Systems analysis of visual memory systems

found a bit more in your reply here:

1 Like

Thanks for watching and reviewing my work. I need other analyst types to help confirm and critique.

I’ve never seen any information on bottom-up vs. top-down methods of constructing palaces. I have seen a mention of having image keys already situated in the background vs. placing your own image keys there, but without noting how important it is for long-term retention and structuring. I’ll give you a brief summary here to get you started if you are planning to use it or just want to know the differences as it relates to my progression in using memory palaces.

In the first few system iterations I tried, in the top-down construction style, where I took a pre-built palace filled with number-peg objects and populated it with stories of Bible verse encoded images, I found it created many bizarre and uncomfortable images, which tended to be unstable over time and created a need for too much review. Another side effect was that, I would say, the objects themselves were very unfriendly with each other without any type of association to their roommates. :grimacing:

The bottom-up construction style resulted from wanting to structure the Bible chapters better into smaller chunks for following the psychological rule of chunking, also known as Miller’s law. I had used common pegs for chapter numbers to help traverse the book in order. Instead, I revisited all the 261 chapters of the New Testament to see if it was feasable to encode them into unique images by way of the book palace context. (Matthew is a Math University building and Matthew 12 is now a dean’s office using the Major system)

I had to extend my list of Major system peg words as much as possible for options to select from that fit the theme of the book. My chart has somewhere between 10 and 30 peg words to select from. That exercise in peg word discovery gave me a mental traversal method for reviewing all the options but after a while I found that the familiarity with the selected peg was getting better.

Then I noticed that my peg words were able to be connected to a background that suggested itself. This is the bottom-up construction method. Some peg words were swapped for better relevancy based on that background. Matthew 12, the dean’s office, is in the administrative area of the math building. Matthew 13 is a team meeting in that area that I participate in.

The backgrounds are easy to traverse with a story as I move through the student commons area, the admin area, the classroom, a buiness partnership area, and the exterior of the building. Even the backgrounds, once detailed out in my imagination suggested improvements to my book image. For the smaller books in the NT, the book was the background.

I hope this shows you why my preference is for the bottom-up method that comes with more analysis, in my case, about three years. Without the time necessary for preparing the palace after the familiarization stage because you’ve chosen a value of competition in your strategy phase, you have to skip over that and emphasize the steps of the preparation phase using a top-down palace.

If I can help with any other details, let me know and your suggestions will always be welcomed.

Edit June '21 - Bottom-up construction I now refer to as a creating and learning purpose or dynamic system for knowledge since it occurs first and does not have a definite end or goal. This is a short-term recall or competition system. Top-down construction is now a using and teaching purpose or a repeatable system (concrete or realized are the software design words) where a system overlays known information for long-term recall.


Not sure what you are doing but interested in what you have done! Thanks!

1 Like

Reading through your replies, it seems that your palaces have 2 additional layers of association. You chose a location based on a substitution word of the book (matthew=math building, peter = church, 1peter = low church, 2 peter = high church) and again chose compartments(rooms) based on major system pegs sub words (12=deans office, 2=inn, 3=Mother Superior[ma]). Three questions: do you teleport through the different locations: Does the deans office(12) have a portal to the team meeting room (13).?
Can you be a little more specific about how “the book was the background”
Do you generate any virtual palaces in your head or otherwise to match the substitute words, or do you find you know enough places to match the books?

1 Like

Good questions, @Mountainmystic. Here’s a few answers and I’ll work up a model for better clarity.

The book forms the general background of the grouped chapter backgrounds (loci if you use the MoL) unless there aren’t enough chapters. But the keys in the background are the chapters. The traversal system (teleportation) is based on a story between the chapter keys which follow numeric order. In the last half of my admin area background story, I know the Dean is the first person I see, who then is leading the team meeting, who gets angry and knocks over his water (14) and people scramble to get paper towels (15) to wipe it up.

There’s many chapters in Matthew because so applying Miller’s law for chunking, I created several backgrounds in Matthew’s palace. If the book doesn’t seem to warrant a background because of seven or so chapters, then the book image itself is the background.

I pre-planned all the palace backgrounds and object keys with as much relevancy as I could using a bottom-up construction method, and create a story for each verse number and verse word sequence. There’s more than enough room for all the verse image objects.

Nobody was really interested in my graphic models but I find them extremely useful in understanding what my mind is doing. I’ll post the model here for my System 7711 of Bible memorization and see if anyone wants to figure it out.

I hope this clarified how I built these memory palaces. It was an incremental project over three years that reminds me of the house that Jack built.


