I'm really glad you brought up spaced repetition since spaced repetition and the techniques described in Moonwalking with Einstein happen to be two of the items on my relatively small list of the most amazing things humans are capable of that they neglected to teach me in kindergarten. I'm sure most people on HN are unfamiliar with both, so I'd like to give a brief overview of what they are and how they work.

The techniques described in Moonwalking with Einstein require serious time investment up front, but allow you to cultivate a memory that rivals that of people with savantism (who incidentally tend to use the same general techniques but do so naturally). Spaced repetition doesn't produce as amazing results but doesn't require initial time investment and is a far more efficient method of memorizing data than conventional approaches. The only downside to spaced repetition is that most software implementations of it require daily review and don't work optimally if you miss a day (recent versions of the proprietary SuperMemo, Piotr Wozniak's own software, are supposed to be much more forgiving), but you don't need to invest much time each day.

So how exactly do these two memory techniques work? Spaced repetition is predicated on the notion that the longer you wait before reviewing data, the longer that information will stick in your head. That is, as long as you are able to recall the data when you review it. If you wait too long, you'll completely forget the data. So there exists an optimal length of time to wait before reviewing a piece of data, and as with the American game show The Price is Right, you can approach that optimal value by increasing your estimate, but the moment your estimate exceeds that optimal value, it becomes less useful than all possible underestimates. The goal of spaced repetition systems (SRS), such as the software programs SuperMemo, Mnemosyne, and Anki, is to adapt to the user's mind and figure out this optimal interval to wait before showing a flash card again, as opposed to conventional flash card systems that review each card every day. Not only does spaced repetition greatly increase the value of each time you review a card by committing it further into your long term memory, but it also greatly decreases the number of flash cards you need to review each day.

The art of memory (aka method of loci), which is the technique described in Moonwalking with Einstein, works by taking advantage of the fact that our visual and spacial memory is far better than our memory for arbitrary facts. The gist of the technique is to convert data that you want to memorize into vivid images (the funnier or cruder the better) and arrange those images at discrete points in a predefined order along a spacial layout you know well (the layout is called a "memory palace", but it can be your home, school, office, neighborhood, favorite video game map, etc.). To recall the data, all you have to do is retrace your steps and remind yourself what each image means. Combining this technique with spaced repetition, you can commit thousands of pieces of data to a memory palace, wait a few hours to retrace your steps to recall the information, retrace your steps again the next day, then again a few days later, then again a week later, then a month later, etc. In this way, about two or three hours distributed over 5 or so weeks (maybe with one more review the next year) will let you memorize thousands of pieces of data permanently. The stuff you can do with your mind once you know more efficient memory techniques is truly amazing.

Thanks for the explanation, very useful!

Recommendations for what to do / read after checking out the book? It sounds like you've put some time and effort into this, so it'd be great if you could provide further recommendations.

To be honest, I know a lot more about spaced repetition than I do about creating memory palaces. My only exposure to the latter is from Foer's book, and while I've begun using my house as a memory palace to easily memorize short lists (15 or fewer items) [1], I haven't been ready yet to devote the time needed to memorize specific paths through hundreds of discrete locations in order to memorize longer things like speeches or several thousand Chinese characters. The best advice I could give you is read Foer's book first, and then you should know what elements of the art of memory you should pursue further. Oh, after you finish Moonwalking with Einstein, you'll probably want to start building memory palaces, and this post has a good suggestion for creating new ones: http://www.michaelcorayer.com/2011/03/31/resource-for-memory...

[1] Even without jumping into the amazing world of memory palaces, you can still take away a lot of little gems from the book, including seemingly obvious things that you probably never bothered to think of. For example, the author says, "Words that rhyme are much more memorable than words that don’t; concrete nouns are easier to remember than abstract nouns; dynamic images are more memorable than static images; alliteration aids memory. A striped skunk making a slam dunk is a stickier thought than a patterned mustelid engaging in athletic activity."

I'm having trouble figuring out what kinds of "vivid images" you would use for memorizing foreign language vocabulary words.

For example, I'm trying to learn German right now. It seems obvious that you would use images of the English translation in your memory palace, but then you would just have the problem of remembering the German word that each image signifies.

Or going the other way, from German to English, what images would you use in your memory palace to represent German words? How would you tie them back to English?

help?

Figuring out how to encode various classes of data into images is one of the more challenging aspects of the method of loci. Also, memorizing an associative array of data is less natural in the method of loci than memorizing ordered data since the way you recall the data is by tracing your steps through the memory palace in your mind. I guess you could place two images at each location, functioning sort of like a LISP alist (list of pairs where the first element in each pair represents a key and the second a value), but that still doesn't answer your question about how to create the images. Maybe go syllable by syllable? For example, if you wanted to remember that kochen is kitchen, you could visualize a macaw perched on the shoulder of Barbie's boyfriend while the latter flipped burgers on a toy grill. The "caw" sound in my dialect of English is identical to the "ko" in kochen, and Barbie's boyfriend Ken would be enough to trigger my memory of "-chen", so I'd be like "Right! This was illustrating that the German word kochen means to cook."

Spaced repetition lends itself a lot more naturally to memorizing foreign language vocabulary words than the method of loci, so you might want to use that approach instead. Within spaced repetition, you can also use vivid images as a sort of scaffolding to increase the odds you'll remember the word before the next review, and you can eventually let that image go once you start recalling it easily. The more connections you form to other thoughts, the more firmly a fact will be anchored in your mind, so you might as well use mental images if you can come up withthem, even in spaced repetition. James Heisig uses visual mnemonics to teach several thousand Chinese characters in Remembering the Kanji[1] (Chinese characters for use in Japanese) and Remembering the Hanzi[2] (Chinese characters for use in Chinese), and communities like AJATT[3] and Reviewing the Kanji[4] advocate combining Heisig's approach with spaced repetition.

[1] http://nirc.nanzan-u.ac.jp/publications/miscPublications/Rem...

[2] http://nirc.nanzan-u.ac.jp/publications/miscPublications/Rem...

[3] http://www.alljapaneseallthetime.com

[4] http://kanji.koohii.com/

Thanks. So I was understanding it correctly then, it is difficult to use this type of memory method with foreign languages.

I will stick with spaced repetition =)

I'm wondering if it might be easier for a language like Chinese or Japanese, perhaps, where the symbols and words are a bit more visually distinct.

That being said: doesn't Rosetta Stone use a technique that basically relates pictures to common words? I've never tried that program, and only looked at demos briefly, but I've heard great things about it.

> The only downside to spaced repetition is that most software implementations of it require daily review and don't work optimally if you miss a day

I don't think the efficiency penalty is that bad if you do it, say, every other day - I doubt the algorithms are precise down to the day.

http://www.ted.com/talks/joshua_foer_feats_of_memory_anyone_... TED Talk on the 2nd method you were describing. (At least I think it's what you were describing)

Exactly! He's the author of Moonwalking with Einstein, the book Bill Gates raved about in the parent article.

Oh haha, I didn't realize he was the author. Watched that video just a few weeks ago.