The Workbench and the Warehouse: Two Memories

Ask someone to dictate eight random words. By the sixth you'll start losing the first. Eight words — and that's the ceiling.

Now recall the route from home to school that you walked as a child. The turns, the cracks in the asphalt, the smell by the entrance, the dog behind the fence. Twenty years have passed — it's all there, in the finest detail.

Eight words don't fit, but twenty years do. Because these are two completely different systems, living by different laws. And half the mistakes in studying come from trying to handle them the same way.

A cramped workbench, a boundless warehouse

Picture a workshop. In it there's a workbench — a small table where the work happens right now. Three or four objects fit on it, no more. Put a fifth down and something falls to the floor. This is working memory: everything you're operating with at the current moment.

And beyond the door — a warehouse stretching past the horizon. Shelves to the ceiling, millions of boxes, room enough for everything you'll manage to live through in a lifetime. This is long-term memory.

Between them works the night shift — a crew that, after the workshop closes, takes everything off the table, carries it to the warehouse, and puts it in its place, next to the similar.

All of studying is logistics between three elements. The workbench is tiny, the warehouse boundless, and the main question is what exactly makes it to the warehouse and whether you'll then be able to find the box you need.

The workbench: why so cramped

The working-memory model that Alan Baddeley and Graham Hitch proposed in 1974 describes it as several separate "surfaces": one holds auditory information — what you say to yourself silently — another visual-spatial, and a control centre distributes attention between them. Hence a common observation: listening to a podcast and reading a text at the same time is nearly impossible, while listening to wordless music and reading is quite doable. The first demands one surface twice; the second, different ones.

The capacity of this table, as we discussed in the article on chunks, is about four items. And here's the key consequence: everything that takes up space on the workbench takes it away from learning.

If you're working through a new topic while also holding in mind an unclosed work chat, an unclear term from the previous paragraph, and the thought "what if I don't finish in time," there are almost no working slots left for the material itself. From the outside this looks like "I'm slow today." In reality the table is simply occupied.

Hence a practical rule: unload the workbench before you start work, rather than heroically enduring the overload. Writing distracting thoughts down on paper, closing tabs, sorting out the unclear term right away — all of this literally frees a physical resource.

The warehouse: there's always room

With long-term memory the situation is the reverse: its capacity is so vast that you'll never reach a practical limit. The problem is never that "space ran out."

What does happen is a broken intake. The most famous case in the history of neuroscience: a patient known by the initials H.M., who in the 1950s, during an operation for epilepsy, had a significant part of the hippocampus removed on both sides. The seizures receded, intelligence was preserved, old memories stayed in place. But he formed no new long-term memories: a doctor who worked with him for decades he greeted each time as a stranger.

His workbench worked — he held information briefly just fine. The warehouse was intact — he remembered his childhood. What broke was the link between them.

That's how science understood that the hippocampus is not a storehouse but an intake: a region that receives new experience and organises its dispatch to the permanent shelves of the cortex. This distinction explains why knowledge doesn't appear in your head the moment you read: reading only puts the thing on the intake table. It still has to be delivered from there.

The night shift

The delivery is done mostly by sleep.

During the deep phases of sleep the hippocampus replays the day's experience — this was first shown in rats in the 1990s: the cells that fired while the animal ran through a maze reproduced the same sequence of activity at night, only sped up. At night the brain literally plays the travelled route again, strengthening the needed connections in the cortex and letting go of the excess.

Hence two conclusions unpleasant for fans of all-night cramming.

Sleep after studying is part of studying. Material learned in the evening and not backed by proper sleep makes it to the warehouse in a noticeably smaller amount.

Sleep before studying is also part of studying. Lack of sleep hits the hippocampus's ability to take in the new: an under-slept person remembers worse not because they're "sluggish," but because the intake is working at half strength. Studying on four hours of sleep is like unloading a truck at closed gates.

Labels matter more than shelves

Now the main thing about the warehouse. Forgetting most often looks not like a box disappearing but like losing the label. The classic example is the "tip of the tongue" state: you know for certain that you know the word, you remember what letter it starts with, how many syllables — but you can't retrieve it. The box is in place, the label is missing.

Endel Tulving showed that recall always goes by cues — hooks — and whether the information will be found depends on how well the conditions of retrieval match the conditions of encoding. And Fergus Craik and Robert Lockhart, back in 1972, formulated a simple thing: the deeper you processed the material at the moment of encounter — tied it to meaning, to personal experience, to what you already know — the more firmly it's retrieved later. Mechanical repetition out loud creates a box with no label. The question "what does this resemble among things I already know?" creates a dozen labels at once.

That's why metaphors, analogies, and vivid images aren't decoration of the text but a tool of warehouse bookkeeping. Every analogy is an extra thread you can pull the needed thing out by.

We rarely lose boxes. We lose labels.

And one more thing: labels fade over time. How fast, and what to do about it, is the topic of the next article — about the forgetting curve and spaced repetition.

In practice

Unload the workbench before you start. Write anxious thoughts and to-dos down on paper, close what's extra, prepare everything you need in advance. Five minutes of preparation give you more than half an hour of fighting an overloaded attention.

Break the material up. Don't try to hold a long chain whole: work through it in pieces of three or four elements, lock in each, then join them. The table doesn't stretch.

Tie the new to the old. On each new idea, ask: what does this resemble among things I already know how to do? where have I seen this? An analogy you came up with yourself works better than a ready-made one — because it hooks onto your own warehouse.

Plan sleep around hard material. Don't leave a heavy new topic for the dead of night, and don't sacrifice sleep before a day with a lot to memorise. The night shift doesn't work overtime.

Check yourself

Close the article and answer in your own words:

Why don't eight random words fit in your head, while a childhood route does?
What did the case of H.M. prove about the role of the hippocampus?
Why does lack of sleep hinder not only recalling but also memorising the new?
What does "losing the label" mean, and how do metaphors help avoid it?

In short

Working memory is a cramped workbench of three or four items; everything extra on it takes space away from learning.
Long-term memory is a practically boundless warehouse; there's always room.
The hippocampus works as an intake, not a storehouse: reading puts material on the table but doesn't deliver it to the shelf.
Delivery is done mostly by sleep, replaying the day's experience. Sleep is needed both after studying and before it.
Forgetting is usually not a box going missing but a label being lost. Analogies and links to the known are extra labels.