Working memory

Working memory is the small, short-lived mental workspace that holds the information you’re using right now, such as the question someone just asked, the point two colleagues are debating or the thing you want to say next, while you work with it. It holds only a few items at once, and they’re easily pushed out when your attention moves to something new. In everyday language it’s “keeping something in your head”: a phone number between reading it and dialing it, or the thread of a conversation while you wait for a gap to speak.

Meetings lean on it harder than most work, for one reason: in a meeting, your working memory is the only copy of the last few minutes. A document stays on the page when you look away. A conversation doesn’t. Once a sentence has been spoken it exists only in the listeners’ heads, and if yours dropped it, the only way back is to ask. That’s why thirty seconds of distraction costs more in a call than at your desk, and why the useful fix is a second copy, not a bigger memory.

What working memory is, and what it isn’t

The term comes from cognitive psychology. The most influential model, proposed by Alan Baddeley and Graham Hitch in 1974, splits working memory into parts: a phonological loop that holds speech and sound, a visuospatial sketchpad for images and layout, and a central executive that directs attention between them. Baddeley later added an episodic buffer that binds them into one scene. You don’t need the model to use the idea, but one detail matters in meetings: the words you’re hearing and the words you’re composing, whether that’s your reply or a Slack message, lean on the same verbal side. That’s one reason typing while someone talks goes so badly.

It gets confused with a few neighboring ideas:

Term	What it is	In a meeting
Working memory	Holding a few things in mind and working with them, right now	Keeping the question in mind while you work out your answer
Short-term memory	Holding things briefly without working on them. Many researchers use it almost interchangeably with working memory.	Repeating back the last few words someone said
Long-term memory	Storage for later: large and durable, but slower to reach	Remembering tomorrow what was decided
Attention	Choosing what gets into working memory and stays there	Hearing your name, ignoring the notification
Meeting recall	Getting back what was said, from your head or from a tool	Finding the decision you missed, during the call or after it

Three limits that shape every meeting

It’s small. George Miller’s well-known 1956 paper put the number of items people can hold at “seven, plus or minus two.” Later work, notably Nelson Cowan’s in 2001, argued the limit on what you can keep in focus is nearer to about four chunks. A chunk is anything you can treat as one unit. That’s why a meeting in your own field is easier to follow than one full of someone else’s acronyms: familiar ideas arrive in bigger chunks.

It doesn’t keep. What you hold stays only while you keep refreshing it. When something else takes over, like a notification, your own train of thought or the next speaker, it can be gone within seconds. Researchers still debate whether it fades or gets overwritten. In a meeting the result is the same.

It’s shared. One small space does every job at once: following the discussion, holding the question you were asked, keeping your own point ready for a gap, and whatever is on your second monitor. Anything that takes a slot takes it from the conversation. That includes things you might not count, like decoding a bad audio line, translating in your head or worrying about being called on.

Put those together and you get the familiar meeting failure. You look away to answer one message, the thread is overwritten, and when you look back the discussion has moved on from a point you no longer have. Context switching in meetings covers what each of those switches costs. This page is about the store that gets wiped.

What fills working memory in a meeting, and where to put it instead

The practical move is to stop holding things that could be held somewhere else. Psychologists call this cognitive offloading: using something outside your head to reduce the mental work a task needs, the way you write a number down instead of repeating it to yourself. Risko and Gilbert’s 2016 review is the usual reference. It isn’t a workaround for a weak memory. Everyone’s working memory is small.

What you’re holding	Why it’s costly	Where to put it instead
Where the conversation is: the topic, what’s agreed, who’s arguing what	It’s written down nowhere, and it changes every minute	A running note of the current topic, or a live rolling summary
The question you’ve just been asked	You have to hold it while you build the answer	Write down its key words, or say it back: “So the question is whether…”
The point you want to make	Held for minutes while you wait for a gap, it crowds out listening	Jot one keyword and let it go
Decisions and action items	You’re memorizing for later while still following along	One person taking notes, or a notetaker. Don’t memorize.
A second task on your other screen	It takes slots from the discussion	Close it, or accept that you’ll need a way back in
Hard-to-hear or second-language audio	Decoding the words uses the space you’d use for meaning	Live captions for the words
Worry about being called on	Worry takes up working memory too	A plan for the moment: what to say when you weren’t listening

The first row is the one most tools skip. Notes and notetakers are good at the fourth row, the record for later. The thread of where the discussion is right now is the thing working memory holds most of during a call, and it’s also the easiest thing to lose.

