Another update of reverse memory palace, animals is the way to go

As I have mentioned in an earlier post, I have switched to a reverse memory palace system. Short summary of (my variation of) this system:

  • a memorypalace with 100 numbered locations;
  • in each location there is an object (some are a person or animal); these objects come from my previous system and they make it easy to identify the number value of the location (the objects act as “training wheels” as it is my goal to remove them from the system when I can directly associate the number with the location);
  • numbers are read using my New memory system for fast translation of numbers and the words/sounds are directly linked to the objects in the locations;
  • a prememorised sequence of elements (objects, persons, animals, or whatever) is placed in the locations determined by the to be memorised numbers;
  • that sequence is finally used to recall those numbers.

I have done some tests using different type of (25) element sequences (so 50 digits):

  • actions:
  • famous people;
  • objects;
  • animals.

I have cheated somewhat in the following way: instead of memorising a sequence (takes to much time for this testing purpose), I simply make a sequence in a spreadsheet column and put a 2 digit number next to each sequence element.

The categories actions and famous people were the least succesful. The category objects was a bit better, but some objects lead to some overthinking (like how do you place a roll of ducktape on a desk in a memorable fashion?)

And the winner by a huge difference in is the category animals. On my first attempt I managed to memorise 50 digits in 50 seconds, despite the fact that this test felt a bit akward as it so different from the usual way of memorising numbers. I have experienced 2 reasons why this category is the winner:

  • all the animals are super fast to visualise and in HD-quality;
  • the imagined animals react to the location in a very natural and memorable way, so no inclination towards overthinking.

To give you a better understanding of the test, here is part of the test;
giraf 19 (= giraf on the toilet)
zebra 37 (= zebra on my office chair in the warehouse)
horse 08 (= horse on my car)
lion 32 (= lion on the coffee machine in office hallway)
cheeta 93 (= cheeta on the kitchen counter)

So in order to finish my system I have to make the 25 animals list twice as big (50 elements for 100 digit memorisation) and I have to put those in the best possible order so the memorisation of the sequence will be easiest.


Would it be a problem if you got the same digits several times? Like 3247329432?


I’m curious, too. If I’m understanding correctly, each repeat of digits sounds like it would mean adding that object to the same location.

But I’m wondering if it’s possible to not need to do that in a combining-objects way, in the same way in real life we can put multiple things into the same box, and just know which box it is in.

What I mean is: When you ask yourself “Where is [object animal 1]?” you get the same answer as for animals 3 and 5, but separately visualised. This isn’t necessarily confusing, any more than when you ask “What is in [location X]?” where multiple values of “X” find the same object each time in a non-reversed MP.

So I’m curious: Has this been tested out?

1 Like

No it would not, because the first “32” would be an elephant (for example) in location 32 (on top of coffee machine in office hallway), the second “32” would be a different animal also in location 32 and likewise with the 3rd “32”

To make it more clear 3247329432:
1st animal in location 32;
2nd animal in location 47;
3rd animal in location 32;
4th animal in location 94;
5th animal in location 32.

In review I don’t look at locations but at the the sequence of animals;
1st animal (let say elephant) goes to location…?
2nd animal (let say hippo) goes to location…?

Having multiple animals going to the same location is no more a problem as having multiple versions of the same item in different locations in the standard memory palace.

Finally a small update:

I have finished my 50 animal collection and I have yet to put them in a easy order. Because of work I don’t have much time to practice, but I intend to make a little bit of progress every day.


You are right.

So I’m curious: Has this been tested out?

In the test I did there were a couple of repeating 2 digit numbers. It did not cause any problems, but I suspect that there is no advantage to this over the normal system in that regard. In other words, the more repeating of 2 digit numbers the more confusing it will be, just like in the normal system. I imagine when you get a lot of repeaters it pays to make some meaningful interaction between all the animals in the same location, but in competition for 100 digit number I don’t expect much problems.


It’s an interesting idea. The main disadvantage seems to be that you need to know your locations extremely well, so you can immediately jump to location 19 or location 37 when you see those numbers. With the scaffolding of index objects and a lot of practice, I’m sure you could get very fast at that, but that’s just for one palace. How many palaces are you planning to have?

I was just thinking that the regular Memory Palace can have multiples of the same object located around it with little confusion (other than doubting it’s the same one again), because it is the location->object association (location “leads to” object) that counts for recall, and each is distinct via the location, so it can support many copies of the object across multiple distinct locations.

The reverse Memory Palace is an object->location association. So instead of needing to combine the animals in the location, this should support many copies of the location across multiple objects, because the association of the object to the location is independent and distinct each time.

Am I making sense?

What I’m saying is: Perhaps there’s no need to combine the animals at the location at all, even if there are many of them associated to the location, because they are distinct associations already. Does that seem possibly true in your tests?

1 Like

I got another idea, I think this method can be enhanced by classifying the animals along with the numbers.

0-10 animals living in rain forest like jaguars, monkeys.
10-20 animals living in cold areas like polar bear, seal.
21-30 animals in grasslands like lion, cheetah
31-40 maybe birds
41-50 reptiles like snake , lizard

I think it would be better and easier to memorise this way.

What do you guys think?

1 Like

an interesting idea. The main disadvantage seems to be that you need to know your locations extremely well, so you can immediately jump to location 19 or location 37 when you see those numbers

I need to know the locations very well, just like in the standard system you need to know your objects very well. I read the numbers as words/sounds and even though the words don’t make much sense in relation to the location, I have come to point (with relatively little training) that I simply know the name of each location. In addition to this I also need to know the sequence of animals very well. This is, admitteldy, a lot more difficult to learn than the sequence of locations in the normal system. So ultimately this system requires a bigger investment.

How many palaces are you planning to have?

Only one, because I’m only interested in being really good in the 100 digit event and later perhaps also 1 deck of cards (I need 52 animals for that, so maybe I add Kermit the frog and miss Piggy to the list).

I have however some ideas about how to extend this system to use it for greater amounts of data. One is to give each animal one or more objects.

1 Like

My list is already complete and it’s in a very intuitive order:
polar bear
grizzly bear
panda bear
koala bear
guinea pig
killer whale
hamerhead shark
tuna fish