Memory hooks and why they matter

I have also made a shape system or rather theoretic over 10 years ago but I didn’t really name it this or even applied it to digits. I think this is the first formal naming of it, I’m sure we all have ideas at some point of time that are named after. For example, when I look at a lot of the neuroscience advancements over the last 10 years, I see at least 20 of my ideas being ‘discovered’ years after I have had them. Realistically though it doesn’t matter too much, if we are really picky on how its named, we can use the name we want while others use another. I used to think the shaper system was called the sharper system. I have just realised you also think this:


I personally never have had this problem, but I imagine this may be also based on how you are placing objects in your palace. When I traverse through mine, I usually go in a very sequential direction so I always know I am processing the images forwards.

The reason I mentioned the system is because the very first time I looked at it, I took around 2 minutes to entirely remember the system. Single 2-digit image recall was also at around 0.4 seconds, whereas when you have an image number association usually the first recall speed is around 1-2 seconds, particularly if you have multiple numbers to learn ‘before you officially train’. What really struck me about this system isn’t so much the system but the fact that you can get to a post training stage in one aspect with this system right away [recall rate]. Something like this is very unusual, so it gave me the consideration that the logic that causes this to function must be applicable on all other stages to some extent.

As for the drumstick and skier, I don’t think it really matters which you are visualising since they both represent the same number, it may even be more memorable to swap between representations.

I just call every shape-based system a shaper system now. So, when I do say shaper system, I don’t refer to the exact numbers, use or etc but the system in general [shape-based system]. For example, I would say shaper system when I am actually referring to encoding japanese characters based on their shape. This however is just out of habit.

I mean, considering that I have spent a large amount of time with rote successfully, I can guarantee that even if something wasn’t memorable you can keep the information for a very long time if you review it appropriately (perhaps forever really). My personal view on this is more or less that when you make something more memorable you shift the forgetting of it to a longer time. So, something that lasts 20 minutes recalled at 20 minutes > something that lasts 40 minutes but is not recalled within 40 minutes. Along with this some interference theory which I have tested myself. As such personally I don’t really feel that spending a couple of days to come up with the right placing for each number/object combination will win in the long run over what is reviewed.

This said :

At first it seems intuitive to dismiss this as, when you are dealing with 10000 digits rather than 100 it will not function, but when I think about it there is a kind of ‘bonus’ to this idea that is easily overlooked. It is the same kind of thing as when you play chess, where you are going through the same moves but start becoming able to chunk the entire board. Perhaps it is indeed possible that this approach will allow you to use such chunking and, in this case, I see it being able to outperform standard memory palace variations, this is of course if it works.

This is actually a very interesting point because single digit numbers mean that if you had a 3-digit system you would have done this is one third the time or otherwise 300 digits in 4 minutes without much training.

Sound is not as slow as it seems but I do agree. My main goal with memory techniques is academic so I have kind of developed an intuition to be able to reason in images and think in images. There are obvious downsides and positives to this, but it is very functional, in the worst case you can visualise language rather than saying it, which I find for me is faster. I find if I were no longer able to use sound to think/encode I could definitely still live with image based reasoning.

I find interference is generally making things harder. If I am doing something entirely visually using visualisation I do not have any problem including language capabilities and vice versa. If I am doing something that requires verbal language and I am visualising text I do not have much of a problem doing these things simultaneously. It’s really where they cross that it has some issues for me.

I do think however that mental arithmetic being verbal isn’t the most functional. Abacus records claim computing 3 digit sums at a time in as little as 0.17 seconds. If I try really quickly to say 3 digit numbers on a stopwatch I can somewhat just scrape 0.11 seconds (this is starting to make me think that verbally sayind threehundredandthirtytwo is faster than threethreetwo ), this is however a far cry from calculating the sum of this along with another 3 digit number. I can visually just about scrape seeing 3, 3 digit numbers in 0.13 seconds visually in my head (using a stopwatch). This is one after another, if I displayed the numbers together at the same time this would be around 9x faster overall. It would be kind of interesting if you verbally read numbers simultaneously and could still functionally calculate but the kind of ‘fast’ pace I can’t really maintain verbally and only do in bursts. visually I at least managed to have 3 iterations. It just seems more efficient to do arithmetic visually. To be honest I do have a little temptation to try seeing where verbal arithmetic ends up when trained, especially since I find it rather easy to train verbal associations and can play multiple language sounds at a time verbally.

