What are the working memory limits in ADHD?

ADHD adults typically hold around 4 ± 1 items in working memory under load, compared to the classic 7 ± 2 estimate for non-ADHD adults — and even Cowan's revised modern estimate of 4 items for the general population is reduced in ADHD samples. The deficit is most pronounced when the items have to be manipulated, not just held, which is why mid-thought drops feel sharper than simple recall failures.

Why does ADHD make me forget what I was about to say?

The thing you were about to say was being held in working memory. Working memory in ADHD has a smaller buffer and a faster decay rate, especially when attention is pulled by a competing stimulus. The thought wasn't lost because it was unimportant — it was lost because the buffer it was sitting in had room for one fewer item than you needed.

Is ADHD working memory deficit measurable?

Yes. Standard tasks — n-back, digit span backward, Corsi block-tapping — show consistent reductions in ADHD samples vs controls. Meta-analyses including Alderson et al. 2013 and Kasper et al. 2012 put the effect size in the medium-to-large range, with the deficit larger for spatial than verbal working memory and larger when manipulation is required.

Can you train ADHD working memory?

Working-memory training programs (Cogmed, dual n-back) show short-term gains on the trained tasks but limited transfer to real-world function — Melby-Lervåg & Hulme 2013 is the most-cited critical meta-analysis. The honest answer is that the buffer can be exercised but not meaningfully expanded. Reliable improvement comes from externalising the buffer, not training it.

What is the externalisation principle?

Externalisation means putting the cognitive load somewhere outside your head — paper, screen, sticky note, voice memo, second person — rather than trying to hold it in working memory. The principle works because external storage doesn't decay and isn't displaced by competing stimuli. The compensation isn't a smarter brain; it's a brain that doesn't need to hold the information in the first place.

Does medication help ADHD working memory?

Stimulant medication produces measurable but modest improvements on working-memory tasks, with effect sizes typically smaller than those for sustained attention. Most medicated ADHD adults still benefit from externalisation strategies — the medication closes part of the gap, not all of it.

A specific question, answered specifically

ADHD Working Memory Limits

Working memory is the short-term holding tank that lets you carry a phone number from the page to the dial pad, a thought from a meeting to the action you take after it, and the four items on the shopping list from the kitchen door to the supermarket aisle. In ADHD adults, the tank is smaller and leakier than the textbook estimate. The deficit isn't subtle and it isn't a metaphor. It's the single best explanation for why a smart person can walk into a room and forget why they walked in, three times in the same afternoon.

What the research says about the size of the buffer

Miller's classic 1956 estimate for short-term capacity — "the magical number seven, plus or minus two" — has been substantially revised since. Cowan's 2001 reanalysis put the working-memory limit closer to 4 items in the general population when chunking is controlled for ([PMID 11515286], Cowan 2001). In ADHD samples that already-modest baseline drops further. Studies using digit-span-backward, Corsi block-tapping, and n-back tasks consistently put ADHD adults around 3 ± 1 items under load when manipulation is required, with the gap widest on spatial tasks.

The most-cited meta-analyses tell a consistent story. Alderson and colleagues found medium-to-large effect sizes for working-memory deficits in ADHD adults across both phonological and visuospatial tasks ([PMID 23627574], Alderson et al. 2013). Kasper et al.'s earlier meta-analysis found the same pattern in ADHD children, and the deficit largely persists into adulthood ([PMID 22245558], Kasper et al. 2012). The size of the gap shrinks somewhat with age — older ADHD adults look more like older non-ADHD adults than younger ADHD adults look like younger non-ADHD adults — but the structural difference doesn't disappear.

Two findings matter for daily life. First, the deficit is bigger for manipulation than for simple holding. Repeating four numbers back is harder than repeating four numbers forward, and the gap between ADHD and non-ADHD performance widens as the task moves from "hold" to "hold and rearrange." Second, the deficit is amplified by competing stimulus. Working memory in ADHD samples is more disrupted by background noise, visual distractors, or simultaneous task demands than in matched controls. The buffer isn't just smaller; it's more porous.

The thought wasn't unimportant. The buffer it was sitting in had room for one fewer item than you needed.