I know no one voted for modeling of systems but take a look at this model of my System 7711 for Bible verse memorization and see if it changes your mind. I use standard software modeling symbols as much as possible. Examples are off to the right.

  • Rectangles are data types,
  • ellipses are systems,
  • solid diamonds indicate composition (thing is a part of diamond end thing)

I’ll be walking through the process of memorization using the model in the topic System 7711 Bible memorization with whoever wants to try it.


I see the source of my confusion now! Wonderful graphic, I would have asked for it first had I known it was so clear. I think most if not all of my confusion revolves around the word “background”.
I typically ascribe the term palace or journey to the set of loci, where each loci hold a single digit number of stored items(if my apartment was a palace, my bed which holds 3 shapes on the bed stand and 2 on the bed posts would be a loci). Backgrounds entered in between these here when I wasn’t sure where they were before.
Palace name-background(locus)-palace key/object-verse content…is the convent the palace and the background the inn, then say a table in the inn a loci?

I think a walkthrough may clarify this and some other questions I’m withholding. Maybe I’m just catching on a little slowly…could I bother you to walk me through a brief portion? I’m dying to see it in action!


I’d be interested in your response when you have read Songlines. I see songlines as multilayered with an integrated knowledge system - all sorts of genres of knowledge intermeshed through highly emotive stories - incorporating song, dance, narrative and art - and a really emotional link to Country. They work much like a memory palace, but far more complex than any other examples I have seen. They are on a massively large scale - criss-crossing the entire continent. Then again, an individual’s knowledge associated with the songline is built up over an entire lifetime!

I am involved with a multidisciplinary large scale study at the University of Melbourne examining not only the effectiveness of songlines / memory palaces / method of loci in education, but also asking the very questions raised here - are they all the same thing? The honours student I am co-supervising is doing a detailed literature review for Honours and then the team hopefully will go on to a large-scale study with university students using various techniques as she goes onto higher degrees. Long term, I am afraid! There are academics from educational psychology and other disciplines involved as well as our Indigenous colleagues. Having listened to the debate in our meetings, I am not willing to commit to ideas on the difference (if any) any more - I need to follow through the research. Such good fun!



@Mountainmystic, confusion was my friend as I was sorting out how other people used palaces. I finally had to apply the system and listen to my intuition to arrive at a result that seemed to be in line with the classical use of them.

The structure on the diagram would map your examples like so:

  • 2 Peter = Convent = palace name and background (locus) (only three chapters)
  • the inn at the convent = 2 Peter Chapter 2 (a traversable object, not a locus)
  • images around the inn = verse numbers converted via Major system to images associated with positions on the traversable object traversed by a story or a journey.

Feel free to ask other questions with other examples to see if it is catching on with you. I’ll be updating more information soon.


@LynneKelly, I’d be honored to give you feedback on Songlines. The book will be arriving next week but I got a Kindle version so I could dig in sooner. I’m doing my own “literature review” and have piled up a few of the great references I’ve seen discussed here by @Josh and others. Moonwalking with Einstein is about 2/3 done and I couldn’t put down Ultralearning today seeing how much good psychology research is discussed when incorporating mnemonic systems into pedagogy and learning.

It’s a shame I couldn’t have been born in Australia. It seems like so many Native American memory systems might have been disrupted by the forcible relocation of the tribes. I’ll be looking for information on them in the future.


Im certainly no expert. I understand your terms now well enough to process your system and ideas

I realized you already defined your terms earlier in the post. So sorry to make you repeat yourself, I should have started f om the top again before asking!

I dont have any more questions, I’ll just be tuning in to your following posts on the topic(especialky your walkthrough)


@Mountainmystic, I updated my diagram a little. It wasn’t really showing the multiple ways I get around in the system.

1 Like

I’m slowly building a better understanding of the system as we go, though not being a software guy has me at a disadvantage. I think the updated graphic helped.

The example you created in the other thread (I linked it already above) had 2 peter (2=high peter=church) chapter 3 (3=Ma=mother superior in a scene from {chapter 2} the inn inside the church)
Verse 9 (9=boy= a boy is running to mother superior)
The story of the text using “lord” “keep” “promise” (the landowner is chasing the boy, catches him at a keep, the boy crosses his heart)(all this happens in front of mother superior)

Because chapters aren’t limited to locations, and the locations aren’t limited to an existing space, your method seems phenomenally flexible and intuitive in its use.

If it’s possible, I still would like to see an existing journey you have stored(actually use) for a few consecutive verses with a note on the set of chapter pegs used in the background and the book image. I’m curious how your text interact with your chapter peg, and how the chapters form the scene.

Thanks for the heavy lifting and constant responding. I’m doing my best to follow and I’m super grateful for what you’re doing, keep up the great work!


