Combining Spaced Repetition and Method of Loci for long-term memorization of information dense content (e.g books)

6 posts / 0 new
Last post
#1 12 February, 2015 - 12:26
Offline
Joined: 2 years 10 months ago

Combining Spaced Repetition and Method of Loci for long-term memorization of information dense content (e.g books)


Greetings fellow memorizers,

I want to submit (what I hope) is a novel method for the memorization of (what I like to call) information-dense content; this method is specifically for long term usage and retention.

I define information-dense content as any content which contains too many pieces of information to adequately memorize (by traditional methods). This more or less refers to textbooks or large novels. When I refer to traditional methods, I am not speaking of mnemonic techniques, but rather, simple rote memorization.

I won't belabor the point with discussions of the method's usefulness, ease, my own experiences, its necessity, etc. etc. I will leave that to the end, for those that are interested. Lets begin.

The Method

Synopsis
Spaced repetition is an incredibly useful tool for memorizing enormous quantities of information with little time investment (some calculations put it at ~2 seconds/card!). More importantly it is the only method which promises and delivers, long term retention.

Having browsed the forums, I've seen a number of interesting innovations to the method of loci (whether they are actually new or not, I can't say; nothing is new under the sun after all, even in the case of this method), of particular interest are Gavino's method (for giving a database like structure to the method of loci) and Bateman's method of memorizing books (using granulation to memorize to desired detail). I've also done the GMS course (only up to Your First Database, from which point everything else was merely supplementary and I ran out of vacation time) which also has a database structure similar to Gavino's.

My method involves the combination of these various techniques, Gavino's database structure, GMS, Bateman's book method, and spaced repetition to create a way to memorize large amounts of information, but more importantly, to retain that information for the long term.

Preliminary Necessities
The user will need:
1. Spaced repetition software (e.g Anki)
2. A database-palace
3. A book (or other information-dense content)
4. Some quiet time every day (around an hour or so)

Step-by-Step method
1. Create the database-palace: You do this by using a familiar (read: real and oft visited) place or journey assigning it with loci. It is very important that this journey is one you take often (even every day) because this will be the foundation for everything else. Once you have your journey, you will need 100 loci within it. This should not be too difficult for an average size journey/palace. The 100 loci must also be divided, in a logical way, into 10 loci slices.

e.g Your house has 10 rooms/parts and each room has exactly 10 loci. Or, your journey has 10 different "stops" (street, train station, transfer, bus etc.) and each stop has 10 loci.

This is to ensure that numbering your loci is easy and simple (if want to know the 31st loci, you just look in the third room or stop and the first loci in that room/stop) and does not require you to stream through all the loci one by one. This trick should be pretty well known, so I will not go into any further detail.

2. "Format" the database-palace: The point of the initial database is to create a structure from which the user can memorize any amount of information without increasing the database-palace's 100 loci. This is done by increasing the granular quality of the palace (I will explain this below). The first 20 loci of the database are essentially RAM, you can use them on the fly to memorize things for the short term (a phone number, a name, a grocery list, a speech etc.) and therefore can memorize "over" previous associations. The other 80 will be categories where you memorize specific information relating to that category (kind of like the catalog of a library). This information will be for the long term and therefore you must not memorize over them. You may create the categories however you wish, I recommend maximizing the loci as much as possible, meaning for every loci, you have a separate category. This way, you still have many loci left over for when you need additional categories.

It is important, to never "write over" an existing loci's association (unless you want to forget all the information, there contained).

3. Create a number of "mini-palaces" a la Gavino: You mini-palaces can be created any which way (I like the video game method myself, the movie method should work also), but must have two restrictions:
a. Always, always, always, have 10 loci per mini-loci (every single time!)
b. Never, never, never, place anything on the first and last (1st and 10th respectively) loci's (this will become clear shortly).

