Great at digits, terrible at 2-back — why?
Someone holds 11 digits in the number test and then scores 40 on 2-back. Someone else stalls at 7 digits but scores 80 on 2-back. Both are called “memory,” so why do they disagree so often?
The answer is simple: the two tests ask for different work.
A warehouse and a conveyor belt
Number memory asks one thing: how many items can you hold at once? The digits arrive once, and while you hold them the contents of your head do not change. They only grow. This measures warehouse size.
Working memory (2-back) asks something else: can you keep swapping the contents and still judge correctly? Every new letter means discarding the oldest and inserting the newest. By the structure of the task you only ever hold two at once, so the load falls on swapping accurately, every single time.
That difference is why the two scores drift apart. A big warehouse does not imply an accurate conveyor.
So different things make each one hard
What stalls number memory is usually count — and here chunking works powerfully. Handling 010-1234-5678 as three groups rather than eleven items cuts the load dramatically. The limit on how many chunks a person can juggle is an old topic (Miller, 1956), with later work arguing that limit is smaller still (Cowan, 2001).
What stalls 2-back is hard to put down to count — there are only ever two things to hold. These are the hard parts instead:
- Discarding. Still clutching the letter from three steps back produces false taps. Letting go is harder than holding on.
- Confusion. When a recent letter resembles the one from two steps back, you keep the feeling of having seen it while the position blurs.
- Pace. It updates every 1.5 seconds, so one slip leaves your queue misaligned for everything after.
This is why the chunking that works so well for digits does not transfer to 2-back: there is very little to group in the first place. (That reasoning comes from the structure of the two tasks; nobody here measured what chunking practice does to a 2-back score.)
Reading the two scores together
If your two results disagree, here are readings worth considering. Two scores cannot isolate a cause, so treat these as hints about what to look at next, not conclusions.
| Number memory | 2-back | Worth considering |
|---|---|---|
| High | High | A run where both tasks’ demands happened to suit you |
| High | Low | Updating may be where it caught. Did old letters feel like they lingered? |
| Low | High | A combination showing that how much you hold does not govern 2-back |
| Low | Low | Suspect the conditions first — noise, fatigue, first attempt |
The third row shows plainly that a low digit span does not mean “bad memory.”
Both scores swing with conditions
Both tests are affected by whether the room is quiet. People who hold items by rehearsing them internally tend to find nearby conversation disruptive (how much it costs varies by person). The first few runs also carry the cost of learning the rules, so do not treat your first game as your baseline.
One caution
Neither value is a grade for intelligence, academic ability, or memory in general. Each is a score from a short game built a particular way. Similar structures have long been used in academic tasks (Kirchner, 1958), but the stimuli, trial counts and conditions in that research are not ours. Use these only to compare yourself against yourself, on the same device.
In short
- Number memory measures warehouse size; 2-back measures accurate swapping.
- So the two scores disagreeing is normal.
- Grouping suits number memory. 2-back has little to group, so the same trick does not fit.
- 2-back only ever asks you to hold two, so the load sits on the swapping.