@Mountainmystic, yes, the system allows for the maximum amount of personal customization because the more personal it is, the better we remember. I have a 2 Peter palace built around combined Austrian and Kansas City area monasteries I’ve been at depending on which object I’m working with. I sketch out the outline and encourage people to enhance the image like it’s a paint-by-numbers template.

I do have over a hundred selections, many with multiple verses. Some are in transition between a previous system and this one but most are usable. You have to let those memory ghosts disappear so they don’t interfere with any new stories. :ghost:

That will be the next two steps I do in the System 7711 Bible memorization process. @LaurianH wanted to do Genesis 1 and 2 which I’m working on since I had just completed the NT and was practicing on it. But I’ll post the process with an example for you there.


On a related note, what about decoding systems? If you are memorising, for example, texts, then you will come across, on revisiting your memory palace, a variety of images. For some of them, your natural memory will have faded. Is there a way of distinguishing what sort of relationship the image has with the text? For example, if you come across someone chopping a cherry tree, and you forgot what it signifies, how could you reconstruct “Washington”?

I guess some of it would come from context and layout e.g. if you are remembering a list of facts about Washington state, and the image is in a room at a place where you put titles, it is easy to reconstruct. But what about lists of unconnected words?

I’ve used memory techniques for non-competitive purposes, so I value stability rather than speed. What I try to do is build several layers of redundancy. Like “metaphysics” is encoded to “met a physicist”, so “meeting Einstein”, but in case you forget, and stumble, also a layer of redundancy can be added, by imagining a conversation with Einstein about the “fundamental nature of reality”.


Encoding digits

Most of these have been well covered in the Wiki under How To Memorize Numbers but I categorized them slightly differently.

One-step systems

These are systems which are a direct one-to-one mapping. This is a disadvantage though because it is difficult to retain the images for numbers much over 20 in my experience. My preference is for a two-step system of pre-encoding with the Major system and then sound-to-word encoding.

By soundalike

See the Number Rhyme system in the Wiki.

digit object
0 hero
1 bun, sun, gun
2 shoe, gnu, zoo
3 tree, sea
4 door
5 hive
6 sticks
7 heaven
8 bait
9 wine, line, mine

By object

Also known as the shape and Shaper system. See the Number Shape system and the Shaper System in the Wiki.

digit object reasoning
0 ball round
1 magic wand straight
2 coat hanger hook on top with straight pant holder
3 butterfly two rounded wings
4 chair person with crossed legs?
5 unicycle rounded bottom with seat on top
6 golf club club on bottom with handle on top
7 boomerang angled
8 snowman two round balls
9 balloon rounded top with string on bottom

By association

See the Association System for Numbers in the Wiki for two-digit mappings.

digit object reasoning
0 ice on thermometer freezing point of water
1 cross, tree count of vertical items
2 twins, jeans count of items
3 dinner triangle, stool sides or pieces
4 compass directions, car tires count of items
5 fingers on hand, glove count of items
6 Star of David, die count of items
7 dice lucky number for games
8 8-track tape, skates part of name, shape from use
9 pool balls, solar system part of name of game, count of items
10 bowling pins count of items

Digit to sound pre-encoding

These systems are the first step in a two-step process converting the digits to sounds which are then encoded to words, people’s names, objects, or locations.

Major system

See Major System in the Wiki.

Major+3 system

This is my addition of three aspirated letters (H, W, Y) to 11/J, 12/Q, and 13/K for use in playing card memorization. 10 is mapped to zero. Ace is mapped to one. Has anyone tried this variation?

Ben system - vowels

@Zoomy (Ben Pridmore) uses this in combination with two of his consonant mappings (below). See Ben System in the Wiki.

# sound # sound
1 short a – cat 6 long A – hay
2 short e – pet 7 long E – bee
3 short i – kitten 8 long I – high
4 short o - tom 9 long O – low
5 short u - puss 0 long U – you

Ben system - consonants

This is a 16-digit/number to letter conversion which provides much more flexibility for different data types but correspondingly more difficult to learn. The system is mostly a digit to letter mapping with a few sounds which improves the efficiency of encoding and decoding. But there are also less secondary encodings of words that can be made.

