Learn to remember everything: The memory palace technique | Hacker News Reader

Learn to remember everything: The memory palace technique

mostlymaths.net

201 pointsRBerenguel68 comments

I've always been pretty frustrated with my inability to remember simple things, while having knack for recalling useless information. I rarely can remember birthdays, anniversaries, phone numbers, important dates in my history like when I graduated or the specific date of when my father passed away. I have hardly any memory of grade school, only can recall two of my teacher's names in high school, etc. Granted, some of those are less useful than others but they're things that people tend to remember and I don't. Instead, I memorize odd things like a Windows 98 CD Key, for example. I must've reinstalled it so many times that it just stuck in my head. Completely useless and yet it'll be with me for many more years I'm afraid.

I have found that I tend to remember things more if I put them into a rhythm, as if they were words in a song. I'm not musically inclined so I kind of find that odd, but it works. I used to work in inventory for a Best Buy warehouse and I'd often have to put UPC codes into a terminal for new inventory. I couldn't see the codes from the terminal most times, so I'd quickly memorize the code in my head by putting it into a rhythm. 1...2...3 4 5 6...7...8....9 10 11 12. Would be curious to know if that's common?

Anyway, very interesting article and I'm definitely going to try and put it to use. See if I can create a memory palace for Rubix cube algorithms as I've recently decided I want to be able to master the 3x3 one. Now if I can only force myself to forget that Kermit the lettuce head is going to greet me when I get home by warning me that Kevin Bacon is chasing pigs through my living room, that someone accidentally formatted my SD card bed and that the sun is supposed to be extra Orange and juicy today. I think I'll keep around Scarlett doing her thing with an Onion Ring hula-hoop though. :)

Look at how the Aboriginees use songs in their memorization techniques. Tribes or families would have songs about locations of food and water, and when two people from different tribes would marry, they would teach each other their songs. Isn't that the best dowry?

I memorized pi to a hundred digits out of boredom in middle school, and I naturally did the rhythm thing, mostly with triplets of digits, sometimes otherwise.

The interesting thing is that I can't remember any of it without using the rhythm to help me; for example, if you asked me "what comes next after the digits '197'", I have no way to answer that question without going through the digits until I get to the part that goes "..795 0288; 4-1-9-7-1 693 993..." and then going, aha, here it is. If you started ten digits in, grouped the digits differently, and read them off to me in a different cadence, I wouldn't even recognize it was pi!

I'm the same way. I had to go through all of pi to find the part after 197. For me, it's broken up like this:

3.141 59 26 53 58 979 323 846 264 3383 27950 288 4197 169 399 375 10 58

So it was even a bit hard to see the 197 in there, because it's "4197" for me.

3.141 59265 3589 79323 846 264 338 3279 50288 41971 69 399

I liked bursts of 3-digit chunks, combined with 5-digit sets that were important addresses in the C64 memory map. (41971 in BASIC ROM, 50288 within the $C000 hidden RAM area).

http://sta.c64.org/cbm64mem.html

The computers you use change how you think, permanently.

FYI, here is a method to solve the Rubik's cube that's optimized for memorization (it contains as few formulas as possible):

http://beust.com/rubik

That's not particularly optimized; it's als from 2003, when speedcubing was not as advanced. If you want a fairly good attempt at an understandable minimum method, look at http://www.speedsolving.com/wiki/index.php/8355_Method

(Although memory shouldn't really matter for such few moves. Blindfolded solving is where it gets interesting.)

This technique is not for speedcubing nor did it claim to be optimized. Its goal is to be easy to memorize.

By definition, speedcubing is the opposite of "easy to memorize". The easier it is to memorize, the less formulas there are, therefore the more you need to do.

Speedcubing is about solving a cube in less than 25 seconds, which pretty much only leaves room for 3-4 formulas. That's a lot to memorize.

There's a book by Jonathon Spence called the Memory Palace of Matteo Ricci, describing the life of an Italian Jesuit who lived in China and was an advisor to the court (I can't recall if it was late Ming or early Qing, but it was in the 17th century). Besides sharing advanced Western science and artistic techniques, another way in which he and other Jesuits inserted themselves into the life of the court was by using the memory palace technique to learn spoken and written Chinese, understand the cultural and social aspects of the court, and impress the court officials with some incredible feats of memory. One trick I remember involved reciting a long passage of Chinese backwards after hearing it once, which utterly astounded those in attendance.