There is something I have always been meaning to investigate but never truly got a chance to because it takes too long to do and is tiring. However, It notably proves to me that high-speed verbal processing is a lot more resource consuming or otherwise tiring than visual processing may seem to be. I think I might make a post on this and one of your posts about high maintenance with a twist on it and investigate both things

2 Likes

I’m not convinced about that either. The more i think about it, the more I can remember from the shape based system I read about. Like for example, number 44 was a sail boat in this version. I imagine it was probably something like “2 digit shapes system”.

personally never have had this problem, but I imagine this may be also based on how you are placing objects in your palace. When I traverse through mine, I usually go in a very sequential direction so I always know I am processing the images forwards.

The problem is not related to the memory palace, it’s just that reverse numbers (83 and 38, 01 and 10, etc) can be translated in the same objects, based on their appearance; so why is “38” a snowman and not “83”?

As for the drumstick and skier, I don’t think it really matters which you are visualising since they both represent the same number, it may even be more memorable to swap between representations.

I think this problem is related to the paradox of choice; to much choice is not good and in high speed translating of numbers any choice may cause friction.

I just call every shape-based system a shaper system now. So, when I do say shaper system, I don’t refer to the exact numbers, use or etc but the system in general [shape-based system].

If my system qualifies as a shape based system, than some of the number to letter systems might also qualify as such, if the number to letter translation is shape based; 0 = o , 1 = l/i, 2 = n, 3 = m, etcetera. I’m afraid any classification method will lead to all kinds of overlap between classes.

I mean, considering that I have spent a large amount of time with rote successfully, I can guarantee that even if something wasn’t memorable you can keep the information for a very long time if you review it appropriately (perhaps forever really). My personal view on this is more or less that when you make something more memorable you shift the forgetting of it to a longer time. So, something that lasts 20 minutes recalled at 20 minutes > something that lasts 40 minutes but is not recalled within 40 minutes. Along with this some interference theory which I have tested myself. As such personally I don’t really feel that spending a couple of days to come up with the right placing for each number/object combination will win in the long run over what is reviewed.

My reason for switching from a word/sound based system to a(n) (actual) number (placed in the locations) based system has nothing to do with memory, at least not in the common meaning of the word. It is mostly about translation speed (I used the word “powerful”, that was a bit vague, I admit); it is my educated guess, that when you make the numbers a real part of the location (as opposed to for example a small label on your front door with your house number on it) like the 2 “4’s” on top of a pallet with paper, you will be able to look at any 2 digit number and actually see (part) of the location. I already know all the numbers of my locations, but it when I actually try to memorise a number sequence the translation speed feels like the biggest limiting factor. My goal is to get close to or break the world record some day (I’m not making any claims of probability in this regard, think of it as an aiming point, like shooting for the stars) and the only imaginable way of reaching this goal is to eliminate every little friction (no matter how small) from my system; the alternative is training (I don’ t even have the time for that) a huge amount with a half baked system (also; I’m not claiming my system is now fully baked) like so many other memorisers are doing likewise.

This is actually a very interesting point because single digit numbers mean that if you had a 3-digit system you would have done this is one third the time or otherwise 300 digits in 4 minutes without much training.

Any 3 digit system I know requires translation (not to mention a great deal of pre-memorisation), so I don’t think it’s fair to attribute that ability to me; I 'm not that special.

