The Memory Palace: How the Method of Loci Works, and What It's Good For

Try to hold twenty unrelated words in your head, right now. Ten minutes from now you'll have seven left, maybe eight, and almost certainly not in the right order.

Now describe the walk from your front door to your bed. Door, hallway, coat rack, first room, the corner by the window, and so on. You did it without concentrating, in order, with your eyes closed, even though you never memorized it.

That asymmetry is the whole of what you need to understand about the memory palace. The technique doesn't make your memory better. It takes information that won't stick to anything and hangs it on information that holds by itself.

A Collapsed Roof and a Poet Who Knew Who Had Sat Where

The story has circulated for over two thousand years, in the form left to us by Cicero and Quintilian. The Greek poet Simonides of Ceos had just finished reciting at a banquet when he was called outside. While he was gone, the roof of the hall collapsed on the guests. The bodies were unrecognizable, and the families couldn't claim their dead. Simonides identified every one of them — not by face, but by the seat at the table where each had been sitting.

Whether it happened exactly this way doesn't matter. What matters is the conclusion the ancient rhetoricians drew from it: the order of the places survives even when what filled them has turned to dust.

The "Rhetorica ad Herennium," a Latin rhetoric handbook written around 85 BC, already lays out the whole recipe. You pick a building you know, fix its locations into a set order, place an image that reminds you of an idea in each spot, then walk through the building mentally while you speak. Roman orators gave hours-long speeches this way, without notes, without skipping a beat. Frances Yates reconstructed the whole tradition in "The Art of Memory" (1966), the book that brought the subject back into modern discussion.

The Latin word for places is loci. That's where the technical name comes from: the method of loci. "Memory palace" is the popular name, but your palace can just as well be your kitchen.

You're Not Building a New Memory, You're Latching Onto One You Already Have

Here's the core of it, and it's simpler than it sounds.

Spatial memory isn't trained on purpose, it forms on its own. Nobody has ever sat down to memorize the route through their own house, and yet everybody has it, complete, in the right order. You've made that walk thousands of times, and your brain kept the order for free.

A shopping list has none of that going for it. Twenty unrelated words have no built-in order, no shape, no address. That's exactly why they slip away.

The method of loci doesn't make your memory for words any better. It takes a structure that already exists and uses it as scaffolding: every item gets a place. The order no longer has to be held in mind separately, because the order belongs to the route, and you already knew the route.

The sign that the mechanism really is spatial is something mnemonists check often: with a properly built palace, you can say the list backward too, starting from the end and walking the route in reverse. With a list worn smooth by repetition, that's almost impossible.

The Second Engine: You're Forced to Make the Image

This part gets lost in short explanations, even though it does half the work.

To hang the word "tax" on the coat rack in your hallway, you first have to decide what a tax looks like. A piece of paper that bites? A clerk dangling from the hook? Making that decision is the work. You can't place a word superficially, because until you have an image, there's nothing to place.

Bower (1970) described exactly this mechanism: items linked by a mental image that pulls them together into a single scene are retained far better than the same items repeated out loud. It isn't the image itself that makes the difference — it's the processing it demands of you to produce it.

That's also why rereading a list gives you so little back. Rereading is fluent, so it feels efficient, but it doesn't force you to do anything with the material. The same confusion between ease and learning sits behind the feeling described in I read but remember nothing.

What the Measurements Show

Roediger (1980) compared several mnemonic techniques on ordered-recall tasks. The method of loci came out among the strongest, and its advantage was largest exactly where order mattered, not just plain recall of the items.

Maguire and colleagues (2003) scanned the brains of some of the top competitors at memory championships. They found no different brain structures, and no unusual intelligence scores. What they found was something else: while these people memorized, they activated the regions used for spatial navigation. Nearly all of them reported using the method of loci.

Dresler and colleagues (2017) took people with no prior experience and put them through roughly half an hour of training a day, for six weeks. The number of words reproduced from a long list more than doubled, and a good chunk of the gain was still there four months later.

Legge and colleagues (2012) tested a practical question: do you need a house you've known your whole life? Palaces built in a virtual environment learned on the spot produced results comparable to familiar spaces, on the condition that the route was learned before anything was placed in it.

It Works on Ordered Lists, Not on Arguments

Everything that's been seriously measured deals with the same type of material: discrete items, held in order. Words, digits, playing cards, names, terms. That's where the technique is among the most powerful known, and it's what memory-championship competitors actually use.

There's no comparable evidence that a palace helps you understand or retain an argument. A book chapter isn't a list. It's a claim, a reason supporting it, an objection, a counterexample — and the links between them are exactly the part that matters. You can put section headings on a route and get a table of contents you can recite, which isn't the same thing as having understood the chapter. This limit deserves its own discussion, and it gets one in can you use a memory palace for a whole book.

There are two more costs nobody mentions upfront. The first placements are slow and feel a little ridiculous, and the palace only becomes fast once the route has become reflexive. And if you reuse the same route for two lists back to back, they blur together: either you keep separate palaces for separate things, or you let one fade for a few days before reusing it.

Four Rules That Separate a Palace That Holds Up From One That Collapses

• The route has to run on autopilot. A place you have to work to walk through eats up exactly the attention you need elsewhere. The house you grew up in, your commute, the gym: anything, as long as you don't have to think about it.
• Distinct locations, in a fixed order. Three identical doors down the same hallway will blur into each other. And the order gets set once and never changes, or you lose the exact advantage you came for.
• Exaggerated images, not discreet ones. The Latin manuals called for imagines agentes, images that do something: big, moving, absurd, sometimes disgusting. A neatly written label on the wall fades within hours. A ridiculous scene stays.
• One item per spot. The temptation is to cram three things into one room to save on locations. The result is that you remember one and lose the rest. Better to use a longer route.

If you want to feel this for yourself, the shortest path is a list of ten objects placed on a route with ten locations: how to memorize a list in five minutes.

The boring part is worth saying too, right at the end. The memory palace isn't a talent you either have or don't. It's a procedure, and what the data on champions shows is that the procedure, not the brain, makes the difference.

See what a reading and memory training session looks like