Good article. I've been studying Dominic O'Brien's techniques for a while now, he's the world memory champ - http://en.wikipedia.org/wiki/Dominic_O%27Brien

It takes some time to get the system down, but it's really handy for memorizing numbers, a pack of cards, or any other structured information. Basically, I've got a journey with 52 stages memorized along with a person and an associated action for each two-digit number from 00 to 99. When I was practicing regularly before I was able to memorize a pack of cards in 3 minutes. It may seem weird practicing this, but it turns out to be a pretty good gauge of your concentration and mental focus, just like how being able to run a few miles is a good measure of your physical fitness.

Highly recommend Joshua Foer's book, "Moonwalking with Einstein", to learn more.

Is there a book that you could recommend on O'Brien's techniques?

"How to Develop a Perfect Memory"

I've been listening to his Quantum Memory Power audiobook. It covers a huge range of memory techniques: memory palaces/journey method, mnemonics, visualization of numbers, etc. Some of his pop-culture references are a bit dated, but otherwise I find it to be a great system.

http://www.amazon.com/Quantum-Memory-Power-Improve-Champion/...

Funny you mention that, I listened to those audio CDs as well! I think that and "How to Develop a Perfect Memory" are the best resources on him, he's the man.

I'm interested in combining the memory palace and peg techniques, specifically for sw dev.

Assuming you understand the fundamentals, I think there is a subset of any programming language, data structures, and algorithms for the programming paradigm you are working with where memorizing some of your "working" tools is a good idea.

From my viewpoint, this might be helpful in a couple of different ways. First, if you are hacking away in a prog language, you need to have some decent set of it "embedded" in your mind. Memorization might help me for those languages I don't get to touch every day, but would like to keep handy in my toolbox. Second, the data structures and algorithms part would be nice to flash against new problems. E.g, here is a prob, can I quickly flash it up against my set of memorized data structures and algorithms to see if something "fits."

I wonder if by the time you come up with an appropriate internal visualization for most data structures and algorithms, you'll probably have enough of a taste for it that memorization isn't useful anymore.

As a side question, I think it might be useful to think in terms of how much code you can write without needing to rely on autocompletion in your IDE or without referring to an outside reference. Just curious if other HNer's have opinions about that kind of prog lang command as a measure of skill . . .

A few years ago I got decent at a similar method so I could memorize a list of 20 items pretty quick. I thought I was the cat's pajamas for awhile until it dawned on me that I had never used it once outside showing off the trick. I've since lost the skill from non-use. Not to troll against the post or the poster, just my personal experience data point. But hey, I love seeing and learning new party tricks, so don't let me discourage anyone from trying out interesting things.

I keep waiting for someone with more spare time than myself to bring the memory palace idea to an augmented reality app.

I have a feeling medical students will pay a decent amount of money to walk around their apartment with an iPhone held out in front of them.

Also, the effectiveness of this idea implies that museums without a single artifact in their collection can be extremely educational by wisely creating a spatial environment.

A tour led through a "Garden of Presidents" or a walkable map could help students on field trips actually retain much more knowledge than in the classroom.

I actually pulled this up on last weeks startup weekend and we made some prototypes.

We tried it first with streetview, where you walk around and learn new words on a different language. We have the best results when we combined it, with a special spelling where the letters were objects also. (like in elementary school, for example "A" is for Apple) Like that you will remember the objects and not just some random characters.

The "let's go for a walk" and the "stay in this room" mobile versions are still in the pocket, so let me know what do you think about this.

Where can I have a look at them? (I don't know what you refer as "pocket")

I am sorry they are not online now. By pocket I mean ideas we haven't tried out yet. I imagine the mobile app version to have 3 separete functionality, and we only covered 1,5

1. "I would rather sit" Is a streetview walkaround with predefined routes. 2. "I want to stay in this room" Is an AR version where you walk around in your room and place the words where you want. 3. "Let's go for a walk" Would give you new words when you stop somewhere in the city.

Of course there are other task and opportunities, like testing, spaced repetition, linkwords etc and unfortunately the team got separeted now (both in space and in thinking) so I am not sure if I would make this project hight priority...

Still I would love to hear your toughts.