Abacus records claim computing 3 digit sums at a time in as little as 0.17 seconds

I feel a need to point at the big fat elephant in the room so to speak. And I get the feeling we share this sentiment: as much as I appreciate the performances of memorisers like Alex Mullen, it is my gut feeling, that the mental aritmetic performances like the flash anzan world records are on a different level, despite this being somewhat of an apples to oranges comparison. I have tried some flash anzan/mental abbacus training and i feel like if I train a whole year I might be on the level of the average 8 year old child in the countries where this a part of elementary school. I think that when entire generations of multiple countries are trained in memory techniques from a young age every current memory world record will be surpassed in ways that will blow everyones mind.

Thanks for the discussion

1 Like

I personally wouldn’t encode as a snowman, however if I did I could imagine a snowman with a trident edge in a 3 shape stuck to it to the right. Therefore I wouldn’t process a trident stuck the other way as 38 but rather a different image, it also wouldn’t make sense as the same trident because the direction would be different (in 38’s case the harpoon edge would be floating).

My experiments have shown that its faster to alternate choice when you have mastered two systems. So having more choice is better than having less choice for speed if you have mastered both variations is what I have experimentally found. Similarly also that for recall too much similarity is not optimal, I found that if I used different images for the same digits I surprisingly didn’t slow down but also that I was much better able to recall the sequence before having to do any reviewing. This is just to say it might not necessarily be bad. A much simpler variation of this experiment is to try writing multiple 1’s as quickly as you can on a sheet of paper and then try writing alternating digits which you may find more difficult such as 12341234. I find even though I would believe writing 1’s to be much quicker the stopwatch clearly makes the distinction that this is not true.

Yes this is true, but this is just like you wouldn’t exactly call the major system a sound system. Surely you could but there are not so many systems that this is a problem either-way. It’s kind of like how you can have animals be animals but also have a poodle, at least that is how I view it.

I don’t really think this is of much consideration though, depending on where I am writing things I would use the appropriate reference.

How do you mean a real part of the location vs label? I find that a label equivalent gives pretty instant location recall at the cost of being easily forgotten, similarly does the visual perception of the image from the digit if it has the same shape. Besides this I have mainly found training is the only alternative.

Perhaps there is a way to avoid this or speed up training.

I think it’s feasible to do 3 digit image placements in 0.1 second. This would completely blow away world records.Using abacus logic this should also be feasible in groups of 3, so something like 9 digits encoded in 0.1 seconds, I believe is feasible. This is then 90 digits in a second. Personally I like the sound of these kinds of results so they get me much more motivated than others.

It is indeed true, I haven’t seen many things comparable to flash anzan where someone who hasn’t practised is not only a lot less accurate but over 100x slower in some cases. It’s also not just one person but the records are actually being beaten by others and there are quite a few at this level.

This said, its not as though people who read books at a young age even excessively for years, start becoming 10x faster than others who are reading books. This comment is just to say that there is more than ‘frequency’/‘starting early’. Perhaps the ‘real reason’ why the abacus is functionally trainable to this extent translates to memory techniques. I have tried to find a lot of logical explanations and have found some but I don’t really have anyone who is below 0.2s 3 digit additions to confirm my deductions with. Perhaps I will take it upon me to master the abacus and confirm for myself or directly deduce the applications to memory techniques.

3 Likes

This discussion about flash anzan/mental abbacus, has in a semi logic way triggered my enthousiasm for the single digit (just placing the actual numbers in the memory palace) method. To start with; the (mental) abbacus is in fact a single digit system (every column represents a digit, by means of the positions of the beads) and it’s potentially amazingly fast.

Just five minutes ago I did a test run while eating diner; thinking of random numbers at what felt like one digit per second for a total of 23 digits (I failed to keep track of the number of digits and I also skipped locations in my memory palace; I haven 't used it in the normal way for quite some time). I managed to recall all without any problems. I realize reading numbers from a paper is more difficult, as it makes the visualisation harder (than when having your eyes closed), but still I feel this may be something to persue. One of the biggest surprises is that is actually more fun, than the translating numbers into objects methods.

