How To Learn The Memory Palace Technique: 11 Science-Backed Steps To Remember More

how to build memory palace
how to build memory palace

If you want to know how to learn the memory palace technique, you’ll love this article.

You read a list once. Ten minutes later it’s gone, apart from the first item and maybe the last. Now picture walking through your own kitchen and finding every item waiting for you: one on the counter, one on the kettle, one stuck to the fridge door. That’s the whole trick. The memory palace, which researchers call the method of loci, borrows the part of your brain that never forgets where the bathroom is and puts it to work on the things you actually need to remember.

The idea is old. The usual origin story credits the Greek poet Simonides, as retold by Cicero. What’s new is the evidence. A large 2025 meta-analysis, a handful of training trials and a 2026 reanalysis of brain scans from world-class memory athletes have given us a clearer picture of what works, what doesn’t, and how long it takes an ordinary person to get good at it.

Below are 11 steps. Some are about building the palace. Some are about filling it. A few are about keeping it from falling apart, which is where most people quietly give up. If you want a wider toolkit, our guide on how to improve your memory covers other methods that pair well with this one.

1. Pick a place you can walk through with your eyes closed

Your first palace should be somewhere boringly familiar. The flat you live in. Your parents’ house. The walk from the bus stop to your office. Not the cathedral you visited once on holiday. You need to “see” the place without effort, because every bit of mental energy you spend rebuilding a room is energy you’re not spending on the material.

Is the palace worth the bother in the first place? The evidence says yes, even if the studies behind it are messier than you’d hope. In a systematic review and meta-analysis published in the British Journal of Psychology in June 2025, Jan Ondřej pooled research on adults and found a large advantage for the method of loci over plain rehearsal when people had to recall items in order (d = 0.88). That’s a big effect by the standards of memory research. Ondřej was also frank about the weak spots. He rated the overall quality of the evidence as low to very low, mainly because of risk of bias, and most participants were university undergraduates. So read it as a strong signal, not a guarantee. The review also lists personal relevance among the reasons the method works, which is one more argument for using a place that means something to you.

Your move: take a sheet of paper and sketch the floor plan of your home from memory in two minutes. If you hesitate over a room, choose a different place.

2. Fix one route and walk it in the same order every time

A palace isn’t a picture. It’s a path.

Decide where you enter, which way you turn and which spots you stop at: front door, shoe rack, mirror, coat hook, kitchen table. Then walk that path in your head until it runs on rails, and walk it backwards too.

The reason is mechanical. When you store an item at a stop, you tie it to a location. Because the locations sit in a fixed sequence, arriving at stop three hands you the cue for stop four, and the order of your list comes free with the order of your walk. Wander in at random or skip rooms and that chain of cues breaks, so you’re back to fishing in a loose pile of images. Brain imaging fits this account, though it doesn’t prove it. In a reanalysis published as an eLife reviewed preprint in April 2026, Jingyuan Ren, Martin Dresler, Isabella Wagner and colleagues found that after six weeks of method-of-loci training, activity patterns in the front part of the hippocampus during an order-judging task were better explained by where items sat along the route than they were in untrained controls. The trained group was small (17 people), and eLife‘s reviewers rated the evidence for some of the paper’s main claims as incomplete. Supportive, then. Not decisive.

Give it a go: choose 10 stops in your home, write them down in order, then close your eyes and walk them forwards and backwards twice.

3. Start with a short list that doesn’t matter

The temptation is to go straight for the exam syllabus or the 40-slide presentation. Resist it. Begin with 10 to 20 throwaway items, like a shopping list or a list of animals, so mistakes cost nothing and you can watch the technique working.