A new book called "Moonwalking with Einstein", by Joshua Foer covers the Memory Palace technique and others.

Check out the review at the New York Times.

http://www.nytimes.com/2011/03/13/books/review/book-review-m...

Also a nice segment with him on NPR: http://www.npr.org/2011/02/23/134003962/Moonwalking-With-Ein...

I like "Your Memory : How It Works and How to Improve It" by Higbee in addition to the Foer book. It is a little more pragmatic and thorough in description of the techniques discussed in "Moonwalking with Einstein."

More general and useful is spaced repetition: http://www.gwern.net/Mnemosyne.html

This is an awesome technique and I can't believe it's not taught in schools. It's better to think of a previous home rather than your current one though, otherwise things can get confusing.

As you say, it is better to use not-often used places. I use my home for grocery lists, as they are usually short-lived lists. For longer lasting stuff I use older places, or not-frequently visited.

Great article Ruben. I've been trying your method out for the past days, but the problem is I don't often have to remember things. I wonder if methods like this one are getting irrelevant in the days of computers and smart phones..

Probably the most useful use case is when you have to shop for groceries in a supermarket. My SO, Helga, usually tells me what to buy and I usually forget everything. But lately I've just been typing everything into my phone and that works. This method might work for that as well - Memory Supermarket, anybody?

This technique works very well with names. Meet a person once, store his/her quirks alongside a name in the mental palace and every time you meet this person you can give a personal greeting.

Now imagine doing that on your first day at work in a big company, or remembering the name of someone you met on a conference two years ago.

I see computers as addressing a different part of the problem.

Lets take as a given that people need a way to recall pieces of information for use in the future.

1. Initially people would probably just wander around aimlessly and hope that their natural capabilities would be sufficient. Apparently this was not the case.

2. To reduce these short comings people developed memory techniques to better organise the information to be remembered. By chunking pieces of information in mnemonics and associating facts to these memory journeys they were better able to remember them.

3. Next people developed the technology of writing. This tool had a number of advantages over the previous techniques: less effort was required to store the information, it maintained its accuracy over a longer period of time, and the information didn't need to be recorded (memorised/written) by the person who would need to use it in the future. The method also had disadvantages: this new external form of memory required the person to take something extra along with them, as the number of pieces of information became larger it becomes harder and harder to sort through them to find the correct piece whereas the human brain doesn't seem to become as `distracted', and the individual pieces of information become less connected.

4. I see technology as helping to reduce these disadvantages. The internet means I can check email from any computer in the world and not just at the one computer it was composed at. Search engines assist in finding a particular piece of information within a greater body of information. I am not aware of a great example of information synthesis but I suppose that certain parts of text mining are making steps in this direction: eg, sentiment mining Twitter or trend detection.

Cognitive scientist Daniel Willingham, in his book _Why Don't Students Like School_, says that memorizing facts is important for developing higher level reasoning. Think of it as building up a large processor cache. If you have the relevant facts in your brain, you can piece them together much more quickly to build up higher knowledge. Always having to look up things on Google is like disk thrashing.

Surprisingly few people grasp this concept, and from my point of view the idea that rote memorisation is a waste of time is remarkably dangerous.

I don't know that this technique is a big win for this kind of memorization. For example, multiplication of digits probably needs to be in L1 cache. You don't want to have to search for the result in your memory palace, which seems more like main memory.

Indeed! There are some things that you need to force feed, and others that you can place in a memory palace or use in an associative array of thoughts. I.e. multiplication tables or foreign basic words need to be in your L1 or L2, but the causes of the First World War (unless it is something you are deeply related to) can perfectly be outside it.

To continue with your example, multiplication of digits gets to be in L1 having lived in main memory for long enough, and being retrieved from main memory often enough, to stay there. The same will be true of anything which is accessed that frequently, but unless things are in the main memory to begin with, they can't possibly get promoted.

There are many methods including google that help retain knwoledge outside one's brain.Cheat sheets, visualizations, intellisense, etc...

And many times, using one of those methods to recall something is faster that using memory. for example using intellisense to see what can i do with a certain object is much faster than using memory.

I think a more useful property of the human memory is the ability to connect different pieces of knowledge, both consciously and unconsciously. Connecting two pieces of knowledge using a computer is much slower and harder, and requires a lot of intellectual effort.

