HomePage RecentChanges


Memory palaces

A MemoryPalace is a use of an imaginary journey through a sequence of places, or loci, each of which acts as a memory peg (PegSystem). Memory palaces are also known as the RomanRoom method, the LociSystem, the Ars Memorativa or Art of Memory, Journeys, and by many other names.

You can use a MemoryPalace to remember large amounts of material. Strictly speaking, it need not be a palace; it might be a theatre or temple, for example. As new material is added, a memory palace might extend into a palace compound or temple compound, or even a whole city.

Ancient Roman orators imagined their home populated with different items, real or imaginary, and each was linked to something they wanted to remember in one of their speeches. (By the way, the use of MemoryPalace``s in oratory is the origin of the expressions "in the first place", "in the second place", and so on.)

A memory palace doesn't have to be a real place, though it may start out as one. The advantage of imaginary places is that they can be systematic. GiordanoBruno and other historical practitioners of the Art of Memory built memory palaces around the Zodiac and other systems of imagery and ideas, many of them drawn from Renaissance occultism. They rivaled or even surpassed SemCubed in complexity; Bruno's basic memory system, described in his "De Umbris Idearum" ("On the Shadows of the Ideas"), used 150 "subjectae" (memory places) which could be modified by the addition of any of several sets of 150 "adjectae" (defining images) to yield an almost infinite number of memory places.

Dominic_O'Brien?, a world memory champion, uses hundreds, if not thousands, of locations in his MemoryPalace``s, which he calls Journeys. He doesn't just use locations; He uses short journeys through locations, that he mentally travels. Each location is a stage that he mentally enacts a scene which links to what he wants to remember.

Dominic uses the journeys to record sequence. Using sequences of loci is better than just linking one item to remember to the next item to remember (the LinkSystem), because a break in the chain doesn't mean you've lost the rest of the chain. You can also remember things out of order, by skipping to another locus, ignoring the rest of the journey.

Try it yourself

Consider a journey which is familar to yourself. A hypothetical journey might start in your bathroom, then on to the bedroom, the landing, the back bedroom, the guest room, the stairs, lounge, dining room, kitchen, back door, side passage, front garden, house opposite, etc. Choose a journey that matches the actual layout of your house. Imagine that you are walking along this journey and don't cross over your path or backtrack since this could cause you to either miss some locations in your journey or use the same ones twice.

At each location imagine yourself always looking at the scene from the same location and perspective. Practice mentally following your journeys and be able to visualise the smallest details in each location. Each time you practice this, you will "see" more and more details in each location. Don't forget the colours and textures. The more details you can see in each location the more mental hooks you can use to attach your facts (LinkQuickly).

Once you have your journeys you can use them for different purposes. Some journeys you can reserve for short term memory and others can be used for long term memory. Dominic_O'Brien? has many journeys which he uses and reuses for memory competitions where he needs to remember them for only an hour or so. He also has journeys to remember facts and figures, for example one is a journey around a stables which he uses to remember the winners, jockeys and trainers for all the Derby horse race winners, each stage of the journey represents a year.

Journeys used for short term memory will nevertheless retain some associations for several days. Once you have used such a journey you will need to 'rest' it for some time before using it again otherwise you will find that previous associations interfere with the new ones. Keep several journeys for competitions and short term use and rotate their use. Make your long term memory journeys appropriate for the subject, for example Dominic's use of a horse stable for the Derby winners. You may remember information about the British Royal Families could be in a journey starting at Buckingham Palace.

Expand your journey / memory palace without building a new wing

You may be able to expand the capacity of your MemoryPalace tenfold without adding extra rooms by using the NookAndCrannyMethod.


I read on a web page, that precise instructions on building memory palaces are found in:

I figure these should be on the Internet, or at least on Amazon. Any links?

-- LionKimbro [[DateTime?(2004-04-12T07:35:41Z)]]

I suppose the problem I have understanding the "systematic" explanation, is because it seems to me that systematic things have less value than those that aren't systematic.

For example, say we're imagining a gigantic lattice world of cube-shaped rooms. This is a systematic design. But it's horrible for memory, because every room looks the same.

When you have pieces that all look similar to each other, the pattern becomes one of just RawDesign?. So I don't understand how systematic memory systems help.

But I also think so many pleased customers can't be wrong. So, I'm still wondering how these imaginary worlds are better than physical worlds. I'd like to better my technique by diving into imaginary palaces, but it seems like exactly the wrong way to go, since they have no day-to-day reinforcement.

