Try this: without writing anything down, multiply 27 by 14 in your head. Notice what happens. You hold "27" and "14" in mind, break the problem into pieces, compute partial products, keep the running total from slipping away, and assemble the final answer. The place where all of that juggling happens — the mental scratchpad where information is held and manipulated at the same time — is your working memory.
Working memory is not the same as memory in the everyday sense. When most people say "memory," they mean long-term memory: the vast store of facts, skills, and experiences we accumulate over a lifetime. Working memory is something else entirely — a tiny, temporary holding space with a capacity measured in seconds and in a handful of items. And despite its modest size, it turns out to be one of the strongest single predictors of reasoning ability we have.
The difference between storage and work
An older idea, "short-term memory," treated this temporary store as a passive buffer — a place where information simply sat for a few seconds before either fading or being transferred to long-term memory. The classic measure was the digit span: how many digits can you repeat back after hearing them once? For most adults, the answer is around seven, give or take two — the famous figure from George Miller's 1956 paper, "The Magical Number Seven, Plus or Minus Two."
But in 1974, psychologists Alan Baddeley and Graham Hitch argued that passive storage was only half the story. The interesting cases — mental arithmetic, reasoning, language comprehension — require you to manipulate information while holding it, not just store it. They proposed a multi-component model and gave it a new name: working memory.
Three (or four) working parts
The model has grown over the decades, but its core components are these:
The phonological loop handles sound-based information — spoken words, digits, the inner voice you use when you silently rehearse a phone number. It has two parts: a brief store that holds acoustic information for a couple of seconds, and a rehearsal process that refreshes it (this is the silent repetition you do to keep a number "alive" before dialing). This is what a forward digit span task measures.
The visuospatial sketchpad does the same job for visual and spatial information — the mental image of a route, the layout of a room, where pieces sit on a chessboard. When you rotate a shape in your mind's eye, this is the system doing the work. The Corsi block-tapping task, where you reproduce a sequence of highlighted locations, taps this component.
The central executive is the boss. It doesn't store anything itself; it allocates attention, switches between tasks, suppresses distractions, and coordinates the two storage subsystems. This is the part engaged most heavily by a backward digit span task — repeating digits in reverse requires you to hold the sequence and mentally re-order it, which is precisely the kind of simultaneous storage-plus-manipulation the executive manages.
The episodic buffer, added by Baddeley in 2000, is a workspace that integrates information from the other systems and links it to long-term memory — letting you, for instance, combine the sound of a word with its visual form and its meaning into a single coherent chunk.
Why it predicts reasoning
Here is the part that connects working memory to intelligence. Across many studies, individual differences in working memory capacity correlate strongly with performance on fluid reasoning tasks — the kind of novel problem-solving measured by tests like Raven's Progressive Matrices. Some researchers have found the correlation so high that they argued working memory and fluid intelligence are nearly the same thing (though most now see them as closely related but distinct).
The reason is intuitive once you see it. Complex reasoning requires holding several pieces of information in mind at once while you operate on them: the premises of an argument, the partial results of a calculation, the constraints of a puzzle. If your mental workspace is larger or more efficiently managed, you can keep more elements active and track more relationships simultaneously. Run out of workspace, and the problem collapses — you lose track of where you were.
Working memory is the desk on which thinking happens. A bigger, tidier desk lets you spread out more of the problem at once. — A common metaphor in cognitive psychology
What working memory is not
It is not fixed in stone, but it is hard to expand. A popular idea in the 2000s was that "brain training" games could increase working memory capacity and thereby raise intelligence. The evidence has been disappointing: people get better at the specific training task, but the gains rarely transfer to untrained tasks or to general reasoning. Your working memory capacity is fairly stable.
It is highly state-dependent on any given day. Unlike a stable trait, your measured working memory on a specific occasion swings with sleep, stress, anxiety, caffeine, and distraction. A noisy room or a racing mind can knock points off your digit span. This is exactly why a single test result is a snapshot, not a verdict.
It is not the same as being smart. Working memory is one ingredient of cognitive performance, strongly linked to reasoning — but it says nothing about knowledge, creativity, judgment, or the thousand other things minds do.
Working memory is the small, temporary workspace where you hold and manipulate information in real time. It comes in at least three flavors — verbal, visuospatial, and executive — and it's a strong predictor of reasoning precisely because hard problems demand that you keep many things in mind at once. It's also fragile and state-dependent, which is why testing it tells you something real but never the whole story.
The next time you lose a phone number because someone spoke to you mid-dial, you've just watched your phonological loop get overwritten in real time. It's a humbling reminder of how small the workspace really is — and how much of our thinking depends on managing it well.
This article is for educational purposes. Cortextest assessments are not clinical instruments. For a formal cognitive or memory assessment, consult a licensed psychologist.