Inverted memory palace

General Memory Chat
1

I tested this idea I had some days ago and it works. That doesn´t automatically mean it´s useful for something, but I have reasons to believe it may have some application. I tested it with numbers and I´ll give the examples with numbers, but some other kind of information could be codified the same way.

In fact this is not something original at all, it´s based in the way I learned the mental calendar from Dominic´s book “how to develop a perfect memory” and someone else may have already though about it, but dominic didn´t give a special name to the technique and I think it´s clearly something different than the normal application of the loci method. I think it deserves more attention and a better examination of it´s possible variations, since I never found anything more about it and it was a very effective way to learn the mental calendar.

This is the idea exactly how I recently tested it: when doing number memorization, instead of using the loci to remember the place and the peg system to remember the number, using the loci to remember the number and the peg to remember in which place it is. This means you need a well numbered memory palace of 10, 100 or 1.000 loci depending on how many digits you want to remember with each image, and an ordered list of items (not necessarily numbered. but you can perfectly use the major sytem) to place on it, long enough to cover the quantity of numbers you want to memorize. Let´s see an example:

You want to memorice the number 103289

Let´s suppose your first 3 pegs are a cow, a monkey and a superman, in that order.

Let´s suppose that in your numbered palace, number 10 is your kitchen, number 32 is your local park and number 89 is your office.

Then you proceed to place the cow in the kitchen, the monkey in your local park and the superman in your office. And with that you memoriced the number. To retrieve the number from your memory, instead of asking yourself “what did I put in my first loci” you ask yourself “where did I put my first peg (the cow)” and you will see it´s in the kitchen, the number 10 loci.

Is there a reason to invert things this way? Some disadvantages are clear: you are going through your palace in a weird and disordered way, kind of teleporting your imagination to place each object in the corresponding loci instead of just walking, running or flying through the palace. And having a well numbered palace is hard (I mean, knowing in an instant which one is loci number X). But there are a few advantages, and some differences that could turn out to be advantages in a given context.

The way you have to think when memoricing a number is different; usually you already know which loci you will use and you don´t know what you will put in it until you read the number; with this method you already know which objects you are going to place and you don´t know where until you read the number. You are looking where to put it instead of looking what to put there when memorizing, and you are looking where did you put X instead of looking what did you put in that loci when recalling.

What happens if a number repeats? in fact it´s much better if this happens and I think this is the main advantage of the technique. You can link the peg with the loci AND the other pegs already in there, forming a more complex and unique escene that will make more connections for all the pegs in there. In the case of long number memorization this would mean taking advantage of the repetition of a pair of digits. If you happen to place 20 pegs in a single loci with the result of a crazy scene, and you ask yourself what did you put in that loci, maybe you won´t remember all of them (they could be linked with each other in a chaotic way), but if you ask yourself where did you put your peg number X you will know it´s there, and that´s all you need in this case. To learn the mental calendar the way dominic teaches it, you need to memorize a number from 0 to 6 for each number from 0 to 99, and you do it with a 7 loci numbered palace, placing each character of your PA list in the corresponding room interacting with all the others on it. I was surprised of how well it worked when I learned it!.

There is another difference. Usually, if you memorize a number, and you want to memorize another number, you need more loci. With this method, what you need is more pegs; you can reuse the same loci and take advantage of the previously created scenes when you have to place a new peg there. The fact that pegs don´t need to be numbered (but they do need to be ordered) can be an advantage to create enough pegs.

With the usual loci method, when you forget a number you have to check all your pegs to see if you can recall which of them was in that loci. In this version, if you forgot where did you put the cow, you run in you palace looking for the cow. I think this is a bit faster.

I think this works better if each peg has some unique way to interact with the environment while it´s alone in a loci. In fact, all the peg selection could be done to work better with this method.

For competitive number memorization, since I didn´t test the technique enough, I don´t know how much can be can improved with practice. Right now I´m far slower than with my usual PA system, but I lose most of the time because of not having a really well numbered palace. The fact that you know beforehand exactly which objects you will use and in which order could mean faster times, but without real experimantation I can´t know if that´s enough to overcome the disadvantages. For longer disciplines you could get a good advantage from the repetition of the numbers, if using a memory palace of 100 loci.