Exactly. You can use intelligence to find setattributevalue (or SetAttributeValue, or was it setAttribute?) easily enough, but unless you remember what it does, you won't ever look.

Thanks Arnor :) As you say, nowadays we have less uses for our memory: that's why paper, pen and iPhones are for. I'm trying to use the memory palace technique to help my language learning (you know, like Icelandic ;), but it is also good to remember things like "hey I should do this by tomorrow" that appear when you are in the shower, or shaving. When you usually have no gadget around (waterproof iPhone, anyone?).

I usually carry many gadgets around (iPad, Ben Nanonote, iPod Touch, pen and paper are usually always with me) and as such, it is far easier to write down important stuff. But when I don't have them available, let's visit some standard palace and put the stuff there.

As for memory supermarket... If you always go to the same supermarket, you can use it as a palace, in the appropriate places you just put some kind of reminder, like a lemon avalanche or the big Bread Man.

We tried this technique out last week for, memorizing new words on a foreign language.

The interesting thing (besides that it is working fairly good) was that you don't have to come up with your own familiar places, routes, etc, but a site can do this for you and it would still work. I can show you a new place on streetview with some visual cues and you will still remember the places there.

And it is actually easier to get started, than when you have to do that on your own.

I used to use this technique as a kid to win money off friends, betting I could instantly remember the order and nature of 50 objects read out to me. If you'd prepared properly it was embarrassingly easy. Recalling the list a week later was trivial for me but astounded everyone else.

The author's place method is much superior to my idea then of using an arbitrary list of objects which I'd previously memorized and then associated with the spoken items by using outrageous mental pictures, Memory Palace technique is to be recommended.

We played a memory game at school once where everyone sat in a circle and said their favourite activity. For years and years after I could remember the entire sequence. I could probably regurgitate the whole thing now if I had a few hints. Another one that works short term for exams and such is 'definitions , lists and diagrams.' On left page of notebook write say 'definition of Agile development' followed by a dash for each point on the agile manifesto, followed by 'diagram of [something agile related. Textbooks always have diagrams]. ' on the right page write the actual definition, followed by the actual list, followed by the actual diagram, but use your own words, not the textbook's. Glance over it, then cover it with a piece of paper and write the entire thing out again from memory. No peaking but at least you can count how many list items are required from the other page. Check you got it right, reread any mistakes, and then you'll find that after having written it out twice you can easily remember the whole thing when the exam comes. I find any sort of thoughtful reinforcement makes memory very effective.

I wanted to ask though, do you guys memorise APIs and keywords and other programming mattered, rather than just picking up or looking up as you go? Does it help you program much faster? Of course it's a lot to remember considering all the rules and contexts and parameters.

Lettuce, Bacon, onion rings, SD card, oranges. This is weird it works so well.

The standard work on the ars memoria is Frances Yates' The Art of Memory, which covers Renaissance practitioners (Lull, Bruno, Fludd, etc.) in some detail. These techniques date back to the classical art of rhetoric, which cultivated the capacity to memorize long speeches or poems.

More recent scholarly work has been done by Mary Carruthers. I have found her books, "The Book of Memory: A Study of Memory in Medieval Culture" and "The Craft of Thought: Meditation, Rhetoric, and the Making of Images, 400-1200", to be more engaging and useful than Yates.

Another interesting view of all this is via John Crowley's novel "Little, Big" (certainly heavily influenced by Yates).

Preston

This works, but I don't know if its web scale.

Like, do you use the memory palace at first, and later it just fixes into unconsciousness? Because there is no way greeks really had hundreds of memory palaces where they structured all their knowledge, right?

We can't be sure about the greeks, and I still don't know how well it scales personally. Memory champions claim to have quite a lot of palaces and ways to cram information in them (for example, for memorisation of digits of pi), but my personal experience is not that big as of yet.

As a matter of fact, an informal list of places I can use as memory palace, as I know them from memory ranges around 70 "palaces" with between 4 and 15 rooms each. If you cram 3 items per room its around 700 things to remember. But you can get better.

You can also use association (I am about to write a post about it) to get facts about books, people, things without using a new palace for each kind of data type.

I've also found it quite hard to "overwrite" memories. At least, you need to let the pegs rest for a while before reusing them.

