holds only a few chunks at once. The three loads, the limits, and how to study psychology terms within them.

Studying13 min read
By Psychology World

Cognitive load theory says your working memory can hold only about four chunks of new information at once. When the total demand passes that limit, learning fails. John Sweller proposed the theory in 1988 after studies of problem solving, and it still shapes how teachers design instruction. The reason it matters for you is direct: every psychology term you learn must pass through that narrow channel before it reaches long-term memory.

  • Cognitive load is the demand on working memory at one moment. Overload is the state when that demand passes capacity.
  • New information moves from sensory memory to working memory to long-term memory, where it is stored in schemas.
  • Working memory holds about four chunks. Miller reported seven plus or minus two in 1956. Cowan updated it to about four in 2010.
  • Intrinsic load comes from the material itself. Extraneous load comes from how it is presented. Germane load is the work of building schemas.
  • In the 2019 revision, germane load is not a third load added on top. It is the good use of capacity freed by cutting extraneous load.
  • You can study within limits by chunking material, sequencing prerequisites first, cutting distractions, and testing yourself in short sessions.

What the theory claims

The theory starts from a simple model of memory. Sensory memory holds sights and sounds for a very short time and filters what passes on. Working memory holds and processes the selected information. Long-term memory stores what survives in schemas, which are organised structures that guide recall and use. The more often you use a schema, the easier it becomes to recall.

Working memory is the narrow part. Miller reported in 1956 that short-term memory holds about seven plus or minus two items. Later work revised the number down. Cowan reported in 2010 that the limit is about four chunks when processing is required. A chunk is not a fixed unit. A single letter is a chunk for a beginner. A whole definition is one chunk once you know it well. This is why prior knowledge changes everything in the theory.

Sweller studied how the way problems are presented changes learning. He found that instruction itself can waste the limited capacity. His proposal was practical: design instruction so that it respects working memory. Teach at the right level, cut what does not help, and leave room for the learner to build schemas. The Medical College of Wisconsin teaching guide from 2022 puts the same point in one line. Schemas, even complex ones, count as one chunk. So prior knowledge sets where instruction should start.

Baddeley and Hitch had prepared the ground in 1974 with their working-memory model. They separated short-term memory, which only holds information, from working memory, which holds and processes it. That distinction is the base of the theory. You do not only store a term. You process its parts, link them, and file them. Each step costs capacity.

The three loads, compared

Ask one question about any difficult study moment: does the difficulty come from the material, from the presentation, or from the schema-building? The answer tells you which load dominates. Intrinsic load is the material and what you already know. Extraneous load is the presentation and the environment. Germane load is the mental work that turns the material into lasting schemas.

The three loads, what each one is, and what each looks like with psychology terms.
LoadWhat sets itPsychology-term example
IntrinsicThe difficulty of the material, the number of linked parts, and your prior knowledgeRetrograde versus anterograde amnesia feels hard because direction, cause, and example must be held together
ExtraneousThe layout, the wording, the noise, and every extra step the design forces on youA term on one page, its definition in another app, and alerts on top split attention and waste capacity
GermaneThe work of linking new material to what you know and storing it as a schemaComparing two terms in your own words and testing yourself builds the schema that lasts

Intrinsic load rises with element interactivity. Interactivity is low when you can learn one part alone, such as the spelling of a single term. It is high when many parts must be processed together, such as the four schedules of reinforcement with their ratio and interval logic. You cannot remove intrinsic load without changing what is learned. You can manage it by sequencing. Teach prerequisites first, break a hard topic into parts, and let each part become a chunk before you combine them.

Extraneous load comes from design. Text placed far from its diagram forces the eye to jump. The same words spoken and shown at once force double processing for no gain. Decorative images, unclear wording, and extra clicks add demand without adding meaning. The 2022 teaching guide lists the fixes: put labels inside diagrams, keep wording clear, link directly to needed resources, and limit background noise. Each fix frees capacity for the material itself.

Germane load is the useful part. It is the effort of connecting a new term to examples, contrasts, and prior schemas. Older pages describe total load as intrinsic plus extraneous plus germane. Sweller and colleagues revised this in 2019: when extraneous load falls, the freed capacity becomes available for germane processing. Do not try to add germane load on top of a full system. Cut extraneous first, then use the room to build schemas. For background on why cramming fades even when it feels productive, see the forgetting curve.