Why items drop mid-thought

The clinical literature describes working memory as a system with three pieces: a phonological loop (verbal information), a visuospatial sketchpad (visual and spatial information), and a central executive that allocates attention between them ([PMID 12626951], Baddeley 2003). In ADHD the central executive is the part most consistently implicated. The buffer storage is somewhat smaller, but the bigger problem is that the executive can't keep all the contents active when something competing arrives.

The lived experience matches. You walk from the kitchen toward the bedroom holding three items in working memory: the laundry needs starting, the bin needs emptying, the email needs replying to. On the way, you pass the dog. The dog needs feeding. By the time you reach the bedroom, the laundry is gone, the bin is gone, and the only item left is "feed the dog," which has displaced the original three. Nothing was lost to disinterest. The buffer simply couldn't hold the original contents while a new item was being processed by the executive.

Multiply this across a workday and the cost is large. Half the small frustrations — "I just had this, what was I about to do, where did I put my phone" — are working-memory drops. The other half look like task-initiation problems but trace back to the same root: the steps of the task were in the buffer, the buffer cleared, and now the task feels like it has to be reconstructed from scratch.

The externalisation principle

The single most useful idea in this whole space comes from Russell Barkley and others working on executive function in ADHD: the working memory deficit is real, structural, and not improvable by effort. The reliable compensation is to put the contents of working memory outside the head, where they don't decay and aren't displaced. Paper. Screen. Sticky note. Voice memo. A second person who's holding the list for you. The brain stops being asked to do the thing it can't reliably do.

This sounds obvious and almost nobody actually does it. The reason is that holding things in your head feels faster, until it isn't. Writing the list down looks like a 30-second tax until you account for the four times that day you'd have looped back, lost time, and reconstructed the steps because the buffer cleared. The 30-second tax is the cheap option; reconstruction is the expensive one.

Working-memory training programs — Cogmed, dual n-back, the various "brain training" apps — produce gains on the trained task that don't transfer reliably to daily function. Melby-Lervåg and Hulme's meta-analysis is the cleanest critical summary ([PMID 23527504], Melby-Lervåg & Hulme 2013). The buffer can be exercised but not meaningfully expanded. Don't waste money there. Externalisation is what works.

Five working-memory compensations that hold up

These are the five that do real work. They share a structure: take the load out of the head, put it somewhere visible, don't trust the buffer.

1. The capture surface

One physical place where every incomplete thought goes the moment it appears. A small notebook. A plain text file always open in a window. A sticky-note stack on the desk. The rule is rigid and small: thought arrives, thought goes on the surface within ten seconds, surface gets reviewed at fixed times. The capture surface isn't a to-do list and isn't a journal. It's a holding tank for items that would otherwise be sitting in working memory taking up space and decaying.

The mechanism: the thought stops needing to be remembered, so the buffer slot it was occupying becomes available again. People underestimate how much working-memory capacity is being eaten by things they've half-decided to do later but haven't written down. Our free EF load gauge is a quick read on how heavily this is showing up in your week — it gives back four numbers and a one-page result, takes about three minutes.

2. The single-step rule for handoffs

When you walk from one room to another, hold one task in working memory. Not three. Not four. One. If three things need to happen, write them on a card and carry the card. The room-to-room handoff is the highest-failure-rate moment in an ADHD day because the change of context displaces buffer contents in a way that sitting at one desk does not. A card or a phone-note in your hand survives the change. The buffer doesn't.

The corollary: the meeting note that says "I'll send the file when I get back to my desk" needs to be written down at the moment of saying, not in the meeting room when you stand up. The walk back to the desk is exactly the kind of context-change that drops items.

3. Externalise the steps of multi-step tasks

For any task with more than three steps, write the steps. On paper, on a screen, in the order you'll do them. Cross them off as you go. This sounds excessive for "make pasta" or "send the invoice" — that's the point. The steps don't need to be remembered, retrieved, or reconstructed. They sit on the surface and you tick them off. ADHD adults often resist this on the grounds that "I shouldn't need a list for something this simple." That's not a working-memory argument; that's a self-image argument. The buffer doesn't care about your self-image. It will drop step three because step five just appeared, every time.