When he writes about this sort of thing Derren Brown (British mentalist) mentions that once you get to a decent level of skill with this you can just make up new rooms in your head; he talks about "dropping" overused rooms and starting fresh ones.

You can see the very impressive results from the previous World Memory Championships here: http://www.worldmemorychampionships.com/MemoryAchievements.a...

  > I don't know if its web scale.

Does it support sharding?

Yes, if you run through the memorized list once a day, in a few weeks you can free up that parrticular palace for other work and keep the memory --- you've memorized by rote.

As for scale, many memory workers use a linked list as well as the memory-palace array. Associate each item with the previous item, and you don't need 10,000 room palaces; just someplace to put the head of the list. That's how I memorize decks of cards: http://blog.diiq.org/post/3366119815/what-are-the-step-by-st...

An excellent resource is Lorrayne and Lucas' The Memory Book:http://www.amazon.com/gp/aw/d/0345410025/ref=redir_mdp_mobil... , which has been lifechanging for me.

Thanks for sharing this. I'm more prone to associative stuff, only that they require more work. Will have to move some lessons I'm memorising to a linked list to free space ;)

I'm pretty sure they have lots of palaces. But think about a subject like "World History". How are you going to store that?

I also noticed a problem I have with using something like the Dominic System to store phone numbers, which is remembering the first image.. after that, this image just "pulls" the others by association.

The only drawback to this method is that it makes it easier for dream-spelunking thieves to commit a heist in your mind.

When I work on a large piece of software I use techniques like this to build a model of the software in my mind. The various layers seem like physical objects or places in my mind. It is a hard point for me to articulate but when I know a software code base really well I am able to immediately answer question about whether certain things are or are not possible. I always thought of this as the "topology" of the solution. I know that is why I have a preference for certain design decisions. Some software seems like a well designed house with layers that do one thing really well and can offer strict guarantees about the state of the system.

Sometimes I am asked to do things to the software that conjure up images of mobius strips or klein bottles and I cringe. Only because of what it is doing to my mental model.

I'm curious if this has any negative impact on the accuracy of long term memory recall. For example, I can imagine that I would place these exaggerated items in a friends house...then, years later, in a discussion with said friend; "Hey, do you still have that SD card bed?"

You need to also learn a way to empty out your mental palace, or learn a way to build new rooms. After learning about this technique over 15 years ago I stored 3 things that I, to this day, know by heart. After learning Derren Brown's technique I still have these items in my mind, probably to the day I die.

http://www.youtube.com/watch?v=mhZcHoU-QR8 (6 minutes Derren Brown technique)

I like how this skill is innate. You don't get born with great memory, you can learn it (not photographic memory). When mastering the memory palace technique, one could store and recall the order of 20 shuffled card decks.

Here is how i remember stuff, with mnemonics

http://seriouslackofdirection.blogspot.com/2008/08/rememberi...

There are a bunch of other techniques like this (which I've used with occasionally surprising success) at http://www.ludism.org.

I recently wrote an article for my company's in-house staff magazine about how Google was my favourite web site because it freed me from having to remember facts.

I think it's a waste of effort to devote time to memorising mere details, when these can readily be looked up. I think it is much more sensible to remember connections between facts (i.e. to theorise and systematise based on experiences) than it is to just store the facts themselves.

The problem is that, unless you know a lot, you can't relate facts. I also prefer associative memorisation (i.e. I can think of subject A, which is related to subject B, caused by subject C and leading also to subject D), but you need to know your stuff, too.

I wish this post was around when i was in the 9th grade so i wouldnt have spent so much on Kevin Trudeau's Megamemory just to learn this.

I still remember the 5 things in this list and I wasn't even trying to... I wonder if it works with bland data?

What do you mean by bland data?

4 hour memory

Why learn tricks to remember stuff if there is paper and pen (or portable computers)?

broswer broken, leaving comment to find later

There is another memory technique where each item is like a member of a linked list. Say you have to remember bread, milk and sugar. So you imagine a loaf of bread that when you cut a slice off a cascade of milk comes gushing out and topples a dam made of sugar cubes. Then you discover that you already have milk, so you alter the image to a colossal slice of bread crashing down onto the sugar cube dam. I'd like a technique analogous to this where you start with 52 cards in your list and as the cards are drawn from a deck they are also dropped from your list. It would make it easier to see what's left

Reply on news.ycombinator.com