Why working memory gives up

Overload has clear signs. You reread the same paragraph without progress. You confuse pairs you knew yesterday, such as CS and US or recall and recognition. You make simple mistakes on material within your level. The 2022 teaching guide notes a clear pattern. Even strong learners fail when the problem space is too large. Problem space means the gap between where they are and where they need to be. The fix is to shrink that gap: break the task down and confirm each step.

Element interactivity explains why some pairs stay hard. Retrograde versus anterograde amnesia is not two labels. It is direction of loss, time of injury, and a clinical example, all linked. Until those links form one schema, they occupy several slots. The same holds for classical versus operant procedures, or for positive versus negative in reinforcement and punishment. High interactivity is normal at first. It falls as schemas form, which is why a senior student sees one pattern where a beginner sees five facts.

The environment adds its own tax. A study of college homework sessions found students averaged 73 minutes of music and about 35 distractions of six seconds or more across three hours. More self-efficacy meant less multitasking in that study. Each switch costs capacity. Stress and low sleep also reduce available capacity, which reviews of the theory note as context that designers should respect. You cannot expand the channel in an evening. You can only decide what fills it.

Three instructional effects from the literature show how presentation changes load. The split-attention effect says separated sources force mental integration that wastes capacity. The redundancy effect says duplicated content can hurt when one source already suffices. The modality effect says well-chosen visual plus auditory input can help because the channels partly differ. The practical rule is not more media. It is the right media, placed together, with nothing extra. A diagram with built-in labels beats a diagram plus a separate legend. One clear channel beats three noisy ones.

How to study within its limits

Use the theory as a study checklist. Each step below manages one load. Do them in order, one list at a time.

  1. 1.Chunk terms into one list per topic. Create a bookmark list for one exam topic, such as conditioning or memory, and add only its terms. Ten related terms in one list form fewer effective units than ten scattered terms. The app generates a random list of about ten words when you need a ready-made chunk.
  2. 2.Sequence prerequisites before details. Learn the base schema first, then add the contrast. Learn what reinforcement means before you sort positive versus negative. Learn what a schema is before you study germane load. Each prior chunk lowers the intrinsic load of the next step.
  3. 3.Study one list in one short session. Open a single list and drill it with one game. Close other sources while you do. This cuts split attention and keeps extraneous load low. Short sessions also fit the segmenting principle: divide content so working memory never faces the whole course at once.
  4. 4.Keep layout clean and direct. When you make notes, place the example next to the term, not pages away. Prefer clear wording over decoration. If you use diagrams, label them inside. If a resource needs a link, link to the exact page. These are the same fixes teaching guides recommend.
  5. 5.Combine visual and verbal with care, and avoid duplication. Read the term, say it aloud once, and write one example from memory. Do not read the same definition while listening to the same words and watching the same text scroll. Use two channels to complement, not to repeat. For a full method built on this idea, see how to memorise psychology terminology.
  6. 6.Test and space, and mix with care. Recall a term, check, and repeat after a gap. Testing builds schemas, and spacing keeps sessions within capacity. A 2018 study discussed in a 2025 review links the spacing effect to working-memory depletion, which fits the theory. For scheduling, use spaced repetition. For mixing topics once each is stable, use interleaving. For vivid single examples, use the method of loci sparingly, not for every list.

Pairs that differ by one word need active contrasts, not rereading. Save CS, US, CR, and UR to one list and drill them until each has its own cue. Save recall versus recognition with one example each. The guide to elaborative rehearsal explains why linking meaning beats repeating form. That link is the germane work the theory protects.

What each study mode does is set out on the home page. Flashcards suit first passes through a list. Multiple choice needs at least four words in the list and suits sorting pairs. Typing suits final checks because it forces full recall. Pick the mode that matches the schema you try to build, then stop the session before fatigue adds errors.

What most pages get wrong

Most overviews get the basics right and four details wrong. Correct them and the theory becomes much more useful.

First, seven plus or minus two is history, not capacity. Miller reported it in 1956 for short-term storage. The current guide for processing is about four chunks. State it with the date: seven in 1956, about four since Cowan in 2010. Do not plan study around seven new terms at once.