To give you an idea of the way I did things, here is the first 10 digits and my intuitive number story:
753 198 469 5
I’m using the 7 to polevault over the office building fence;
the 5 picks up my car and trows it an the side;
the 3 is scratching the glass sliding entrance door;
the 1 is piercing through the reception desk;
the 9 is sliding down the stairs;
the 8 is growing inside the elevator and is getting stuck;
he 4 is slicing the toiletpaper in the toilet room;
I place the 6 over the toilet and use it to pull it away;
the 9 is jumping on top of my locker;
the 5 has hooked my locker door and has pulled in in to the air.

To end my enthousiastic review of this single digit system, I would like to say that it seems that it may not be important to obey the 1 digit per location rule. In my test run I skipped multiple locations without negative effect and I imagine, that I could do the opposite as well, like hook the car with the 5 and let it rest on top of the 3 or pierce the 1 through the desk and hammer it further with the 9 as a hammer.

The end.

3 Likes

You’ve written a lot of words here on the shaper without having to read the whole thing :slight_smile: . No probs, I don’t mind explaining it further it it helps.

The answer is no. I don’t think It’s original, I also mention that in the thread, so please have a read and you’ll have a better understanding of how the whole thing had started.
I also thanked all the board members who were involved in the brainstorming of creating and coming up different shapes. I’ve spent a huge amount of time thinking about this, and a huge amount of time finding the numbers shapes that was above 0-9.

to give you a quick idea; It’s an improvement to the basic number shapes, but also with a few added methods to how it can be used in many ways, and has a few benefits one of which is that it can be used in most languages. So it’s an advanced version of the basic one to generate images from the numbers that covers 00-99 which I think is simpler than major. It’s up to the individual to decide it’s better, slower, faster, etc as we are all different.

One of the added methods is The Clue Shape method.This is where you’d a see the 2 digit clue on the next object as it’s planned with it’s premade actions and the clues it would leave. The clue method can be used with any system with anything that makes sense to the individual. With the shaper, those images with its clues can be used in major or dominic system as well its ownself. Because we already have the shapes, and the clues, it’s kind of a ready made system in a box. Out of the few of the added methods to it, there is one more which I have yet to release but working on it is a method I’ve not seen on the forum or on google searches, but it’s so powerful it would be difficult to forget what you have just memorised. So it’s still in its unfinished but almost finished stage, and I’ll create a thread on releasing all the methods and how they can all be used.

Thank you. I have spent over 3 months on them, and a further 6 months to match the real images from the shaped images.

The shapes/images are what I see, and most people are able to also, but there are some people who sees different shapes in different numbers. So the idea is to change number shapes to the ones that is easier for the individual to see. There is no restriction :slight_smile: In fact while it was in a brainstorming session, other board members created their own, and we used each others shapes.

Well, change it to what makes sense to you. Those images are there to guide and to use if you like to. Another thing I do is to use any other shape that could be used again later on, so I actually use 2 or 3 images for the same digits if I have them or added them in my shapes library. It may be confusing how this is used, I’ve already explained this in another post, but to give a quick example; 11 would be to used as drumstick on the next object, and then if 11 is seen again, this time you’d ski on the next object. Do you get confused when recalling? I don’t. And I have a student who use the shaper also dont get confused with it, but if anyone does, then simply use one shape/image.

It does. An elephant looks the same to all people in the world, so if 19 is elephant, its is very easy to imagine it soon as you see the digits. So I think another advantage is that it takes less time to learn, some learn it almost instantly by the time they get to 99.

I strongly agree, you can in fact use single digits and place them in a memory palace and recall them with almost no issues. If anything; its is the easiest method and will work. Not only the single digits, you can also use that method for memorising cards. Simply use an image for every a single card at each spot in your memory palace. 52 image & 52 locis. it’s is a very powerful method.

However, if we want to memorise more digits a little faster, and with using less locis & rooms in a memory palace, then we have to “compress more digits into a single image”. This of course can be accomplished by using the 2 or 3 digits systems available, and by applying the methods that works for you. When I’m competing I memorise 4 digits at a time. If I’m not competing, to compress more digits into a single image/scene I use 5 digits at a time which is far easier than 4 digits with using the middle digit as the action. So I think this all depends on the individual and what works best for them.

