How much can your short-term memory hold?

You hear a phone number, walk across the room, and by the time you reach the phone half of it has gone. That small frustration is the subject of one of the most famous papers in psychology.

Seven, plus or minus two

In 1956 the psychologist George Miller published a paper with a catchy title. It pointed out that across many different tasks, people could keep about seven separate things in mind at once, give or take two. Seven digits, seven letters, seven words.

The headline travelled well. The more important part of the paper often gets skipped: Miller stressed that the unit isn’t the digit. It is the chunk. If you group digits into meaningful bunches, each bunch counts as one item, and you can hold more digits in total.

Then came “four”

Later researchers weren’t sure seven was right. When people were stopped from silently repeating items to themselves, or from grouping them, the number that came out was smaller. A major review in 2001 argued that the true limit is closer to four chunks at a time.

Both numbers can be true, because they describe different conditions. Seven is roughly what people manage when they can repeat and group freely. Four is nearer what’s left when they can’t. The honest answer is “a handful, depending on the material”.

Short-term and working memory

Psychologists separate two ideas that sound alike. Short-term memory is holding something for a few seconds. Working memory is holding it and doing something with it, like adding up a bill in your head.

The classic model came from Alan Baddeley and Graham Hitch in 1974. It treats working memory as a small workspace with separate parts for words and sounds, and for pictures and positions.

How people hold long numbers

A famous training study from around 1980 followed a volunteer who began with an ordinary span. He was a keen runner, and over many sessions he learned to link digits to running times and ages, turning strings of numbers into meaningful chunks. In the end he could recall a very long string.

The details are the interesting part. When he was tested with letters instead of digits, his span dropped back to ordinary. He hadn’t grown a bigger memory. He had built a better method for one kind of material.

You see the same idea in daily life. Mobile numbers in India are usually said in two groups of five, and cards and bank codes are printed in blocks of four. People group them without thinking about it.

Why digits slip away

Most losses come from a few habits:

• Distraction. Memory span is easily thrown by a noise or a notification.
• Rushing. Looking away before the number has settled.
• Holding each digit alone. Ten separate digits are ten items. Two groups of five are far easier.

Digits are harder than words

Numbers give your mind very little to hold on to. “Cat, river, lamp” comes with meaning, sound and a picture. “4 7 1 9 2” comes with nothing. Words bring associations along, which is why a short sentence is easier to remember than the same number of random digits.

It also explains how most memory tricks for numbers work. They give digits something to hang on to: a date, a price, a shape traced on a keypad.

What your result does and doesn’t mean

Your longest number is the longest string you got right on this try, in this room, on this device. Span moves with sleep, stress and attention, so a result on a bad day isn’t your ceiling. And it can’t say how good your memory is in general. It captures one narrow kind of holding: short, in order, and in digits.

Ways to hold more

Try grouping the number the way you would a phone number, and say the groups to yourself in a steady rhythm. Link a group to something you know, such as a date, a price or a house number. Take a second before you start, instead of racing in. If you want a quick way to see the difference, take the number memory test once without grouping, then again with it.

For the same idea applied to pictures, see is photographic memory real?. And because any score is a snapshot, what psychology tests can and can’t tell you is worth a read before you compare results with anyone.