Create at least 80 mini-palaces of 10 loci each, 1 for each category-loci in the database. (You can also do an additional 20 for the RAM-loci, but it's not absolutely necessary.) You don't have to do them all at once (in fact, I recommend you pace yourself and do 10 a day over a period of a week or so), especially if you have not decided what all your categories will be.

You must now associate each mini-palace to its particular loci of your database-palace. You can do this whichever way feels comfortable (Having been through GMS, I prefer the simple image overlap but use whatever works), but make sure that the connection is clear and allows you to "enter" your mini-palace. The first loci of the mini-palace is then, in turn, associated with the loci of the database palace. Note that the database palace connects the mini-palace using an alternate image, but that the first loci of the mini-palace associates the database-palace's loci.

e.g Database loci- window -> associated with an old black car -> goes into mini-palace where the first loci is a steering wheel -> associated with the window from the database loci.

The last loci of the mini-palace is also kept free. This is so that a second mini-palace can be connected to the first. In this case, as before, the last loci contains an association to enter the second mini-palace, whose 1st loci is associated with loci 10 of the first mini-palace. In this way, an unlimited number of mini-palaces can be tied together under one database-loci. But we are far from done.

4. Fix your information into the palace: Each of the "free" (i.e not 1st and 10th) loci in the mini-palace is essentially a base for another memory palace of however many loci are necessary. I recommend keeping these palaces short (20-30 loci is enough for most purposes, e.g textbooks) as they will only serve as a skeleton for carrying the the meat of the information. As an example, lets take the textbook "Accelerated C++" which has 16 chapters. This method works well on textbooks, because the content is usually broken down into a branching hierarchy (e.g chapter, topic, paragraph, sentence).

To memorize the contents of this book, we use Bateman's method 4 (Title, Author, Chapters, Topics). Note that to memorize this book, you must have actually read it :P. I discourage memorizing word for word (though it is, of course, possible) so that the memorized material only helps you to recall what you should have already read. It is a crutch, not a leg.

a. We create a 17 loci memory palace (use whatever works, movie method, video game method, etc.) linking the first loci to the title
b. We transform each of the chapter titles into a series of 3 (or less) linked images. It is important to keep the image count less than or equal 3, to decrease the strain on your memory. You must link the images by chaining them (one on top of the other).
c. We fix the first linked image to each of the loci (now the chapters are memorized)
d. To memorize the topics/summary of the chapters: divide each image (from the chain of images) into three parts then fix the appropriate topics/images for the summary on these parts. You have space for 9 images, if more are necessary (usually they shouldn't be) you may form chains on each of the image parts (make sure these chains do not exceed 3 images, as before). This allows for 27 images.
e. Now to memorize specific details under a topic (sub-topic), do the same as above, this time using the image of the topic as the base.

Now an example with the book I mentioned:
(Title- Accelerated C++ --> A car filled with water associated with the first loci)

Chapter 1 - Working with Strings --> A spool of thread with a needle associated with the 2nd loci; has three parts: top part of spool, white thread, needle
i. Input -> Zoom into the top part of the spool (the flat part with a hole through the middle), associate the image of a computer keyboard
ii. Framing a name -> Zoom into the thread, associate a picture frame made out of nameplates
iii. Details -> Not necessary to fix, since it is a summary of the chapter and is in every chapter

Chapter 2 - Looping and Counting --> This chapter has 7 topics, so we will use 2 images for the chapter along with the sub-chain (though you could just use three instead): a racetrack, an abacus
i. The problem -> Zoom to one part of the racetrack (the starting line), associate an image to it (a banana hanging over a box)
ii. Overall Structure -> Link an image of a marble pillar with the previous image (banana with box below it)
iii. Writing an unknown number of rows --> Link an image of a bunch of pencils lined up in a row with the marble pillar
iv. The complete framing program --> Zoom into another part of the racetrack (a F1 car), associate an image with it (a gun with fingerprints on it)
etc.