-- LionKimbro [[DateTime?(2004-04-12T21:36:04Z)]]

Well, as the "Hotel Buzan" and "Hotel Dominic" analogy illustrates (SemCubed), there is more to systematic memory systems than featureless lattices. For example, SemCubed can be thought of as a 10x10 lattice of suites containing 100 rooms, but each of those rooms has a different mnemonic. Room 8001 is the Red Day Room, whereas Room 8101 is the Orange Day Room. The regular features of each room provide enough detail for all the pegs you need. It seems that people such as Giordano_Bruno? who used the Zodiac and other regularised systems as the bases for their memory palaces might have had more in common with today's mentats who use SemCubed and the DominicSystem than they did with classical users of the ArtOfMemory?.

-- Ron Hale-Evans [[DateTime?(2004-05-20T02:39:37Z)]]

Then, I suppose, the Memory Palace is not so much about places, as it is about structured arrangements of memorable things.

Like, I could take my peg list, and call it a memory palace, because I can go from the Tie room, to the Noah room, to the Ma room, to the Rie room, to the Law room, to the Shoe room, to the Cow room, to the Ivy room,...

Does this say something more than just LinkSystem or PegSystem?

I think so: Pegging is just about having easy to use hooks. Links are between two things.

A MemoryPalace is a graph of mnemonics, then, by this looser concept of a MemoryPalace.

But I would want a stronger notion of a Memory Palace.

If you were an astronomer, and knew the positions of stars, then you could make a memory palace out of the zodiac. This is because you'd have lots of stars in each "room" to hang things on, provided you know the stars well enough to assign a personality to each one.

But if you're someone who's just memorized the zodiac, would we want to say that this was a MemoryPalace? I'd say, "no," because you don't get all the goodies that you get from a MemoryPalace- lots of space to hang things from.


If you could build a systematic memory palace that had lots of hooks to it, I think that would be very interesting.

Hmm... Still thinking about this.

-- LionKimbro [[DateTime?(2004-05-20T07:23:13Z)]]

Well, the traditional Zodiac, as it was understood in Bruno's time, has, among other divisions, twelve signs: Aries=Ram, Taurus=Bull, Gemini=Twins, Cancer=Crab... Plenty of pegs there.

You seem to be having trouble reconciling the idea of a matrix or lattice with the idea of such a matrix having features. I think your confusion is coming either from (1) your thinking about the Zodiac in today's terms (as a bunch of stars that are essentially featureless from a mnemonic perspective), or (2) your not realising that there's a lot more to a system like Bruno's than just these twelve featureful signs. IMHO, you can think of the signs as corresponding to the broadest divisions of SemCubed (Colors, Animals, Birds, the Solar System...), each of which has many subdivisions, the lowest level of which is the "room".

Think of a system like Bruno's as a matrix or lattice, but not a featureless one.

-- Ron Hale-Evans [[DateTime?(2004-05-20T08:15:16Z)]]

Oh, no- Actually, you made it pretty clear to me, in your responses, that we are not talking about featureless patterns of rooms. I should adjust the Wiki:DocumentMode? text to reflect what I understand now.

But my new misunderstanding is about: "How does taking a list, and calling it rooms, help?"

My understanding of how the MemoryPalace works is this: They work because we see these places all the time (in our day to day lives,) and are thus very familiar with them. They also work because our rooms have lots of details, details that we know very well. For example, we know where the booksshelf is. We know where the TV is. We know where our cupboard is. We know where the couch is. We know where the lights are. We know all these little details that we hang stuff on.

But when you take something abstract, like a peg list, or the list of signs in the zodiac, then it seems to me you miss all those details. You also miss out on the fact that you don't daily walk through these structures.

-- LionKimbro [[DateTime?(2004-05-20T18:59:16Z)]]

I've just developed an interest in Memory Palaces and would appreciate any Info, help or tips anyone has to offer. I have found an English translation of Quintilian's "Institutio Oratorio" at : http://penelope.uchicago.edu/Thayer/E/Roman/Texts/Quintilian/Institutio_Oratoria/home.html Havnt found any English translation of "Ad Herrenium" or Cicero's "De Oratorio" but I'll keep looking. Angela

According to Frances Yates' "The Art of Memory," Robert Fludd, who was a practitioner of the art of memory back in the 17th century, insisted that real places were much better memory palaces than imaginary ones. I haven't yet been able to score Fludd's treatise on the art of memory, so can't comment on his reasons, but I'd agree with him precisely because every room in a real place is different from every other room, as suggested above.

I used to use some of the old buildings at the University of Washington as memory palaces -- the old ones, because they had been remodeled several times and no two rooms or hallways were the same. With practice, it takes only a couple of walkthroughs to get a good chunk of a building stored in memory, and then you can map your memory images onto it.

-- John Michael Greer

(A public response I've written to respond to a question I received in email.)

Sure, but then what use the memory palace?

Just say, "pick these three things by this indexing system, and then associate your item with those three things."

I don't see how calling it a "room" helps, and I don't see how "connections" between the room function. It is, instead, rather, just a change of indexes and search for neighbors, to get from room A to room B, rather than an actual mental traversal from one room to another room.