As mentioned at the top, i´m talking about a numbered palace for number memorization, but instead of a number each loci could codify some other kind of information.

Any opinion, idea or advice is welcome. Maybe someone already tested this and can talk from experience much more than me, or maybe this technique is well documented somewhere and I missed it. Thanks!

4 Likes
Best Posts of Mnemotechnics
The First 4-Card System
2

What if 89 came first? How would you know? Especially with memorizing long numbers, you wouldn’t know the exact order of the loci.

It’s an interesting idea, there have been posts on it before. Using places to represent numbers. An extreme version of this would be to have multiple palaces of 100 loci, then every second number store the images in whatever locus fits.

The advantage to just using the memory palace the original way is you always know where to go, you always know where to progress, and ultimately that’s faster than having to think of where the next locus is. It’s also easier to change backgrounds when the loci are very close, as opposed to a random distance away.

On a calendar, the order you memorized the images in doesn’t really matter, what matters is where they are placed. When memorizing numbers and such, the order does matter, and unless you mark which one you memorized first, second, third, etc, there is no way to tell for longer disciplines. You could combine it(as some already do, using normal memory palaces) with linking, as in linking 3 objects per locus and then linking the last object in that locus with the first object in the next locus after that. Then you could follow the chain, provided it doesn’t break.

Bateman

1 Like
3

Yes, you know exactly which one came first. The example I gave was a bad one because the numbers happened to be ordered. During the recall process, you don´t go trought the palace in the order of the palace, you also go thought your pegs in order and find each one. I memoriced much longer and disordered numbers without problem. If instead of having 103289 you had 891032, (your first 3 pegs being a cow a monkey and a superman in that order) you place the cow in loci 89, the monkey in loci 10 and the superman in loci 32. When recalling your first look for the cow, then for the monkey and then for the superman, there is no possible confusion in the order.

4

Oh, my bad. I misread, I thought the cow, monkey and superman actually represented numbers.

Bateman

5

Your system is very interesting, but I think not suitable for competitions, because it’s slower than ordinary locus->number method.

You are true that it’s easier is to move in palace from a locus to the next locus (and this way scanning them all through), rather than going through the pegs in order. But that’s only going to speed you up when you forget where a peg was. The big disadvantage is that most of the time now you have to go through your peg list - when memorizing number and when recalling it. And teleporting in the numbered palace will also slow you down.

Each time you teleport into the locus you already used before you have to recall the last peg you placed there. Another thing that will slow you down. Maybe just place the next peg there without thinking about the previous ones?

That’s the most interesting aspect of your system: Is it easier to answer to the question :“Where did I saw this image” rather than “What image did I see here?”. Our braing geotags all our daily events, it’s not difficult to answer to the question “Where the event took place?” (because an event cannot happen twice in different places).
I have pondered over whether the same applies in memory palaces. Deducing the background that this peg had might be a bit harder, because it’s not so specific than the image itself (locus isn’t so noticeable). In case of locus you know exactly spot where you place the image, but if you want to recall the locus from peg you don’t know in which part of the screen you focus your attention (the background is all around the peg). But if you link the locus with peg then the recall will be easier and in that case I don’t know which way the recall (locus-> peg or peg->locus) will be faster. Depends on the link.

Creating new peg lists is harder than creating new palace.

Still very interesting ideas! Thanks for racking your brains over it and sharing!

6

I´m 99% sure that you are right. For all the reasons we mentioned, I don´t think this can be fast enough for a competition, and when I test it it´s clearly slower. But I would like to know how much of that speed difference is because of problems in the technique and how much because of lack of practice. Maybe with enough training going throught the list of pegs and teleporting in the numbered palace becomes much faster; if it becomes fast enough to make this speed difference of the transitions less relevant, and this method happens to be faster in some other aspect, you could end up winning time. Again, I don´t think this is the case, and I find this method much more interesting for other purposes.

Yes. I was pointing to the advantage of recovering a number you forgot much faster (I think you also have better chances of recovering it, but I´m not sure) but the general speed disadvantages of the method are always there. Maybe this speed difference to recover what you forgot allows you to be less careful when generating the images and win time in the memorization process.