The cross-link to task initiation paralysis covers the related angle — the inability to start often traces back to the buffer being asked to hold "what does starting look like" while also trying to think about the task itself. Putting the steps on paper means starting becomes "look at step 1," which is much cheaper than reconstruction.

4. The voice memo for the in-between moments

The walk to the car, the shower, the drive — the moments when paper isn't available and the brain produces three useful thoughts in five minutes. Voice memos to yourself capture what working memory will lose in transit. Tap once, talk for fifteen seconds, send. Review later. The number of useful ideas an ADHD adult has in a week that never get acted on because they appeared in a no-paper context is uncomfortably large. The voice memo is the externalisation surface for those contexts.

The weak version of this is "I'll remember." The strong version is "I will not remember, and that's fine, because the audio file will."

5. Trust the second brain over the first

When the calendar says 3pm meeting and you "feel like" it might be 3:30, the calendar wins. When the list says "buy eggs and milk" and you walk into the supermarket "remembering" only eggs, the list wins. Working memory in ADHD is consistently optimistic about its own contents — meaning the felt sense that "I've got this in my head" is often wrong, and the cost of trusting the felt sense is small failures across the day. The compensation is to default to the externalised version even when your gut disagrees. Over months, the pattern of external-trust produces fewer dropped items, fewer late arrivals, and less low-grade anxiety about what you might be forgetting.

The compensation isn't a sharper buffer. It's a brain that has stopped pretending it has one.

What about medication?

Stimulant medication produces measurable improvements on working-memory tasks in lab settings, but the effect sizes are smaller than those for sustained attention or impulse control ([PMID 27262877], Cortese et al. 2018). The buffer expands a bit on a good medication day; it does not become typical-sized. Most medicated ADHD adults still benefit from the externalisation strategies above. The medication closes part of the gap, not all of it. Treat the strategies and the medication as additive rather than alternative.

What this looks like over a year

The shift from "trying to remember" to "externalising by default" doesn't happen at a moment. It happens in small substitutions over months. A capture surface that becomes second nature. A list that gets used even for two-item tasks. A meeting note that goes into the phone instead of into the head. The cumulative effect is large and quiet — fewer dropped balls, fewer "I just had this" moments, less mental tax across the day. The gain isn't the kind that feels dramatic at the time. It's the kind you notice when you compare the year-end version of yourself to the year-start version and realise the small failures are happening less often.

The reverse pattern is also worth naming. ADHD adults who continue to rely on the in-head buffer often hit a wall in their thirties or forties when life accumulates enough simultaneous demands — work, parenting, ageing parents, finances — that the buffer can't keep up with the load. The wall feels like a sudden cognitive decline; it usually isn't. It's the same buffer being asked to hold more items than it ever could.

One thing to do today

Pick a capture surface. One. A small notebook in your back pocket, a text file always open, a sticky-note pad on the desk. For the next 48 hours, every incomplete thought goes there within ten seconds of arriving. Don't curate. Don't organise. Just capture. After 48 hours, look at the list. Most of it will be either done, irrelevant, or worth one specific next action. The exercise isn't about productivity. It's about feeling the difference between holding the items in working memory and not holding them. Most ADHD adults underestimate that difference until they feel it.

If you want a sharper read on how heavily working memory is loaded in your week, the free EF load gauge is a quick run. The cross-link to the kits includes the executive-function workbook, which uses externalisation as its core mechanic.

The honest summary

Working memory in ADHD is real, structural, and smaller than it should be. Effort doesn't expand it. Training expands it on the training task only. Reliable compensation comes from putting the contents of the buffer outside the head, in a form that doesn't decay and isn't displaced. Capture surfaces, single-step handoffs, externalised steps, voice memos, default-to-list. None of it is exciting. All of it works.

If you've been operating on the assumption that you should be able to remember things and have been losing the small fights for years, this is part of why. Stop asking the buffer to do work it can't do. Build for the brain you've got.

If this lands, the ADHD Executive Function Kit is built around exactly these compensations — capture surfaces, externalised steps, re-entry. It's $4.99 right now in the launch sale (was $9.99–$49). Sale ends May 31. See all 5 kits →

Last updated: May 8, 2026