If I use my apartment for a memory palace, the entrance is connected to the living room because, in reality, the entrance is connected to the living room, and I long ago memorized that.

But if I use a mathematically constructed cube, with different items for indexes, then I don't "walk" from room to room. Rather, I'm adjusting an index in a process, and then recollecting association from there.

It's not a memory palace, and, I usually find that it doesn't work so well.

If I instead construct memory palaces manually, out of places I've seen or spent a lot of time and effort constructing, and then tie things to those constructions, I think it will (A) be true memory palace, and (B) work better.

-- LionKimbro

I am inclined to agree with LionKimbro's comments, especially that the real world memory palace would have more 'pegs' and ultimately richer detail with which to bind new knowledge. A second advantage to the physical world memory palace is that if you forget a part of it, you can walk over to that part and refresh your memory.

If learning is the process of making mental connections and here I am thinking specifically about learning factual knowledge like a foreign language word's meaning, then the world we see every single day has the potential of more modalities to link. Look at it this way, if there was only a few imagined details in a mental construct, how many ways can you merge new knowledge? Example: a red room in a memory palace of 10,000 rooms. What do you know about it? It is red. It is a room. It seems like a make-work activity to create an imaginary room when you already have in long-term memory your own home to use as a memory palace or as a journey.

Outside my window I see the topography, the trees, the bitter cold snow, smell the salt of the sea, hear the gulls crying and watch my neighbor's tubby belly jiggle as he blah blah's on the cell phone. That is a rich & sensual impression that has many more natural pegs to link to than anything I could make up.

However as a final note, I want to say that anything that works is the thing to use. I just wonder about the durability of memories using both methods. Just how long does Brian O'Dominic remember a single memorization of a pack of cards? Does this have real world application beyond a very fascinating parlor trick?

One thing to take into account in this discussion is that different people have different visual memories. If you have strong visual memory and can recall most places you've been in extreme detail, then real-world places are better pegs for you than constructed places. However, other people (including me) don't recall visual details that way; instead, I remember things I've seen more as abstractions than as rich details. So, stylized or iconic images (like cartoon characters or cartoon representations of places) are easier for me to recall than even what my own room looks like. A constructed place approach is just fine for me.

Another thing to consider is the organizational principle. When you visualize either a journey (going to work) or a constructed place (a Red Elephant room) to memorize a shopping list, neither method is going to work unless you also remember "I stored my shopping list in my route to work/the Red Elephant room". You need a systematic way of finding your MnemonicCues to trigger the recall.

-- JohnLaviolette [[DateTime?(2007-10-06T03:43:41Z)]]

Though I'm very new to memory techniques, I'd like to add some ideas. Please delete my thoughts if they seem wrong to you or aren't appropriate. I think that LionKimbro could be right about real-world places being richer and therefore easier to use than systematically constructed places. I'd disagree though about peg systems being the same as a "room".

The first example I'll give might seem a bit far fetched, but I think it has something to do with our topic. Please, think of the so called "bell curve" (http://en.wikipedia.org/wiki/Gauss_curve) associated with Carl Friedrich Gauss, also known as the normal distribution. At least for me, it was quite easy to memorize the shape of the curve and also not too difficult to adjust it visually given mean and variance. Although the associated formular is only another representation of the same information, it's already harder to memorize for me. Though not as hard as memorizing all the numeric function values of the curve would be. This is interesting in itself, because by simply remembering the shape of the curve and drawing it in a standardized coordinate system, approximate values for the function values could be easily derived without having to remember only one single number. My conclusion is that it is much easier for our brain to remember points or lines which have certain spatial relationships to another than mathematical formulars or even plain numbers.

Or imagine you're given a large dataset of numbers in a table to analyse. If you want to find relationships between such values, often it can help to visualize them in two or three dimensional space. Again, it seems to be much easier for our brain to process the spatial relationships of points or lines in space than it is to look at just the plain numbers. Often, you'd suddenly just "see" something in that graph. In such a case, one really could say that an image isn't only worth a thousand of words, but a thousand of numbers.

The relationship to LionKimbros? problem is that while going through a peg system, "adjusting an index" IMHO is an activity rather associated with the brain operating in "left-mode", so to speak, than in "right-mode". It's much like counting: "What number comes after 7?" isn't much different from asking "what peg comes after the lion in the major system", for example. Trying to imagine the bell curve though - at least for me - is very much "right-mode": I can see the curve as a whole when I'm visualizing it. I'd not go through the areas of the coordinate system, asking myself "how does it look in quadrant 1, 2, 3, 4".

