Good Citizens · Usability Harness · Experiment ·

Working memory: what a page asks you to carry

We can only hold a few things in mind at once, and one interruption is enough to lose them.

You will see a code, answer a small question on the next screen, then type the code back. It grows by one character each round, the way a reference number or a code from a text message does.

Round 1 of 6 · 4 characters

Here is your code. Take as long as you like, then move on.

7YFE

I · How much we hold

Not seven. About four.

about four chunks
Cowan (2001): held at once, without rehearsing or grouping, about four things stay up. A fifth costs one of the others.

In 1956 the psychologist George Miller wrote one of the most quoted papers in psychology, “The magical number seven, plus or minus two”.

It is often read as a rule that people can remember seven things. Miller was careful to say something subtler1: that memory is limited by the number of meaningful chunks, not the number of letters or digits, and that the sevens turning up in his data were probably “a pernicious, Pythagorean coincidence”. In 2001 Nelson Cowan reviewed the evidence and put the real limit, when people cannot repeat or group what they are holding, at about four chunks2.

Chunking is why a phone number is written in groups. “4821 7730 2291” is three things to hold; “482177302291” is twelve.

II · Who has less of it

All of us, on a bad day

four places to hold thingsstressa lost nightagefewer on a bad day
Acute stress, lost sleep and age all shrink what we can hold at once. The capacity you have at your best is not the one you have when it matters.

The capacity you have at your best is not the capacity you have when it matters.

Acute stress reduces working memory, according to a review of 51 studies3, and so does losing a night’s sleep. It declines steadily with age from our twenties, even as what we know keeps growing. An estimated four in five children with ADHD show a significant working-memory difficulty.

Researchers have also argued that money worries take up mental room: in a much-discussed 2013 study, poorer shoppers did worse on reasoning tests after thinking about a large bill. Later attempts to repeat it have had mixed results, so it is better taken as a question than a settled fact. Either way, the person filling in your form may be anxious, tired or interrupted, often all three.

III · On a page

Every code you make someone copy

your code isenter the code482917482917by hand: two digits fallautofill: the whole code
Every code you make someone copy spends their working memory. Let it be pasted or filled in automatically, and none of it is dropped.

Each time a site makes someone carry something from one screen to another, it is spending their working memory.

A verification code sent by text, an order number from an email, a price from the page before, a choice made three steps ago: some people will drop them, and blame themselves. The accessibility guidelines now say so plainly. A sign-in that makes people retype a code by hand, with no way to paste it, does not comply8, and a process must not ask for the same information twice9.

Menus are the opposite case. People do not memorise a menu they can see; they recognise what they want in it. That is why Miller’s number says nothing about how long a menu may be.

IV · What to do

What it means for your design

  1. 482917482917
    01

    Never make people retype a code

    Let one-time codes be pasted and filled in automatically (autocomplete="one-time-code"), and let password managers work (WCAG 3.3.8).

  2. 02

    Don’t ask twice

    If someone has already given you something in the same process, fill it in or offer it to choose (WCAG 3.3.7 Redundant Entry).

  3. A-7731
    03

    Show it where it’s needed

    Keep the order number, the account details or the question being answered on screen at the moment people need them.

  4. 4821773022914821 7730 2291copy
    04

    Group long strings

    Show long codes and numbers in short groups, and give people a one-click way to copy them.

  5. 05

    Design for the tired user

    Plan for someone stressed, short of sleep, older or interrupted, not a rested volunteer in a lab.

  6. 06

    Save progress

    Let people leave and come back without losing what they entered, and warn them well before anything times out.

V · Prototypes

An experiment on you

measures younot the page

This one runs here, on this page: it measures you, not a design. Use what it shows you when you review a prototype or a live page with the other tools.

VI · Myths

What people get wrong

Menus should have seven items, plus or minus two
Miller wrote about remembering and judging, not menus. People recognise items in a menu they can see; they do not memorise it.
Working memory holds seven things
Closer to four chunks when people cannot rehearse or group, and it varies with the person and the moment.
Poverty lowers IQ by thirteen points
That comparison comes from one much-debated study. Several attempts to repeat it found no effect.
Multitasking permanently damages memory
The evidence is weak. Heavy multitaskers may filter distractions less well, but repeat studies largely failed.

VII · Sources

The research behind this page

Every figure above comes from one of these 10 sources. The small numbers in the text point here, and each source points back.

  1. Peer-reviewed study

    The magical number seven, plus or minus two: some limits on our capacity for processing information

    Miller ·

    Psychological Review

    psychclassics.yorku.ca · cited in I

  2. Review

    The magical number 4 in short-term memory: a reconsideration of mental storage capacity

    Cowan ·

    Behavioral and Brain Sciences

    researchgate.net · cited in I

  3. Review

    The effects of acute stress on core executive functions: a meta-analysis and comparison with cortisol

    Shields, Sazma and Yonelinas ·

    Neuroscience & Biobehavioral Reviews

    hmlpubs.faculty.ucdavis.edu · cited in II

  4. Review

    A meta-analysis of the impact of short-term sleep deprivation on cognitive variables

    Lim and Dinges ·

    Psychological Bulletin

    med.upenn.edu

  5. Review

    Moderators of working memory deficits in children with attention-deficit/hyperactivity disorder (ADHD): a meta-analytic review

    Kasper, Alderson and Hudec ·

    Clinical Psychology Review

    sciencedirect.com

  6. Peer-reviewed study

    Poverty impedes cognitive function

    Mani, Mullainathan, Shafir and Zhao ·

    Science

    collaborate.princeton.edu

  7. Peer-reviewed study

    Poverty and economic decision-making: evidence from changes in financial resources at payday

    Carvalho, Meier and Wang ·

    American Economic Review

    aeaweb.org

  8. Standard

    Understanding WCAG 3.3.8: Accessible Authentication (Minimum)

    W3C

    Web Content Accessibility Guidelines 2.2

    w3.org · cited in III

  9. Standard

    Understanding WCAG 3.3.7: Redundant Entry

    W3C

    Web Content Accessibility Guidelines 2.2

    w3.org · cited in III

  10. Guidance

    Memory recognition and recall in user interfaces

    Nielsen Norman Group

    nngroup.com