That’s roughly the setup Monash University used. At its rural health campus in Churchill, Victoria, a team led by neuroscientist David Reser, working with Tyson Yunkaporta, gave 76 first-year medical students a single short training session (Monash’s own reports put it at 20 to 30 minutes) in either the memory palace, an Australian Aboriginal narrative technique, or no technique at all. Then the students learned 20 butterfly names. The study ran in PLOS One in 2021. According to Monash’s press release, students in the palace group were about 2.1 times as likely to get a perfect score after training as before it, while the untrained group improved by about 1.5 times just from repeating the test. The Aboriginal method beat both. A six-week follow-up drew only eight students, too few to say anything about long-term retention. Reser himself admitted the palace can be “intimidating to get started”, which is exactly why a tiny first list makes sense. And if a voice in your head is already saying you’re just not a memory person, read our piece on how to stop underestimating yourself.

Start small: put tonight’s shopping list on your 10-stop route, then shop without looking at your phone.

4. Make every image alive and moving

Each item needs an image, and the image needs to do something. A loaf of bread sitting on the doormat is forgettable. A loaf of bread waddling across the doormat and kicking your shoes over is much harder to lose.

This specific point has been tested. Janell Blunt and Joshua VanArsdall reported two experiments with novice users in Memory & Cognition in 2021. In the first, people recalled more words with the palace than with a pleasantness-rating strategy, and words for living things beat words for objects. In the second, participants were told how to picture each word, as animate or not. Words for living things, and words people imagined moving, were both recalled better, and the boost seemed to add to the palace’s own benefit rather than replace it. One caution: this is a single paper using lab word lists, and our research turned up no newer replication specific to memory palaces.

Picture this (a hypothetical example): your list reads “stapler, milk, passport”. By the front door, the stapler snaps at your ankles like a small, bad-tempered dog. Milk crawls up the mirror. And at the coat hook, the passport flaps about in a panic, a trapped bird made of paper that keeps slapping against the wall.

Next time, do this: give every item a verb before you place it. If it can’t move on its own, make it.

5. Give each spot one image, and keep the spots different from each other

Think of your loci as coat hooks in a hallway. One coat per hook and you grab the right one without thinking. Pile three coats on one hook and you’re rummaging. Fit two hooks so close together they look identical and you’ll hang things on the wrong one. The comparison holds for crowding and confusion. Where it breaks down is capacity: a mental spot has no fixed size, and experienced memorisers do sometimes store a small cluster of images at a single location. As a beginner, though, stick to one per hook. It’s the same logic as getting organized at work: a place for each thing, and each thing in its place.

The analogy has some backing from the brain-imaging side. In the same 2026 eLife reanalysis, Ren and colleagues found that after training, people produced more distinct prefrontal activity patterns for each word they placed. The more distinct those patterns became, the better trainees recalled word lists four months later, although the link was modest (a cross-validated correlation of r = 0.23 across 44 people). The authors read this as the palace working partly by making each memory unique. The same caveats apply: a reviewed preprint, small groups, and claims the reviewers asked the authors to tone down.

Before you place anything, walk your route once more and ask of each stop whether you could mistake it for another. If two spots look alike, two identical bedroom doors for instance, drop one and choose something with its own shape, colour or texture.

6. Link each image to its spot in a way that makes sense

Random pairings feel creative. They’re also harder to hold onto. Where you can, make the image fit the place: milk by the fridge, a passport by the front door, a toothbrush in the bathroom. Your existing knowledge then does half the work.

This seems to matter more as you get older. Laure Debroux, Christine Bastin and Emma Delhaye at the University of Liège tested young adults (18 to 30) and older adults (60 to 75), and published the results in the Journal of the International Neuropsychological Society (August 2025 issue, online October 2025). Their study of the method of loci with meaningful versus mismatched links found that both age groups improved when the pairings agreed with what people already knew. With mismatched pairings, the older adults did no better than with no strategy at all. With sensible pairings, they recalled as much as the young group. It’s one lab study with word lists, but it’s a useful nudge, especially if you’re over 60 or teaching the method to a parent.

A quick experiment: build a second 10-stop route. Learn one 10-item list on route one with sensible placements and another on route two with deliberately random ones. Test both the next morning and see which held up.

