Cognitive Load
Date: 2026-08-17
The mental effort an interface demands. Working memory is small and easily exhausted, and every element competing for attention spends from the same limited budget — which is why removing things frequently outperforms improving them.
Cognitive load is the total mental effort required to use something. Working memory — the space where conscious thought happens — holds only a handful of items at once, and anything beyond that has to be re-read, re-derived or written down.
Three kinds, and only one is worth cutting
INTRINSIC the inherent difficulty of
the task
→ choosing a mattress IS hard
→ reduce by breaking it up,
not by pretending
EXTRANEOUS effort caused by the
PRESENTATION
→ unclear labels, hidden
information, inconsistent
patterns
→ ALL OF THIS IS WASTE. Cut it
GERMANE effort spent building
understanding
→ learning how your sizing
works
→ worth protecting
Extraneous load is the target. Every unit of it spent decoding your interface is a unit unavailable for the actual decision — Choice Overload.
Where it accumulates on a retail site
HOLDING INFORMATION ACROSS SCREENS
"was the one I liked £42 or £48?"
→ show the comparison, don't make
them remember
VOCABULARY THEY DON'T HAVE
"select your Tariff Band"
→ they now have to work out what
that means before they can answer
INCONSISTENT PATTERNS
filters on the left here, top there
→ the pattern must be relearned per
page
UNPREDICTABLE OUTCOMES
"will this button add it or open a
modal?"
HIDDEN INFORMATION
delivery cost at step 3 of 4
→ they've been reasoning with an
incomplete picture
The last one is the expensive one commercially. Withholding a cost doesn’t remove it from the decision; it defers the decision to a point where the person has already invested effort and now feels misled — Checkout Design.
Reducing it
- Recognition over recall. Showing options beats requiring people to remember them. This is why a visible filter beats a search box for browsing
- Chunk. Group related fields; break long forms into stages with visible progress — Progressive Disclosure
- Defaults. A sensible default removes a decision entirely — The Default Effect
- Consistency. The same thing in the same place means learning once — Design Systems
- Say things in their words, not yours — Taxonomy and Labelling, Microcopy
- Front-load the constraints. Delivery cost, stock, size availability — everything that could change the decision, before the decision
Where “reduce cognitive load” is applied badly
The advice is sound and its cargo-cult version isn’t:
- Hiding necessary information to look clean. Progressive disclosure hides secondary information; hiding a price is not a load reduction, it’s a deferral
- Removing choice people want. Someone comparing three moisturisers needs the comparison. Reducing them to one is not a simplification
- Minimalism as an end. Whitespace that pushes essential information below the fold has increased effort, not reduced it
- Fewer clicks as a proxy. Three obvious clicks beat one ambiguous one — the count isn’t the load
The measure is effort, not element count.
Signals of overload
In research — Usability Testing:
- Hesitation before clicking
- Re-reading
- Scrolling back up
- Asking “what does that mean?”
In data — Form Analytics:
- Long dwell with no action
- Repeated back-and-forth between two pages
- Form abandonment on a specific field
A pause before a correct click is the most under-read signal available. The person got there, so the funnel shows success — and the hesitation says the label didn’t do its job, which will cost you on the visitors who didn’t persist.