# sound # sound
0 s 8 f
1 t 9 b
2 n 10 p
3 m 11 d
4 r 12 h
5 l 13 sk, sn, sm
6 g, j, ch 14 st, sp
7 k 15 sh, sl, sw

Dominic system

Dominic O’Brien pre-encodes digits to letters, and is not a true digits to sounds pre-encoding, so is more of a competition system and then finishes with Person-Action pegs. See Dominic system in the Wiki.

digit letter reasoning
0 O looks like zero, last letter of zero
1 A first letter of the alphabet
2 B second letter of the alphabet
3 C third letter of the alphabet
4 D fourth letter of the alphabet
5 E fifth letter of the alphabet
6 S first letter of six
7 G gee, I have no clue (DH)
8 H almost the equivalent of a long A
9 N first letter of nine

Phone system

A system lacking usefulness for encoding.
2=A,B,C; 3=D,E,F; 4=G,H,I; 5=J,K,L; 6=M,N,O; 7=P,Q,R,S; 8=T,U,V; 9=W,X,Y,Z

Numerology system

Another useless system for encoding.
1=A,J,S; 2=B,K,T; 3=C,L,U; 4=D,M,V; 5=E,N,W; 6=F,O,X; 7=G,P,Y; 8=H,Q,Z; 9=I,R.


@Niten, you raise some good questions. I’m not a memory athlete myself. I grew up on a farm without brothers so sports was not my thing. I could barely handle bowling.

The encoded images should be unique so they can be distinguished. If they can’t then you need to change the context or the object itself. Exaggeration or providing a different set of enhancements changes the object. But maybe the object might try to slip back to mediocrity. In verbatim texts, the background (locus) provides the scenery to the objects and activities of my journeys and should be well distinguished from any other background.

So you are right about how I would go about encoding information about Washington state, however it might end up as a room (a memory object) within a locus of the Northwest states. I’d have to have mapped out the info that I wanted to retain and created a traversal scheme for every room based on that information.

Lists of unconnected words pose a problem. They have no natural order as in verbatim texts. So they must be associated with an ordered system of pegs. The next large post I do will review the peg systems.

I’ve found that using redundant peg systems, which I call doubling-up, is a great way to provide assistance when reaching for that next part in traversing my journey with just a story. I double-up with number-pegs on every paragraph of Bible verses or famous verse which helps me get back on track as I traverse to the next used number-peg.

It seems to me that your style of redundancy is what I would call enhancing the image and is a category on the poll I took at the beginning of this topic that I’ll address in the future. It’s often just a sketch in my mind that I create at first for a story and not being that memorable; it fades easily. The big enhancements are personal images and relevant or strong actions. Dance and music are also used. The big techniques are writing down your story and telling your story orally to another person (if you can find another person who would tolerate it. My wife made it through only one.)

I hope this helps you see from my perspective how I do my memorization. Maybe other folks have some other ideas. There’s quite a large range of opinions here which is why I like participating. I look forward to hearing more ideas and questions from your background and sharing here on the site!

Happy encoding, Doug


I think I’m comfortable in sharing the analysis of memory systems in general now. I had to use the various systems, learn from everyone, and test everything before I made decisions. The three big variables that define a memory system, it seems, are how encoded images

  • are grouped
  • create their own or use an existing system
  • are traversed

Here’s the basic visualization system categories:

Type single association narrative sequence rule-based sequence
Words / digits Simple association Story Pegs
Object Symbol Memory object (palace) Journey

and here are the system construction methods:

Top-down, key first Bottom-up, value first
Simple association Key-value pair Tradition driven systems suggesting links.
Story Pre-written narrative Keys become values to link to next key suggesting narrative.
Pegs Peg system A personal implied order of items.
Symbol Shelving, pigeonholes. Aggregate object, souvenir cabinet.
Memory object (palace) A selected background, familiar locations, photo books, an unordered method of loci. A composite memory image where chunked values suggest a background or a scene of a play.
Journey Locations having a common and distinct background(s) for encoded images or other systems. Winter counts, songlines, an ordered method of loci. Associated values suggest an imagined background or scenery for multiple encoding types in each scene. Layering of types. Lukasa. Free-form art forms – plays, movies.

Rules which can be applied to the ordered groups are one or more of the following (not complete, I’m sure):

  • pre-existing path
  • sequence (alphabetic, numeric, etc.)
  • peg system (imposed sequence)
  • path by proximity, alignment, contrast, or repetition of markers
  • small to large
  • front to back
  • left to right
  • top to bottom
  • external POV outside to inside
  • internal POV low (starting near your feet) to high
  • external POV high (starting near your head) to low
  • internal POV: inside to outside
  • clockwise (north, east, south, to west or 12, 1, 2, etc.)
  • 6-sided die - turn right, rotate forward
  • pitch

I’ll continue to discuss the systems in these terms in my posts. I welcome any discussion on my organization of the memory systems.

Edit: I’ve made a major revision to this summary at post 58 in this topic.