7. Practise every day for about six weeks, not once in a burst

This is the step nobody likes. The palace is a skill, and skills take reps.

The best-known training trial gives a concrete dose. In Neuron in March 2017, Martin Dresler and colleagues, including memory athlete Boris Konrad, trained 51 people with ordinary memories. After 40 days of 30-minute daily sessions, the method-of-loci group went from recalling about 26 words out of 72 to about 62. Four months later, with no further training, they were still recalling roughly 22 more words than at the start, according to press coverage of the study. People who trained short-term memory instead gained about 11 words, and the untrained group gained about 7 just from repeating the test.

That trial is now more than nine years old. We’ve kept it because it’s the foundational study, and the 2026 reanalysis by Ren and colleagues uses the same data and spells out how the training actually ran. Participants first learned a 50-stop route in and around the Max Planck Institute of Psychiatry in Munich, then built three routes of their own in the first two weeks. Daily online practice started with five words and added five more after every perfect run. Sixteen of the seventeen trainees reached the top level of 40 words.

Here’s a simple way in: block 20 to 30 minutes a day at the same time and treat it like any of the good habits you want to keep. If your commute is when you’d normally doomscroll, walk a route in your head instead. Start at five items and add five whenever you get a list perfect.

8. Test yourself by walking the route without notes

Recognising your list when you look at it is not the same as remembering it. Close the notes, walk the palace, say or write what you find at each stop, and only then check.

Jeffrey Karpicke, a psychology professor at Purdue University and one of the leading researchers on retrieval practice, sets out the case in a 2025 review chapter for Elsevier’s Learning and Memory: A Comprehensive Reference (third edition). His summary is that a test changes what you know, because the act of retrieving strengthens it, and that the effects are robust and long-lasting. He points to an earlier meta-analysis by Christopher Rowland (2014) in which about 81% of 159 comparisons favoured retrieval over restudying, with an average effect of g = 0.50. Two details in the chapter matter for palace users. Retrieval helps most when it takes effort but still succeeds, so recalling from cold beats rereading. And in Karpicke’s own lab work with Megan Smith (2012), plain retrieval practice outperformed imagery strategies such as the keyword mnemonic for learning vocabulary. That doesn’t make the palace weak. It means the palace stores the material, and recall practice is what makes the storage last. Use both. These sessions work best when you can concentrate fully, so our tips on getting into a flow state are worth a look.

Test it this week: after each session, wait ten minutes, then walk the route aloud from memory. Mark every stop where you stalled. Those images need rebuilding, not rereading.

9. Space your reviews, and try a week between them

Once a list is in, don’t hammer it five times in one evening. Come back to it later. Then later again.

The newest classroom evidence backs this up. Rhys Mawson and Sean Kang at the University of Melbourne published a 2025 meta-analysis of spaced practice in real classrooms in Behavioral Sciences in June 2025. They screened more than 3,000 articles and analysed 22 reports with 31 effect sizes, covering more than 3,000 learners. Spreading practice out beat bunching it together by just over half a standard deviation (d = 0.54). Two patterns are useful here, though the authors stress they come from grouping the studies descriptively rather than formal testing. Fixed seven-day gaps between sessions gave the most consistent benefit. And studies with fewer re-exposures, around three, tended to show bigger effects than those with many. The studies varied a lot, so your results will too.

If reviews keep sliding off your calendar, these strategies to beat procrastination can help.

Your move: for anything you need long term, put three review dates in your calendar, the next day, a week later and a week after that. On each date, walk the palace cold before you look at the list.

10. Reuse a palace only after the old images have faded

Sooner or later you’ll run out of routes and want to recycle one. You can. Do it too soon, though, and here’s what happens. Each stop already has an image tied to it. Place a new image there and the stop calls up both, so they compete. The more you reviewed the old list, the stronger it is and the harder it pushes into the new one; hobbyist memorisers call the result “ghosting”. Images you stop reviewing fade over time, which frees the spot again. Interference of this kind is well established in memory research generally. The exact timing for palaces isn’t.