Why a summary offloads better than a transcript

A transcript and a summary are both copies of the conversation outside your head. They cost very different amounts to use.

A transcript is every word. To get the thread back from it you have to read back through minutes of dialogue and rebuild the gist yourself. That uses the working memory you were trying to spare, while new speech keeps arriving on the same verbal channel. Live captions have the same problem in a smaller window, and they scroll away.

A summary hands you the gist already chunked: the topic, where it has got to, what’s being asked. That’s a few items, which is about what working memory holds anyway. A multi-resolution summary adds the choice of window. The shortest answers “what’s happening now” and the longer ones answer “how did we get here” without making you read it all. Live transcript vs live summary goes through the difference in full.

Timing matters as much as format. A recap you can only read after the call helps long-term memory: tomorrow, you’ll know what was decided. It can’t help working memory, which needs the second copy while the call is still going. That’s the core of real-time vs after-the-meeting AI notes.

Can you improve your working memory?

Not in a way that helps in your next meeting. Reviews of working-memory training programs have generally found that people improve at the exercises they practice, with little convincing evidence that the gains carry over to everyday tasks like following a discussion. What changes how much you can hold in practice is mostly load: how tired or stressed you are, how late it is in a stacked day, how familiar the topic is, and how many other things you’re holding at once.

That’s why “concentrate harder” rarely works. Gripping the thread tighter takes working memory too, and you end up less present, not more. Meeting recall: never lose the thread and staying present in back-to-back meetings cover the habits that lower the load. The other lever is the table above: hold less.

Working memory, ADHD and other differences

Everyone’s working memory is small, but it isn’t equally loaded for everyone.

• ADHD. Working-memory difficulties are among the most commonly reported cognitive differences in ADHD. Not everyone with ADHD has them, and they vary with the task and the day. Holding the thread of a meeting across moments of drift is where they show. How do I stay focused in meetings with ADHD?, the best meeting notes app for ADHD and why do I zone out in meetings? cover this in depth.
• Anxiety. Research on anxiety has long suggested that worry itself takes up working memory. That’s why nerves about being called on can make you worse at following the discussion you’re nervous about. See how do I stop worrying about being called on in meetings?
• Hearing, processing and language. When decoding the words takes effort, that effort comes from the same space as keeping the thread. The cause might be auditory processing disorder, a bad line or a meeting in your second language.
• Autism. Many autistic people describe meetings as several jobs running at once, each drawing on the same space. Making meetings easier as an autistic person covers that.

This is general information, not medical advice. If holding things in mind is hard across your whole life rather than only in some meetings, that’s worth raising with a clinician.

Where Canary fits

Canary is built to be the second copy. Canary is a real-time, bot-free meeting summarizer. It captures your computer’s system audio (no bot in the call, no plugin) and shows a live, multi-resolution rolling summary — from what’s being said right now to the whole call — so you can catch up the instant your name is called.

In practice, a small widget stays on top of your meeting app. You switch it between time windows (now, the last 2 minutes, the last 5 minutes and the full meeting) and three levels of detail, and it minimizes to a pill that shows the latest one-line summary. When you drift and come back, you glance at it instead of rebuilding the thread from memory. What did I miss in the meeting? walks through that moment.

The limits:

• It doesn’t improve your working memory. It holds the thread so yours doesn’t have to. You still need to listen to the parts that are about you.
• It isn’t a treatment. For ADHD or anything else, it’s a tool, not care.
• It’s a summary, not the words. It paraphrases. If you need the exact wording during the call, keep captions on too. After the call, Canary’s review view has the full transcript, which you can search and copy.
• It’s listening, so say so. A tool that holds the thread for you hears the whole meeting and keeps a transcript. Tell the people in the call you’re using it. How to tell participants you’re using an AI notetaker has wording that works.

For how this differs from notetakers that summarize once the call ends, see Canary vs Otter and Canary vs Granola.

Related

• Context switching in meetings: what each switch between the call and other work costs.
• Meeting recall: getting back what was said, during the call or after it.
• Rolling summary: the running copy of the conversation that sits outside your head.
• Why do I zone out in meetings?: the causes of drift, and why re-entry is the expensive part.
• What to say when you’re called on and weren’t listening: the words for the moment working memory lets you down.