Spaced repetition is a more robust method memorizing large amounts of material than more fashionable accelerated learning techniques. You can apply it to learning just about anything. I’ve used it to learn a programming language, shortcut key sets for the software I use, retaining information in new articles I read online, and a host of other subjects.
How Spaced Repetition Works
The spaced repetition system (SRS) is a advanced form of rote memorization that uses electronic flash cards in an algorithmically determined sequence. With traditional flash cards, you work through a deck, one card at a time, until you reach the end. More astute learners will sort through the cards at the beginning of each session to eliminate the items they already know well, so they can focus their studying on the material that needs it most. A spaced repetition program automates this elimination process, but that’s only one advantage.
One of the most important findings in memory research is that forgetting happens exponentially, following a predictable curve that varies from item to item. An item that you remember well has a flatter forgetting curve than an item you remember tenuously.
Suppose you tested yourself on two facts you learned five minutes ago. You recall the first fact without any effort, and give yourself an “A”. You recall the second fact, but only after putting in some effort, so you give yourself a “C”. The forgetting curve would predict that you would completely forget the “C” fact days before you would completely forget the “A” fact.
That seems like common sense, but it points to a huge improvement opportunity in traditional flash card drills. Repetition spacing schedules an item for review right before the forgetting curve predicts you would have less-than-perfect recall, and deliberately avoids scheduling a review in the interim. This not only minimizes the number of items you need to review, but also minimizes the number of reviews for each item.
SRS programs ask you to score your recall of each card on an A-F or 1-5 scale. Items you recall with no effort get the top score, items you miss completely get the bottom score, and the scores in between correspond to ease or difficulty. The higher you score each item, the less frequently it appears in your daily flash card sessions. The review schedule varies for each item, based on the recall score assigned to it. You only spend time reinforcing the items that actually need reinforcement.
If I were learning Spanish vocabulary, and “inteligente” for “intelligent” came up, and I graded my recall for it “A” on first two or three exposures, it might literally be years before the system queued it for review again. An item with consistently high scores has a flat forgetting curve, and needs virtually no reinforcement. It’s already part of my long term memory. A word I’m less familiar with, like “aburrido” for “boring,” would receive a “B” or “C”, and I might be tested on it again in the next day or two.
Actually using a spaced repetition program instead of conventional flash cards can be strange at first. You have items to review every day, but only a fraction of the topic database. When you complete a session, you might think to yourself, “Is that it?” You don’t get to decide which items you’re going to review in a session. The algorithm determines the schedule.
Here’s the cool part. Because you only review a fraction of the database each day, you can add more items if you feel insufficiently challenged. Instead of accelerating how fast you learn individual items, spaced repetition accelerates the number of items you can learn at once. You can organize your material into topic specific databases if you want to isolate the content of each study session, or you can throw all items into a single database.
I have a catch-all database full of information I find in articles I read. If I read an article on using Google’s search operators, I might create a new card for each operator (Question: “What is the operator finds a term in a domain name?”; Answer: “indomain:”) and add it to the database. Some articles only have one or two items worth remembering. others have more. Any articles with no facts worth remembering probably aren’t worth reading in the first place.
Using Spaced Repetition Software
Most software for repetition spacing works in a similar fashion, despite differences in layout and terminology. You create a database for each topic, like “Spanish Vocabulary,” create optional categories for each database, like “Adjectives,” then create Question/Answer pairs for each item you want to learn. Q/A pairs can use yes/no questions, open-ended questions or fill-in-the-blank statements.
After you create your items, you need to “commit” them. Some programs have a “drill” mode that will test you on each item and repeat the items you answer incorrectly until all of them are answered correctly. Drills don’t offer grading on how well you’ve learned each item; they’re only designed to ensure initial exposure to all items. Then you commit them to the algorithm. The first time you use “test” mode, you’ll be tested on and grade all items. The next day you’ll only see some of the items, the next day you’ll see other items, and so on.
In most cases you’ll create Q/A pairs directly in the program, but most programs have other methods of creating these items: flat text files with line items beginning with “Q:” and “A:”; spreadsheet files with the questions in the first column and the answers in the second column; clipboard utilities that let you copy and paste questions and/or answers from web pages. Many developers of these apps will have free libraries of preexisting databases on their website: periodic tables, foreign vocabulary, MS Office shortcuts, and dozens of other topics.
The first SRS software I used was SuperMemo for Palm PDAs, which was limited to text files. Most current apps, on the desktop at least, allow you to import image and sound files. If you’re learning to recognize anatomy or spoken vocabulary, support for these files is essential.
The desktop version of SuperMemo is the most famous SRS app available, but many of its once-unique features, like support for richer media files, have been duplicated by commercial equivalents, like FullRecall, and free/open source ones like Anki and Mnemosyne. I actually prefer the simplicity of Mnemosyne and use it regularly, even though I paid for SuperMemo, which has rather dated and cluttered interface.
Incremental Reading
The one feature SuperMemo does have, which I have yet to use enthusiastically, is support for incremental reading, a superset of the spaced repetition concept. In incremental reading, you import an entire article into the program and read it through once. Then you grey out sections of text that you don’t need to learn and commit the remaining sections for reviewing individually, according to SRS scheduling.
As with traditional SRS, incremental reading doesn’t allow you to read material faster, but it does allow you study and memorize more material at once. Obviously, you would only want to do this with material you were going to study anyway, not casual reading. Personally, I’d rather take notes, distill them into Q/A pairs, and use the flash card method of SRS than repeat larger blocks of text.
Have you tried any SRS software? What has your experience been like?
spaced repetition, spaced repetition learning, spaced repetition software, spaced learning technique, repetition and learning, repetition learning, spaced repitition, learning by repetition, does spaced repetition work, accentuate software
Discussion