(Another) Using a Memory Palace to Memorize a Textbook

spencer_green

(Another) Using a Memory Palace to Memorize a Textbook

post by Spencer_Green · 2021-06-14T22:12:26.697Z · LW · GW · 4 comments

  Approach 1: Picture-in-the-mind
  Approach 2: Story-in-the-mind
  Intermission
  Approach 3: Extending Dominic
  Approach 4: Chained Palaces
  Soo, what?
None
4 comments

Why do this? I was a year out of graduate school, but I could already feel my knowledge leaking away. This is a frustrating experience that will be familiar to any of you who've switched fields, if you are no longer working with your hard won knowledge/skills, they seem to vanish. The mind is a leaky sieve, without constant refilling it empties quickly.

Like, if you asked me to write down the Schrodinger equation (my PhD was in physics), right this instant, I'd have a 50/50 chance of getting it right (I just tried this, by the way, and I failed, now all I'm left with is a wrong equation and a slightly hollow feeling like what Scrooge McDuck might feel if he opened up his vault and all that was there were a few quarters. Canadian quarters.)

My experience with memory palaces were that:

They are a bitch to get right, you need to put in significant practice to get proficient, but...
They give more permanence to memories

This permanence was exactly what I wanted: I wanted to be able to remember things like the Schrodinger equation, even if I hadn't thought about it in years.

So, instead of memorizing the entire textbook, I narrowed my vision-- could I memorize the important equations and figures in a chapter. The goal would be to be able to deliver a lecture on the chapter without looking at written notes-- being able to move from important equation to important equation, and being sure of your derivations.

The usual format of a palace is: you take a place you are familiar with, and you mentally 'place' objects there, and then you walk through the palace in order, visiting the objects that help encode memories.

Approach 1: Picture-in-the-mind

I first tried just taking a 'snapshot' of an equation, and placing it on pedestals around my palace (which was just my dingy basement suite apartment). Unfortunately, this was an abject failure. My powers of visualization were not enough to create permanent 'mental' snapshots, they disappeared to dust when I wasn't focusing on them. I needed something more memorable...

Approach 2: Story-in-the-mind

For every equation, I tried creating a little visual 'story' for it. This was fairly free-form. Say I wanted to memorize y = x/2 +1, I might picture an "x" sliding down a divisor "/" into the waiting arms of a "2". This was easier for me to recall than a static picture of the equation, the visual story allowed me to 'move' through the equation, the same way you might if you read it off the page. The problem was my visual language wasn't consistent and had to be invented on the spot-- this made the storage process slow and the retrieval process prone to error.

Intermission

I took a break from this for a year or so. I took up another challenge-- memorizing the names and dates of office of all American presidents. In testing out approaches, I came across the Dominic System of memorization. Used by famed memory athlete (who looks, in the best way possible, like a pornstar from the 70's) Dominic O'Brien to win the Memory Olympiad multiple times, it is a refreshingly straightforward scheme.

Take the following letters (0, A, B, C, D, E, S, G, H, N) they correspond to the numbers (0, 1, 2, 3, 4, 5, 6, 7, 8, 9)
Take all combinations of these letters (giving 10*10 =100) unique combos (AA, AB, ...HA, HB)
For every combo, think of a celebrity who has those initials (AS-> Arnold Schwarzenegger) (this is surprisingly difficult-- like, who has the initials H.E.? (I floundered on this for a while, now I picture the Joker giggling "He He")).

Now you can memorize things with 2 digits! Say you parked your car in spot 62, this becomes SB, which could be Simone Biles, so you imagine Mss. Biles tumbling over your car. Useless though. You can't even memorize a date (4 digits).

To memorize 4 digits, you use the Object Action system. Each celebrity is intrinsically an Object, and for each celebrity you give them an Action. Say you wanted to memorize the date 1990- (AN)(NO) (Amber Nash) (Nick Offerman), so you might picture Pam (from the cartoon Archer) making a canoe (this is the Action I've given Nick).
You can expand this even further by using the Object Action Item system (each celebrity now also has an item associated, and you can memorize 6 digits.

The basic premise is, by drilling the conversion between numbers and celebrities until it is second nature, you can quickly construct memorable, visual stories that all have the same common visual language. By placing these stories around a route (or palace), you can memorize very long series of digits.

But how does this help memorizing equations?

Approach 3: Extending Dominic

One hundred things is actually a lot of things. Turns out, you can fit most of the common symbols of mathematics into 100 boxes. So I did.

You need to include the English alphabet (26 boxes gone: (0A-BS)) (this seems weird, as you are actually increasing the information here-- 'j' is encoded as "AB". But "j" is really encoded as "Anthoy Bourdain", so it is still encoded as one entity.
You need to include the greek alphabet (21 boxes gone: (C0-E0) )
Basic arithmetic (10 boxes gone: '=', '+', '-', '*', '^', 'root', '(', ')' "|" ) S0-SN
Trig (8 boxes gone: e, ln, sin, cos, tan, sinh, cosh, tanh, ) (G0-GG)
Calc (4 boxes gone: d/dx, integral, del, Summation) (H0-HC)

Most physics equations can be written as a combination of the symbols encoded above. You'll notice I have room to spare! I still have the 'N' column. I also have gaps, where for symbol hygiene, I try to start new categories of symbols on new columns. All in all, I still have 31 boxes left, if I want to extend this system further!