Btw, I do like your reverse memory palace method. A lot!
I also have a good idea of the order of which the digits can be memorised while using the reverse place method. I think it can be a powerful method, and perhaps can also be used in competing. I’ll be doing some experiments I really like the idea.

1 Like

You are right, but I was asked what I thought of this system and I had to choose between speed reading your Shaper system thread (I had allready read your original post, but not the entire thread) and reacting in some interesting way or reading it really well and not having time to react at all (or at least not that same evening).

However, if we want to memorise more digits a little faster, and with using less locis & rooms in a memory palace, then we have to “compress more digits into a single image” .

Those are 2 assumptions that I imagine the major part of memorisers hold to be true and I used to be on the majority side untill recently. But now I feel both assumptions might be wrong. Let’s look at both:

  • memorising faster means compressing more digits in an image.

Allow me to state the obvious: if you can memorise a single digit (using a single digit system; numbers as they are) more than twice as fast as a 2 digit number (using a 2 digit shape or number to letter method) the statement is not true. I have not done enough testing to make any hard claims, so the following is just my intuition translated in words. I think that it is possible to create a single way to memorise each single digit (without changing the look of the numbers; so no translating into objects), that will allow for a speed that is more like that of the mental abbacus (flash anzan) stars than of the typical 2/3 digit system users. The mental abbacus is a single digit system (for calculation), but I don’t think these practitioners manipulate only one digit at the time because of that.

  • When you want to use less locis & rooms in a memory palace you have to compress more digits in a single image.

Allow me a chance to change your mind on this statement by means of an example. I will first give you some ways to memorise a couple of single digits (keep in mind these are literally prototypes; I intend to do some brainstorm sessions trying to come up better ways for each single digit).
0: a ball with magical qualities; it can bounce, but it can also penetrate trough every object;
1: a speer/pin;
2: a sharp hook, always pulling in a downward way (5 is also a hook, but is pulling upwards);
4: a mini skateboard ramp.

Let’s use one textbook loci, “the front door of your house” and the number sequence: 41102.

This translates into:

jumping over the 4 (ramp) like a skateboarder, throwing 2 spears in the right side (just based on intuition) of the door, throwing a ball through the left side of the door (resulting in a round gap in the door) and finally watching (“2” acts autonomous, again just a prototype, so who knows how this may change) the 2 cut the middle of the door and and slicing it all the way to the ground.

So that’s 5 digits in one location.

I have a lot more positive/promising intuitions about this system, but I don’t think I know how to put these into words in a convincing way. So this is basically what I can offer you in terms of making a case for this system.

Thanks for your reply.

This seemed like a bold statement. I’m afraid, you didn’t change my mind at all. I am still sticking to what I have written. With No disrespect.

Your example displays the basic Linking method. That is the basics of memorising a single or more items in a single location.

You can in fact add countless amount of digits in a single locus if you can keep the story going by linking them.

I would like to invite you to join the Memory League if you are not there already. Practice all your systems, be it 1 digit or whatever, try to compete with other users every so often, and then come back and tell us your experience with your scores and timings along with the systems and methods you have used.

1 Like

My intention is to aim at sort of a hybrid between linking and connecting to the memory palace. In my example everything is connected to some part of the same loci but it is obvious that the story is vital in fast memorisation and remembering of these 5 digits. Simply putting every single digit in a location and disregarding the surrounding digits will probably not work as well.

I’m afraid, you didn’t change my mind at all. I am still sticking to what I have written. With No disrespect.

I don’t think my former self (even from just from a few months ago) would be convinced when reading my post (from the present me).

I would like to invite you to join the Memory League if you are not there already. Practice all your systems, be it 1 digit or whatever, try to compete with other users every so often, and then come back and tell us your experience with your scores and timings along with the systems and methods you have used.

I hereby promise, that if my 1-digit system gets up to some speed I will join and tell all about it. It won’t be very soon I’m afraid; work has been kind of crazy (in a good way) so I don have the luxury of spending much time on a daily baisis to memory sport.

1 Like