This is one more reason to think this technique is not for competition, but can be very good for something else. I think that if you place the peg without thinking in the previous ones you can erase the previous ones. But when you think of the previous pegs placed there and keep linking them with the new ones in any way, you are refreshing what you previously memorized and making new connections that make them much harder to forget. In a speed competition this is most likely overlearning and a waste of time; in another context it´s something good.

Well, when I have only one peg in a loci, I can forget it if I don´t link it properly with that loci because deducing the background is hard; that´s a mistake I often made in my first tries with this method. And, in fact, I didn´t totally forget it, I could usually recover it going from the loci to the element. But I think the best aplications of this technique are about placing many elements in the same loci, linked with each other and the loci as much as possible; and in my experience deducing the background becomes easier with each added element, because the scene itself ends up kind of merging with the background, and it becomes highly noticiable.

That´s true. But the fact that creating new palace is easier could also mean that you can create a very big numbered palace to remember more digits with a single image (if you happen to have a numbered palace of 10.000 loci, you can remember 4 digits per image with this method).

Thanks to you for your comments!.

These are some applications I was thinking about:

  • If for any other reason you want to have a numbered palace, memorizing a long number or a list of numbers with this method is a good way to train with it, since you have to find the loci from the number and then the number from the loci. And you train with some peg system at the same time.

  • This one is the main reason I got interested in this method. When learning the mental calendar from dominic, I was amazed by the fact that I permanently learned a total of 112 things (the code for the years from 0-99 and the code for the 12 months) using only one palace with 7 loci; and I could have learned even more if needed. And the same peg system (my PA in this case) can still be used in any other palace!. I think this technique is very powerful to store huge ammounts of information in few locis, and that can be very important to learn things permanently without having to create a lot of loci for the information. For example, if you have 1.000 pegs, you could learn 1.000 digits of pi in a palace with only 10 numbered loci (with around 100 pegs on each one, forming a massive escene) and then be able to awnser in an instant which digit is in X position. If you want to learn more digits, you can add 10 more loci for each 1000 digits you want to learn. This is something I´ll actually try as soon as I complete my level 3 system, to see if it really works as I think it should. Since you go one digit at a time the memorization is slow, but there is no hurry.

7

And from here we get another mind thrilling question: Do we remember better What is the next peg in your link chain? or In what chain the peg was? Your system is just overwhelmed with its unique reverseness :slight_smile:

By the way I think In what chain the peg was? is easier to anwser, because we need to recall what was the peg before OR after it - recalling just one of them is enough (if you forget one you can still use the other). In case of What is the next peg in your link chain? you have to recall linearly. It seems that in your system (when using linking) forgetting the next digit is less probable than in ordinary locus->digit->digit->digit or digit->digit->digit systems. Right now it seems that the system is more foolproof but slow.

8

I just had an idea: When creating new numbered palace and new peg list then you place each peg to its corresponding locus (peg 0 to locus 0, peg 1 to locus 1,…peg 99 to locus 99). Now you can

  1. memorize the new peg list with less effort and also go through it much faster (you move in the palace)
  2. its easier to recall number X locus
  3. you do recalling the pegs and memorizing numbers in the same palace (don’t have to jump between palaces)

When memorizing number you take peg N from current peg-locus-palace, study the number X you want to memorize, teleport to number X peg-locus, put peg N there and link with the last peg you placed there. Then continue with peg-locus N+1 and number Y.
Now you can recall peg N by thinking: Where did I see peg N?, To what locus did I jump from locus N? and In which chain did I see peg N?

The only possible disadvantage that I can think is if the first peg you place into locus X is peg X, and you forget it because it’s already (naturally) there. But maybe it’s the opposite and will stick to your memory (because it doesn’t happen often).

Your system is getting more and more exiting, Nerto! :slight_smile:

9

Also see this for an alternative way to use memory palaces.

