Thursday, May 13, 2010

When repetition does not improve memory

The above is the title of a very interesting post on TheSuperMemoryBlog, a blog written by a fan of the spaced-repetition system SuperMemo named Gérman. Although he's using SuperMemo, his post applies to any spaced-repetition system.

The crux of his post is this:
[S]ometimes repetition is ineffective in promoting learning. … [U]nder certain circumstances … [repetition] not only doesn’t help, but could even impair your recall.
The first such circumstance is repetition without attention or more elaborate processing of the material. He begins with the example of Americans and the U.S. penny.
[M]any of us [Americans] have seen [the penny] enough times to know every detail on it, yet we only pay attention to its brownish color to differentiate it from other coins and thus that’s the most we remember about it. If I were to give you an old roman denarius, and tested you a week from now about details on it, you would be better at it than what you can currently remember about a penny (it would be the same with the currency of any other countries than the USA).
Is this example accurate? For you Americans, try to recall what's on the penny, and then compare it with what you find here. For non-Americans, try to do the same with a very common coin or bill in your country. How'd you do?

I did a lot better than his example would suggest. In trying to recall the Lincoln Memorial penny that was produced from well before I was born until 2008 (I don't recall ever seeing the newer union shield penny), here's what I could recall before seeing it at the link above: Lincoln's profile on the front (I couldn't recall the direction he was facing), some words above him, the year to the right of him, the Lincoln Memorial on the back, some words both above and below the Lincoln Memorial, and, on both sides, a raised ridge around the circumference. I could also recall that the outer edge was smooth, as opposed to the ridged edge on the dime and the quarter. I guessed that some of those words that I couldn't recall were "In God We Trust", but I can't fairly say that I remembered that. If my recollection is typical, I'd say that his example isn't the greatest.

That said, the point does not seem to be an unfair one. And he backs it up with some info on rote memorization:
In many experiments, where participants have to repeat items aloud over and over[, w]hen they are given an unexpected memory test on the rehearsed words, there is almost no relationship between the number of overt rehearsals devoted to an item and later memory. Simply repeating an item over and over has little benefit for memory in the absence of attention or more elaborate processing of the material.
Remember all those rote repetitions you did back in your language classes? Yeah, that's about how useful they were.

The second such circumstance that he sets forth is when you've got to memorize multiple sets, say, Set 1 and Set 2, and each of those sets have multiple but overlapping items, say, A, B, C, and D in Set 1 and C, D, E, and F in Set 2. If you study Set 1 first, you'll have more trouble recalling that Set 2 also contains C and D because they are also included in Set 1. Why? Because "[i]f an item has been presented in several contexts, it may become difficult to retrieve the occurrence that is being tested". That makes the earlier repetitions of Set 1 detrimental to the later repetitions of Set 2.

How to resolve that problem? Gérman suggests using associative techniques. I'd put those items together; seeing a graphic of Set 1 right next to Set 2 would make it easier for me to associate and remember the differences.


  1. I think it's more the kind of repetition than the repetition itself. If I hear a word said in a tv show many times, it's much easier to remember than one I learned just by studying in an SRS.

    When learning sets of words, and trying to remember the set, it doesn't surprise me that muddying the sets makes it harder to remember them. You aren't learning the words, you're learning the SET. Which should be obviously wrong to anyone.

    I've had a lot more luck using sites like and that give you the word and the word in context and ask you for information about it. (For ReadTheKanji, that's just the pronunciation, but Renshuu asks much more.) Just studying from an SRS or wasn't nearly as useful and I had a hard time remembering the terms even between sessions. (Which is odd, because tries to do the same thing as those other 2, with the only difference being that it keeps the learning to a set of 10 words, instead of mixing them in as needed.)

  2. I don't know that the penny is a very good example. Being around something is not the same as actively trying to study and review something you want to know. If you cared about the details of a penny, put them all into your SRS & studied it regularly, I can guarantee you'd be pretty familiar with the features. Asking someone to describe the details of a penny out of the blue would be like quizing them on a book they've had on their shelf for years but have only flipped through once or twice.