After updating the above summary another thousand times, I considered that it wasn’t quite ready after all and now post the current status of the overview which has much simpler names and consistent descriptions based on sentence structure similarities. If you find it easy then my difficulty has been worth it. In analysis, simple is never easy. Suggestions are always welcome. @LynneKelly has provided much input via her Songlines book.

Memory systems

Visualized by single association narrative traversed sequence rule traversed sequence
Words Sentence Story List
Things Symbol History Journey

Creation methods

Type Key first (using and teaching) Value first (learning)
Sentence Logical sentence. Subject performs relevant action with relevant object. Nonsense sentence. Any subject acting in any way with an object.
Story Interpretation. Imaginary subjects and actions perform narratives on objects to achieve an outcome. Improvisation. Subjects and objects follow narrative logic with relevant actions without knowing an outcome.
List Pegs. An ordered list of subjects perform relevant actions to a sequence of objects. Pattern. Seen as a whole, a group of objects suggest an order for imaginary subjects or actions of a sequence.
Symbol Feature. A natural or created part of a background is associated in a relevant manner with an object. Monument, memento, icon, tag. Method of loci without a traversal rule. Projection. An object acts on a natural or created part of a background for an imaginary reason.
History Adventure. An existing background controls the narrative logic to the actions and objects without knowing the outcome. You are standing in an open field west of a white house, with a boarded front door. There is a small mailbox here. What does the mailbox do? Dreaming. A background is the result of a narrative of locations with imaginary actions and subjects to achieve an outcome. People or gods did stuff here to this object, who and what was it? Songlines.
Journey Marked path. An ordered set of locations within a real background are associated with actions and objects. An ordered method of loci. Art form. An ordered set of locations using an imaginary background develops out of imaginary subjects doing actions on a group of objects. Plays, movies, dance. Lukasa, winter counts.

The difference between the two styles of visualization systems in the bottom chart are distinguished when the system is created. If the knowledge that is being stored is known beforehand, then it is what I used to call a top-down system or a key system based on using a memory system as keys to provide storage locations for the knowledge. People impose a rational order on top to retain for use and to teach others about the knowledge. It’s more of a scientific approach.

Any kind of system for using and teaching information by location is essentially a method of loci. The feature has no traversal system and items are plopped into a room as features to be seen, the adventure has a narrative approach to walking around in your memory palace finding objects as you go, and the marked path tells you by rule what location to go to next.

The other style of visualization system is the bottom-up style where information is stored as it is acquired and blended in to a system. When too many pieces of information or values are needing to be stored, they have to be understood and chunked using the imagination to provide a way to organize it. It is a learning process as new pieces of information are blended in and the ordering can change. You could call this the analytic approach.


I received my copy of Margo Neale and @LynneKelly’s Songlines recently and thoroughly enjoyed discovering how knowledge has come to be encoded and visualized in her part of the world by the indigenous Australians. As an analyst, I wanted to see if I could visualize the memory system known as the songline. Here’s my first draft of the model:

All of my information is from one reading of the book, so if I don’t have all the details, any other person including Lynne, is welcome to suggest edits.

The sites of significance exist on trails and sometimes will appear also in the constellations. The sites became descriptive names or sometimes creatures made through associations to add to a rich composite memory image at that site. Simple visualizations follow a sentence structure usually. The imaginative visualizations are what I think are called the Dreamings. The composite memory images also contain the visualized imagery that has been encoded from plants, animals, migration times, agricultural methods and more. The diamond refers to the relationship of visualized images being aggregated into the composite in a way that doesn’t depend on the composite to understand it.

As a custodian travels along the paths of the songline traversing the specific locations, composite memory images come to mind. The dreaming is a narrative-based sequence of objects recalling a history of significant events and objects, in this case, locations followed in a particular order along a path that existed before the songline was created. This is a learning type of recording of history drawn from the imagery of the Country. Today, we generally bring information to associate to locations which is more of the adventure purpose used to teach others or use for ourselves. When you bring knowledge about history in a narrative sequence, as Lynne does, to her path around the house and into town, that is her adventure. If the information was more disconnected along a path it would be a marked path style.

The filled diamond is a symbol of composition which means that the composite memory images can be traversed as a dreaming by travelling along the locations in order. But the composite memory images can also be traversed as a legend which puts the words on the images so they can be told in stories, sung as a song for easier recall, and performed as a dance in a ceremony. Ceremonial objects. the art form style of journey, also have locations on them, traversed by a rule, which represent information and are included in the legend traversal when the performer enacts a songline.

The benefit of having redundant traversal methods is that songs can be sung as the sites are being visited. I liked a part of the book where a song told two local Aborigine people of a waterhole obscured by an escarpment while driving through Country even though they had never visited the area.