Bateman

10

Thanks r30 for the ideas! I was going to awnser this yesterday but I couldn´t load the forum

At this point I think we must make an important distinction: it´s not the same using this method to form long-term memories than using it for fast memorizing, the approach must be different according to the objective. More links can improve retention but cost speed, and some methods to make the memorization faster may be unnecesary if you don´t mind too much about that; and there are other differences I will try to cover in this post. Also, we already know this method is useful to form long-term memories (the calendar is a clear example) but we still don´t know if with enough training and improvements it can become fast enough to compete.

About forming long-term memories, there is something very important I want to clarify, because I think it´s one of the main reasons to use the inverted palace:

I think you are right about that. But I usually do something even slower and even more foolproof than a chain (and this is how dominic tell´s you to memorize the calendar). I wouldn´t call what I do a chain, it´s more like a net, or a big scene. We could call it chaotic linking or free linking. It´s a single situation in which all the pegs can interact with each other (but without forcing all of them to interact with all the others). There is no need to link each peg exactly with the previous one and the next one, you don´t even have to link it with exactly two, you can make as many links as you want between all the pegs in the same loci, because you don´t have to link them in order (the order the pegs already have between them and the number of the loci gives you all the order you need). And this freedom is good; the more links you make for a peg the better chances you have to remember in which net it was.

I usually set the main scene with the first pegs I put in the loci and when I need to add a new peg I see how it could fit there, making sure that it interacts with AT LEAST one other peg or has an important interaction with the loci itself; and when I finish I like to have at least two links for each peg, and try to make sure that no interaction between pegs is too isolated from the rest. The result is usually hard to forget: you have each peg with multiple linkings in the context of a massive unique escene going on in that loci. And if you find some peg hard to remember, you can always link it with more pegs. Characters are specially good for this, because they can interact in so many ways or comment about what they are looking.

The wonderful thing is that, as far as I know, there is no limit to how many objects you can place in a single loci this way. I tried with around 20 per loci and it works, but I plan to do it with much more and share my experience.

In a speed competition, using a simple chain instead of this “net” could be more effective because it´s more systematic and you make sure each element has two linkings with pegs (and it´s also placed in the loci itself if you make the links there) without having to think about that. That´s one of the main differences I wanted to point out between fast memorizing and long-term memorizing.

Very nice! This is a good way to overcome the slow movement between pegs and the slow teleport in the loci at the same time. I think this is far better for fast memorization than my initial method. I see a few limitations you didn´t mention, but nothing impossible to solve:

  • If you use the same palace for both things, the number of digits you can memorize is limited by the ammount of numbered loci you can make. For example, if you have an enumerated palace of 100 loci with 100 pegs asociated with it, you can only memorize up to 200 digits (you remember two digits per peg, and you have 100). One way to overcome this would be to have more than 100 locis for the pegs, even if you only use 100 enumerated locis when memorizing the number, being always able to memorize “number of pegs x 2” digits. Another way to overcome this would be to use 1.000 enumerated locis with 1.000 pegs, pushing the limit to 3.000 digits and making it equivalent to a 3rd level system at the same time (3 digits per peg, and you have 1000). Yet another way to do it, memorizing less digits per peg but becoming virtually unlimited in capacity, would be having more than one palace to use the same pegs again (if you have 1.000 pegs, and 10 palaces with 100 enumerated locis each, you can memorize 2.000 digits on each of them for a total of 20.000. But if you use linear linking between pegs you could get confused here).

  • One of the possibilities of the inverted method (in theory) would be that if someone could create a huge palace of 10.000 or even 100.000 enumerated loci (don´t ask me how!, I don´t even know if that´s possible), he would be able to memorize 4 or 5 digits per image, something very hard to do with the conventional methods (I only know one case of someone with 10.000 pegs). But if you want to have each loci asociated with a peg, then you are again limited by the number of pegs you can come up with and can´t take so much advantage of the fact that making more loci is easier than making more pegs. Someone with such a huge number of loci maybe should use the same pegs many times (to make it easier to get to the exact numbered loci) but then focus more on the loci when memorizing than in linking with the peg placed there.

I think that even with your improvement the transitions are still slower than with the usual loci; you still have to do some teleports. But I see two possible advantages to give it some competitive hope:

  • If the method is foolproof enough to skip revision, you win a lot of time there
  • If someone can make a huge enumerated loci and use it fluently, he can memorize more digits per image

So, the questions that should be awnsered to know if this method can compete with normal loci would be:

a) Is it foolproof enough to skip revisions? (or at least reduce them a lot)
b) Is it possible to make a numbered palace of 10.000 and use it fluently? and what about 100.000?