So the advantage of putting pegs into a room could be that in comparison to a plain peg system, the brain gets additional information (position and spatial relationships) which can be accessed almost immediatly by visualizing the room, because our brain is very good at that. Imagine for example putting a lion on the top of the bell curve, a skunk at its left end and a horse at its right. In case one has really visualized such an image (it might be best to really draw it), I'm sure that to recall the pegs it's only necessary to visualize the picture you've drawn again. One could immediately "see" that the lion is on the top and where the other pegs are. One doesn't need to go through each position systematically, asking like "What was on the left?" or "What was on the right?". When I'm recalling the first scene of my favourite movie, Star Wars IV, I'd know immediately that the destroyer comes from the right and that the rebel ship is in the middle. I'm sure that the effect of visualizing a room with pegs several times could lead to the same result of immediately knowing where they are. And it doesn't seem to play a role how "rich" the scene is.

Nevertheless of course one needs a very good ability to visualize things and I'm not sure if anyone of the "old masters" ever used those systems in such a way.

--Dp [[DateTime?(2008-27-02T03:43:41Z)]]

Use google sketchup to build a palace, and populate it with items from google's 3d warehouse. After you do a few walkthroughs, you'll remember it quite well. The 3d warehouse has just about any image/object you can think of...furniture, statues, etc. If I could get google earth to work on my computer, I'd upload my ( rather small ) palace for all to see. It's all free, just in case you're wondering.

Using old D&D dungeon modules as places to frame hooks

I know this might sound bizarre. But I recently wanted to try a variant of this method. But I did not want to use my house or some other such mundane place. I wanted something fantastical. Well, I don't have the time to design something truly fantastical and filled with things and pegs. But I do have one document that does contain all those things. It is called the Tomb of Horrors. There also is another great selection: The Temple of Elemental Evil. But the former is smaller even though it is a vicious killer of your D&D characters.

I figured I would just build a plan and aspects on paper. Then construct a model in blender of something (I have linux) My Rhino distro is useless along with anting I can get for Lightwave or Maya. I don't need anything fancy. If Google ports their app to linux, then I would be very happy. Meanwhile, I guess I need to get used to Blender argghhh.

This campaign is filled with colors, portals, and choices. Lots of pegs. I will have to invent a random number generator from the time to make some of the portals work properly.

If it does not work, I will have at least undergone a fantastic mental exercise -- and my be the only man in the world who has the tomb completely constructed inside his memory.

Any ideas about how to make this work? Does this seem like too much work to adapt to a memory project?

Oh.. forgot to mention all the first edition modules shipped with ink drawings of each room and hallway of note.

---DintVerge? AT g mail

I think you might be on to something here, DintVerge?.

I note that Tomb of Horrors is readily available as a PDF bittorrent, for example from http://thepiratebay.org. Arrr!

-- Ron Hale-Evans [[DateTime?(2008-06-24T20:52:01Z)]]

Don't you all think that something like the maps in some computer games would be an easy and fun (if you enjoy FPS or RPG games) to construct a quite detailed memory palace?

-- laddiebuck

Yes! I have tried using real life locations, made up imaginary locations, and video game levels for memorising. I find video game locations win out and are the most effective method for me, to the extent that for the last few years I have stopped using locations other than video games for memorising.

Some games work better than others. Ideally a game should be:

-- ThufirHawat

Before talking about real loci vs. imaginary loci, I want to ask if people have the same experience I do with fidelity vs. speed. Like, the richer your actions and connections, essentially the longer they last in memory without review, but the longer they take to peg. I've always used my major list 1-100 as like a back up for when I run out of room lists. I have some real lists, and some imaginary lists. I don't think that there's much difference to me in routine use, but you know if I'm trying to do it as fast as I can, the imaginary lists seems to work better. When I'm trying to make something incredibly vivid, I think the real lists are better. Just my experience. -Cassox

Hi everybody - I learned a lot by reading your wiki. I'd like to receive your advice: I am not interested in competitions, nor in simply learning long tedious lists. I'm more interested into the language. So, to ask for some specific info, do any of you know about using the Palace for:

... sorry, they are 2 VERY different questions - but I'm very interested. Thank you for your help. -- Alessandro

Alessanadro, on learning PinYin? (Chinese phonetic pronunciation/writing system for anyone who isn't aware familiar with it), I've been thinking about exactly the same thing after happening across this site after watching a TEDx talk on memory techniques. I've made a first draft of an encoding system for PinYin? text but I haven't spent too much time trying it out yet. It would be interesting to discuss it further. Maybe we should create a PinYinMemorySystem? page? My general approach is to look at all the independent elements that could vary in each PinYin? syllable. I map each element to a different visualizable attribute. The hard part is that they all have to be independent. So I break it down the following way. The PinYin? word structure is <initial consonant> {<trailing vowels/sounds> could be several of these} <tone>.

Like I said I've only started with it but I think it could work. I'd be interested in talking more about it if anyone has ideas. I'll fill this in more later when I get using it more.

-- MarkCrowley?