This post is getting long, so I'm not going to write down examples of this. It was a pain to set everything up, but I encoded everything in flash cards and worked through them instead of hitting up reddit when I was in the bathroom, it only took a week or so to learn. The problem was density. Equations can have a significant number of symbols in them, even fairly simple ones. (And why bother to memorize simple equations. What you're really after, you greedy little STEMLord, is the ability to draw some equation from Jackson Electrodynamics and Magnetism at the drop of a hat and pistol-whip the insouciant lout who dared question you into complete submission.)

So, if an equation has 12 symbols, then you have still have to build a visual story containing 12 elements, on the fly, and keep it in one location. It was too much to manage, which led me to my current approach.

Approach 4: Chained Palaces

The way to order and remember long visual stories is to use a memory palace (duh). So, now what I do is I have a 'main' palace (like a friend's house). Every place within this palace encodes two things: the equation number I'm trying to memorize, and a link to another palace (this is surprisingly easy. I don't need to do anything special to encode the link, weirdly.) In that linked palace, I divide up the equation and place the visual elements in some familiar route.

I thought I would have trouble coming up with enough palaces, but it hasn't been an issue so far. I also re-use the palaces, and as long as the context between the uses of the palace are dissimilar, it doesn't appear to be a problem.

Soo, what?

I mean, it works. I can successfully encode equations, and have long term recall of them. I actually did encode a chapter from Jackson a year ago ;), and though the equations are a bit rusty, they are still there.

This approach does take a while though-- you have a lot of setup time to drill the visual language enough that it is reflexive. Then you have to build the visual stories (this gets quicker with practice, but still). Then you have to drill the stories a few times (easy with Anki) to make sure you've got it.

The big win though is verification. Instead of taking a stab at writing something down that looks right (and hoping you have an even number of sign errors), you can write down an equation, and check the answer against the visual story in your palace.

4 comments

Comments sorted by top scores.

comment by ryan_greenblatt · 2021-06-15T12:09:11.307Z · LW(p) · GW(p)

First of all, interesting post. This gave me a better understanding of the process of creating a memory palace and updated me toward thinking memory palaces are much harder than I expected.

This post has made me think that memory palaces are not useful for me; typically, I want to memorize things either for recall faster than internet lookup or to make it easier to build intuition and connections.

This makes me wonder why you went through this and what other benefits exist. Why not just use the internet as slow memory given that memory palaces require slow reconstruction anyway?

Replies from: Spencer_Green

↑ comment by Spencer_Green · 2021-06-15T16:05:07.804Z · LW(p) · GW(p)

Great question! First, for me at least, memory palaces are asymmetric in effort, they can take a fair amount of effort to build (especially if you are trying to encode a lot of information), but retrieval is fairly constant (not instantaneous, but on the order of seconds.) As with anything, this gets better with practice.

And I don't want to mislead you! Memory palaces are fairly easy to get started with (like if you want to remember simple things, like a grocery list, you can plop groceries around a path in your house). The complexity in my techniques comes from developing a consistent visual language to encode things (like numbers with the Dominic System, or equations with the Extended Dominic System).

As for why I tried (and am still trying to do this!) there are a couple of reasons, First, I have been thinking about what it means to know something in the age of the internet. Like you pointed out, why not use the Internet as an external memory? By that definition, do I know all of Roman history?

Obviously not. Even if all the facts are there, they aren't in a format that allows for connections to be made. So I have a very similar idea as you do-- making knowledge internal allows us to develop intuition and connections.

I still think that this is broadly true. I used a memory palace to memorize the dates of office of American presidents and Canadian prime ministers, and I've noticed that it gives history a lot more context-- they act like meridian lines throughout history, reference points. My next project is to memorize English heads of states (they go back a lot further, so extending my temporal grid lines).

The second reason I'm trying this is more personal. I have always been amazed at human's mental capacity-- Greek poets reciting the Illiad by memory, Aleksandr Solzhenitsyn[1], who wrote a book in his mind using a rosary as a kind of word-abacus, the feats of calculation of human computers like Katherine Johnson, even reading about the Mentats in Dune was, and continues to be, inspiring to me.

If you look at the great strides we've made in understanding the human body in the last century, and our ability to train and condition strength, pack on muscle, whatever metric you want, you (or at least I) can't help but wonder if there is a similar revolution possible with our minds. (I know there is a bunch of studies that suggest we can't improve our working memory/IQ, but I take that more as a challenge).

The third reason is more... ahem... practical. People listen to you if they think you're smart. They will give you jobs and raises if they think you're smart. It also feels good when people think you're smart. And you can contribute more to a conversation if you don't have to say "Hold on, I need to google this." or "I don't know, but give me a day or so to think about it/google it."

Sorry for the word vomit

[1] https://nevalalee.wordpress.com/2016/07/11/solzhenitsyns-rosary/

Replies from: ryan_greenblatt

↑ comment by ryan_greenblatt · 2021-06-18T12:17:33.589Z · LW(p) · GW(p)

Thanks for the response.