Second, germane load is not extra load to add. Older diagrams show three bars stacked. The 2019 revision says germane is the good use of freed capacity. When you read an older page with three additive bars, translate it. Cut extraneous, keep intrinsic at the right level, and spend what is left on schema-building.

Third, more media is not always better. A video with captions plus the same slides plus a transcript can trigger the redundancy effect. Music with lyrics during verbal learning adds extraneous load. The modality effect helps only when the second channel carries distinct, needed information. When in doubt, remove one source and test again.

Fourth, multitasking does not expand capacity. It divides it. Notifications, open tabs, and background shows each claim a share. The homework study above is typical: dozens of micro-switches across an evening. One task, one list, one game. That is the whole trick, and it is enough.

Critics add two honest cautions. De Jong argued in 2010 that load cannot be measured directly and that self-report scales dominate, which makes claims about exact loads hard to test. Kalyuga argued in 2011 that intrinsic and germane loads are hard to separate and that germane was added on theory more than on direct results. Read the theory as a design guide with strong support, not as a meter that reports three exact numbers. The central fact stands: working memory is narrow, design changes learning, and schemas carry expertise.

Keeping the terms apart

Students mix five pairs. Keep them apart from the start.

  • Load is demand. Overload is the state when demand passes capacity. You manage load so you avoid overload.
  • Intrinsic is difficulty in the material. Extraneous is difficulty in the presentation. A hard topic can have clean presentation. An easy topic can have messy presentation.
  • Germane is schema-building effort. General effort is not germane. Staring longer at a cluttered page is effort without schema gain.
  • Working memory holds and processes. Short-term memory only holds. The theory concerns the first.
  • A schema is an organised structure in long-term memory. A chunk is how much of it you handle at once in working memory. A large schema still counts as one chunk once formed.

These distinctions turn exam questions into sorting tasks. If the question asks what depends on prior knowledge, answer intrinsic. If it asks what a better layout would fix, answer extraneous. If it asks what testing builds, answer schemas through germane processing.

Frequently asked questions

What is cognitive load theory in simple terms?

It says working memory holds only a few chunks at once, so instruction must fit that limit. Intrinsic load is the material, extraneous load is the presentation, and germane processing builds lasting schemas. When total demand passes capacity, learning stalls.

What are the three types of cognitive load?

Intrinsic load is the difficulty of the content plus what you already know. Extraneous load is the extra demand from poor layout, noise, or split sources. Germane processing is the useful work of linking new facts to schemas. The 2019 revision treats germane as the good use of freed capacity.

Can you give an example of cognitive load?

Learning retrograde versus anterograde amnesia carries high intrinsic load because direction, timing, and examples must be held together. Reading the term in one app while the definition sits in another adds extraneous load. Writing one contrast in your own words and testing it builds germane processing.

What did Sweller actually propose in 1988?

Sweller studied problem solving and argued that instruction should be designed around working memory limits. He proposed guidelines that reduce wasted capacity so learners can build schemas. Later work added the three-load language in 2010 and revised germane load in 2019.

How do I reduce cognitive load while studying?

Chunk terms into one list per topic and learn prerequisites first. Study one list in one short session without other tabs. Keep notes clean with labels inside diagrams. Test yourself and space repeats. Each step cuts extraneous load and protects room for schema-building.

Is cognitive load theory still accepted?

The core claim is widely used in education: working memory is limited and design changes learning. Details are debated, including how to measure loads and whether intrinsic and germane can be separated. Treat it as a practical guide with dates attached, not as exact numbers.

Sources

  1. 1.Baxter, K. A., Sachdeva, N., & Baker, S. (2025). The Application of Cognitive Load Theory to the Design of Health and Behavior Change Programs. Health Education and Behavior.
  2. 2.Medical College of Wisconsin. (2022). Cognitive Load Theory: A Guide to Applying Cognitive Load Theory to Your Teaching.
  3. 3.Koenig-Workman, I. (2024). Cognitive Load Theory. The Decision Lab.
  4. 4.Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science.
  5. 5.Miller, G. A. (1956). The magical number seven, plus or minus two. Psychological Review.
  6. 6.Cowan, N. (2010). The Magical Mystery Four. Current Directions in Psychological Science.

Look the terms up yourself

18,000+ psychology and psychotherapy terms in English and German, on iPhone, iPad, Mac and Apple Vision.

See how Psychology World works
All posts