To memorize sub-topics or details, simply zoom into any of the images for the topics, divide them into three parts and chain images on to them. Again stick to three images (though I believe 5 can also work).

5. Retain information for the long term: For the most part, frequent use and recall of the information will be enough to fixate the information into the long term. Unfortunately, some information is not encountered enough in everyday life for this to be possible. Thus we need to have a systemic recall schedule to ensure efficient retention. We do this by using spaced repetition. The flashcards will have two numbers on the front representing a particular loci. The first number is from the database loci (ranging from 20-100), the second is the mini-palace loci (ranging from 1 to some multiple of 10). On the back of the flashcard will be a short description of the data memorized in that loci path, as well as the memory palace.

e.g Front - 31.6 Back- Book: "Grapes of Wrath by John Steinbeck", Palace: Office from CS:S

The user simply moves to the given loci, recalls whatever he can, then rates himself based on how difficult/long it took him. The program then schedules a review of that information accordingly. When new information is memorized, it is a simple matter to create new flashcards.

Discussion

This method is more aimed toward people who want to memorize a diversity of information rather than one particular book or textbook with extensive detail. Again, I must stress that the method is only a crutch, not a leg. You have to actually read the book, while or before you memorize it. I think the structured nature of this method and its expandable nature make it a good choice for a holistic memorization system.

I wrote the description of the method both to share it with you all, as well as to put it down in writing for my own reference. I hope it helps some of you, and I'd very much appreciate your feedback. Thanks.

3 December, 2016 - 02:26
Offline
Joined: 1 year 3 weeks ago

Quote:

8 December, 2016 - 07:08
Offline
Joined: 5 years 1 month ago

Thanks a lot for sharing this method !

I intend to adapt your method to store anything : your categories are general areas of knowledge, and the link between loci are organised to get a multi-nested structure.

I wanted to find an intermediate between Gavino's MMP (for its granularity) and Mayarra 1.000.000 loci palace (for the category-based approach). But, to my understanding, Gavino's MMP is not made to be indefinitely scalable and Mayara 1.000.000 loci palace, while providing many loci - as its name suggest - is a bit static and may give me problem of extensibility and maintenance. This method sounds highly dynamic making it suitable for a long term usage

More specifically :

- I like the idea to have many precise categories. If some are missing, categories may be added by extending the main journey.

- Using anki with the numerical index of the palace is quite smart, I like it.

- The RAM journey is also something I wanted to implement, but I see it more as a separate journey than within the main one.

Quote:
You must now associate each mini-palace to its particular loci of your database-palace. You can do this whichever way feels comfortable (Having been through GMS, I prefer the simple image overlap but use whatever works), but make sure that the connection is clear and allows you to "enter" your mini-palace.

"Image overlap" ? I am not sure whether I know this technique. You are somehow merging your loci ?

8 December, 2016 - 04:16
Offline
Joined: 1 year 9 months ago

By my experience, it isnt so neccesary having the informatiin so structured if the exam is by test, or by problema.
Instead of it, you must be sure to recall when you still remember it, and point when you dont, so I think that using the minute/hour/day/week/month rule is better to get it on long than rewriting everything on flashcards. Furthermore, you hace memorized it on memory palace so, you dont need flash cards to recall. At most, review the weak links between memory palace and issues.

8 December, 2016 - 05:05
Offline
Joined: 2 years 6 months ago

Quote:
I am not sure whether I know this technique. You are somehow merging your loci ?

I believe he means that he takes the same approach as in Gavino's MMP, where he has a 'parent memory palace' of which each loci leads to another memory palace. :)

8 December, 2016 - 08:12
Offline
Joined: 5 years 1 month ago

Mayarra, I was more specifically interested by what he meant by "image overlap". I edited my post for more clarity. Do you think it's the same kind of link as in gavino's MMP ?

Download our free ebook! Just click the "Sign up" button below to create an account, and we'll send you a free ebook with tips on how to get started.

Related content: