Cognitive accessibility in a digital product

Perceivio·22 January 2026

What cognitive accessibility covers

Most accessibility work is about how a product is perceived and operated. Cognitive accessibility is about how it is understood and remembered. It covers attention, memory, language, learning and the executive function that decides what to do next.

The two overlap. A form that takes four attempts to submit is a keyboard problem for one reader and a memory problem for another, and the fix is often the same fix.

The ADA (Americans with Disabilities Act) does not name cognitive conditions, WCAG (Web Content Accessibility Guidelines) has no section for them, and neither does any other standard we test against. That does not make WCAG silent. It means the criteria that carry cognition are scattered through the four principles, which is exactly why they get skipped.

The criteria that carry it

• 3.3.7 Redundant Entry (Level A). Do not ask for information the reader has already given in the same process. An address entered at step one should be filled in at step four, or confirmed, not typed again.
• 3.3.8 Accessible Authentication (Level AA). No cognitive test to sign in. Pasting into password fields must work, and a password manager must be able to fill the form.
• 3.2.6 Consistent Help (Level A). Where a help route exists, it stays in the same relative place on every page, so a stuck reader does not have to search for it again.
• 2.5.7 Dragging Movements (Level AA). Anything dragged has a single pointer alternative, because dragging needs a steady hand and a good aim.
• 2.2.1 Timing Adjustable (Level A) and 2.2.6 Timeouts (Level AAA). A time limit a reader can extend, and a warning before an idle session throws work away.
• 2.4.6 Headings and Labels (Level AA), 3.3.1 Error Identification (Level A) and 3.3.4 Error Prevention (Level AA). Names that say what a thing is, errors that say what happened, and a chance to check or reverse anything that costs money or data.

That list is not a separate standard. It is the part of WCAG 2.2 that a testing plan can quietly skip if nobody is looking for it.

Where it shows up in a real product

Four examples, all taken from flows rather than pages, because cognition is a matter of sequence.

• A checkout that times out. The reader gathers a card and an address, comes back, and the basket is empty. Nothing was broken. Everything was lost.
• A slider that only drags. A price filter a keyboard cannot set and a hand cannot hold steady, on the screen that decides whether a purchase happens.
• A form that repeats itself. The same address asked for three times because each step was designed on its own.
• An error that says something went wrong. No field named, no reason given, and the reader left to guess which of eleven boxes to change.

How we test for it

Tooling cannot find most of this. A markup checker sees that a label exists, not that it says the right thing, and it cannot see that step four asks for step one again.

So we test the way a person uses the product. We run a real task end to end, then run it again while reading every word out loud, then run it with a screen reader and with the browser at a large text size. We count the decisions a reader has to hold in mind between starting and finishing, and we look at what happens when a step fails. Details of the method are on the how we test page.

What to read next

The working memory side of this is in what cognitive psychology says about interface memory load. The habits that keep teams from acting on it are in cognitive distortions that hide accessibility problems. Where the product is a mental health service, the accessibility work has its own shape, which we wrote about in accessible digital CBT (cognitive behavioural therapy).

If you would rather see it on your own product, the accessibility audit and testing service covers cognitive criteria with everything else, and the free one-page audit shows you the shape of the findings on one page.