If one of the awnsers is yes, then, in theory, it COULD be faster than normal loci; If not, then I can´t find a reason to think the method is likely to compete against the normal loci.

There is still one more thing to take in consideration about the inverted palace for fast memorizing: the idea is not explored enough to think we already came up with the best possible version of it. Few days after I posted about it you already made it faster.

Thanks :slight_smile: , but I took the idea from a book, it´s too much credit to call it MY system. I´m just exploring more areas of application for a technique already known.

I still don´t understand clearly what´s the difference with the usual loci, but i´ll keep reading. Thanks!

11

hi! long time without entering here. I had forgotten my password, the irony.

I just want to say that with 10 locis and 300 pegs I was able to memorize 300 digits of pi with this in a way that is very effective to then awnser “which is the xth digit of pi” instantly.

12

Hi,

this inverted memory palace is basically the same as the reverse memory palace I have made several posts about. To begin with, the idea is at least 5 years old, but I don’t know who came up with it originally. Some of my experiences:

  • teleporting is not the problem at all; it’s just as fast to translate a number into a location as into an object (like in the normal system);
  • recall seems easier (see: Asymmetrical memory connection);
  • biggest problem is learning a peg list in such a way, that you can visualise it really fast in the right order when you do the actual memorisation. I never succeeded in doing this, so I abondoned the system.
  • I tried an alternative in which there is an object in every location; you memorise a number sequence by placing a starting object (this object is always outside the palace at the beginning) at the first location determined by the first 2 digits and than you take the object from that location and you place it at the next location (determined by the next 2 digits). In order to retrieve the whole number sequence you basically rewind the movie: first object went to … and the object from that location went to … ? This system works better because you don’t have to keep track of a peg list. I’m not sure if it can compete with the normal system.

For what it’s worth: as a result from a recent discussion I have become fascinated with the possibility of a single digit number system with just numbers as they are see (Memory hooks and why they matter).

Anyway, I’m interested to know how well the inverted system will work for you
.

1 Like
13

Hi!

I never really put much effort in making this method able to compete in speed with the normal loci, it just doesn´t seem plausible and maybe we should use this method for whatever it´s good instead of trying to make it work for what it´s bad. So I just thought more about using it for something else, like long term memory, recall speed of some kind, some variety in or tools, or just fun, and not memorizing speed for competition.

But, if one decides to try using it fast anyway, about the problem of having to visualize the pegs very fast, I think one could test the system with some visual help to see how fast it could potentially work. For example, I could have the string of images at my sight bellow the number i´m memorising. That would be cheating, but if at some point I can stop looking at the images and recall that process (or i´m not in a tournament) it´s not cheating anymore. If even doing something like this it´s still slow, then I thinkk it would be hard to make it go fast.

But, as I said, I didn´t put effort in making it fast for memorizing and replacing the normal loci, I´m more interested on it´s strenghts. I´m not sure if I was correct when I said that this is the same as what dominic does for the mental calendar… I just thought about this when I noticed that you could learn the mental calendar in a much less efficient for recalling and boring way if instead of placing 112 characters in 7 loci you use 100 loci to remember the code of each year.

I wasn´t that much into mnemonics for the last years, but I did start to make a third level peg system, when I had up to 300 I used this method with success to learn the first 300 digits of pi with 10 loci, in a way that, when asked for any number from 1 to 300, I can inmediatly awnswer which digit it is. I didn´t even think about doing it fast, I took my time, I even made kind of lot of little stories with the pegs that end up together in the same loci to improve long term memory (I think this won´t work very well if you use the method with the same pegs for many different things, but it works well if used once). It does take time, it won´t be the fastest and not the best method too learn a lot of pi either, but´s it´s hella funny and good enought for memory demonstrations. For me it was also a nice way to practice with my pegs and motivated me to have more: the more my third level peg system grew, the more the 10 locis for Pi got filled with stuff and stories. But I never completed my third level system, so maybe i´m not a good example.