The classic test is old. In 1988, Rossana De Beni and Cesare Cornoldi of the University of Padua had 105 university students learn three lists of word pairs on the same loci, reporting the results in Perceptual and Motor Skills. The palace still beat the control strategy, but performance suffered from interference between the lists. We found no modern replication with accessible results, so treat the effect as real and its size as unknown.

The practical rule follows from that: keep anything long term on routes you never reuse, and rotate two or three scratch routes for throwaway lists, leaving at least a few days between uses and a longer gap if old images still pop up.

11. Build a palace for real work, and know where it fits

Once shopping lists feel easy, move on to something you care about. The key points of a talk. A procedure at work. Vocabulary. The names of people you’ll meet at an event. Give each subject its own palace.

Early evidence suggests the skill can carry over into knowledge work. In a randomised trial posted to SSRN in June 2025, Evangeline Wandag of the Research Data Alliance in the Philippines assigned 96 remote visual artists and copywriters to either six weeks of online memory palace training or a productivity course. The palace group improved on work-related memory retrieval tasks (d = 0.52), and most of the gain held at a three-month follow-up (d = 0.47). It’s a single-author preprint that hasn’t been peer-reviewed, so call it promising rather than settled. Stay realistic about scope, too: the authors of the 2026 eLife paper note that the method suits ordered or arbitrary material and transfers poorly to tasks that don’t involve that kind of memorising. It won’t make you better at chess or spreadsheets.

Give it a go: put the five main points of your next meeting or talk on a five-stop route around your desk, then speak without notes. For events, store names the same way. Remembering someone’s name and what they told you is half of a good conversation.

What to do when you slip

You’ll miss days. You’ll open a palace and find a stop empty. Neither means the method has failed you.

Start with the long view. A systematic review led by Ben Singh at the University of South Australia, published in Healthcare in December 2024, pooled 20 studies with 2,601 participants. In the studies that measured it, new health habits took a median of 59 to 66 days to form, with individual estimates ranging from 4 to 335 days. Singh’s advice was to “not give up at that mythical three-week mark”. Coverage of the review also reports that one of its trials (Keller and colleagues, 2021) found that missing a single opportunity didn’t materially change the overall habit curve. These were health habits, not memory training, so the parallel is indirect.

The memory research is reassuring in its own way. In Dresler’s 2017 trial, the trained group still recalled far more words four months after training stopped, so a skill built like this seems to fade slowly. Even that trial didn’t demand perfection: when someone missed a session, they were simply asked to add extra time the next day (as Ren and colleagues describe in 2026).

So when a session goes missing, do a normal one today with five extra minutes, and don’t try to cram a week’s worth into one sitting, which is the massed practice that step 9 warns against. When a stop comes up blank, rebuild the image on the spot: bigger, moving, and linked to the location in a way that makes sense. Then test it again ten minutes later. If frustration is building, shrink the list back to ten items for a few days, or calm your nervous system before you try again. If the whole day has gone badly, turn the bad day around first and come back tomorrow. And treat the next session as a small promise you keep to yourself, not a test you can fail.

Conclusion

The memory palace isn’t magic, and it isn’t only for memory champions. It’s a learnable skill with a fairly clear recipe: a familiar route, one vivid and sensible image per stop, a few weeks of short daily practice, and self-testing spread out over time. The research has holes, with small samples, lots of student volunteers and a couple of preprints, but it keeps pointing the same way.

Pick your route tonight and put ten things on it. Then browse our Education section for more science-based ways to learn faster.


Przemkas Mosky
Przemkas Mosky started Perfect 24 Hours in 2017. He is a Personal Productivity Specialist, blogger and entrepreneur. He also works as a coach assisting people to increase their motivation, social skills or leadership abilities. Read more here