Almost everyone has heard that working memory holds seven plus or minus two items. That came from a 1956 paper by George Miller, and Miller himself framed it more cautiously than the citation suggests — his title referred to the magical number seven partly in jest, and his examples mixed several different kinds of task.
The figure has since been revised sharply downward. Later work, most influentially by Nelson Cowan in 2001, concluded that when you strip out rehearsal, grouping and long-term memory support, genuine working-memory capacity is closer to four chunks. Some estimates put it lower still for complex items.
Four. Not seven. That is the number to hold your open work against, and it explains why a to-do list of eleven things produces paralysis rather than productivity.
A chunk is whatever your long-term memory can treat as one unit. The string of letters F B I C I A N A S A is ten items and beyond capacity; read as FBI, CIA, NASA it is three chunks and trivial. Nothing changed except what you recognise.
This is what expertise actually is. A chess master does not have larger working memory than a novice; they see one familiar structure where the novice sees sixteen separate pieces. It is also why the same task is exhausting for a beginner and effortless for a practitioner, and why the fix for overload is often practice rather than organisation.
Cognitive load theory, developed in instructional research by John Sweller, splits load into three:
| Type | What it is | What to do |
|---|---|---|
| Intrinsic | Inherent difficulty of the material | Cannot reduce, only sequence — break into stages |
| Extraneous | Load from how it is presented | Eliminate ruthlessly — this is the wasted portion |
| Germane | Effort spent building durable understanding | Protect — this is the productive part |
Extraneous load is where the practical gains are. A badly organised codebase, a form with unclear labels, notes scattered across four apps, or an explanation that makes you hold two things in mind to understand a third — all of that consumes the same four slots as the actual problem, and none of it produces anything.
Count what is genuinely open in your head right now, not on your screen:
| Source of load | Count | Chunks consumed |
|---|---|---|
| Unfinished tasks you are tracking mentally | 3 | 3 |
| Conversations awaiting your reply | 5 | ~2 if grouped as one queue |
| Open decisions with no deadline | 2 | 2 |
| The task you are actually doing | 1 | 1–2 |
| Total | 8–9 against a capacity near 4 | |
The overflow does not vanish. It manifests as rereading the same paragraph, losing the thread mid-sentence, and the low-grade background tension of holding things that will not fit. The single most effective intervention is external storage: writing every open item down reliably frees the slots that were being spent remembering them rather than working on them.
Cognitive load is not directly measurable outside a laboratory. Researchers approximate it with pupil dilation, secondary-task performance, heart-rate variability and validated questionnaires — the NASA Task Load Index being the best known, scoring mental demand, physical demand, temporal demand, perceived performance, effort and frustration. All of those require instruments or structured self-report.
So a web page counting your open tabs is producing a heuristic prompt, not a measurement. It cannot know whether your five open threads are one familiar chunk or five unrelated problems, it cannot distinguish intrinsic from extraneous load, and it cannot see fatigue, stress or sleep debt — all of which reduce effective capacity substantially and none of which appear in a count.
Persistent difficulty concentrating that does not respond to reducing open items is worth discussing with a doctor rather than a calculator, since it has many possible causes. This page is general information, not medical advice.
No. What feels like parallel work is rapid switching, and each switch leaves attention residue on the previous task. Measured throughput on two interleaved tasks is reliably worse than doing them in sequence.
Because it converts working memory items into external storage. The relief is not psychological comfort — you have literally freed capacity that was being spent on retention.
Unrehearsed information decays within roughly 15 to 30 seconds, which is why an interruption during a phone number is so destructive.
No. Your counts stay in your browser and nothing is transmitted or saved.
Every tool comes with a written guide, and every category is one click away.