    Likewise ~ classroom repetitions are not generally covering things the students care about learning. At most they want to learn the facts for a test, but don't intend to use them after that. Yet the same student could bombard you with details about an artist/celebrity/sports team they are interested in. Every card someone enters into their SRS may not hold quite the same level of fascination, but generally cover information that the person wants to learn bad enough to devote years of self-study to.

  3. Jonathan MahoneyMay 14, 2010, 4:31:00 AM

    I agree w/ Tharvey. I have never deliberately studied any currency, whereas when I work with SRS I'm always alert to every detail. I don't think it's a great comparison.

  4. Learning words is not acquiring a language. The kind of repetition that helps you acquire a language is repetition of new items in context. Flashcard programs won't do that.

    If you want a highly repetitious classroom-based system that covers items that students are interested in, check out TPRS. There are also teachers who do TPRS tutoring via Skype (I teach Chinese this way, for example). Principled repetition (what I called Optimized Immersion) works a lot better than the traditional methods out there. Comprehensible input is the basis for the brain acquiring a language; making the CI interesting is the job of the teacher and the student together.

  5. @William: I completely agree that different kinds of exposures to a word result in different memory strengths. More passive exposure, such as an SRS repetition, are weaker, while active exposure, such as you saying or writing the word in context, do a lot more to get the word in you head.

    I've actually got Anki set up to include the sentence from which I got the words I'm learning. One thing I like a lot about Anki is how customizable it is, although achieving such customization is generally not very intuitive.

    @Terry: Learning words is not acquiring a language, but learning words is key to acquire a language. What do you do when you don't get enough reps of words in context? That's where an SRS comes in.

    Based on what I've picked up from your other comments on this site and a few basic Google searches, it certainly does seem to me that TPRS would be much more efficient that the traditional classroom based system. That said, TPRS + SRS would be even more efficient. Outside of your system of principled repetition, do you make systematic use of SRSs?

  6. Jonathan MahoneyMay 15, 2010, 6:41:00 PM

    Sounds more to me like Terry doesn't know what SRS is.

    One can and should make cards in w/ context.

  7. @Jonathan: I'd bet that she does know what it is, but it sound like her method doesn't make use of it.

    And I fully agree; adding context to items in your SRS is a great way to make things easier to remember.

  8. Well, we do use SRS in a sense, since TPRS and the input-based methods emphasize "spiraling" back and constantly re-using words and items that have been acquired, and continuing comprehension checks until we're sure everyone really really really has those items solidly. But it's not an algorithm-based process; we're human teachers doing the repetition and the spacing in this case.

    But I'm talking about activities geared to acquiring a language, not activities like flashcards primarily suited to memorization of lists. And that's because even making cards in context is not going to be as effective as having a skilled speaker of the language present the language to you in meaningful pieces that you can understand. That's what the brain needs to acquire the language and put all the pieces together to be able to use it automatically, without thought.

    I do encourage my students to make use of a good flashcard program that uses SRS. I've been using SuperMemo for PalmOS for years, and transitioned to using the flashcard program from (I don't get paid by him, but his support has been astoundingly good and he's willing to implement new features users ask for) on my iPhone. In fact, I post ready-made sets of audio-equipped flashcards for my students for download after each session (if they're individual learners) or for anyone in the class if it's a class. Only problem is that for classes, not everyone has the same technology available. But it's a start.

    Don't get me wrong -- for rote memorization, I swear by SRS. That's what got me through interpreting school, memorizing chunks of my native-speaking classmates' language for me to appropriate wholesale and trot out later to sound good. ;-) But man shall not acquire a language (as efficiently as possible) by flashcards alone. That's all I'm saying.

    And since many folks haven't heard of the input-based methods like TPRS, I like to put the word out there and let folks know that there are people teaching that way if they are so inclined -- or even that they could look into the basis for the method and pick the "active ingredients" out for their own use, if they're forced to self-study, as I am right now for the language I'm currently working on, where there are no speakers available.

  9. It looks good,I have learn a recruit!
    Recently,I found an excellent online store, the
    are completely various, good quality and cheap price,it’s worth buying!

  10. When learning a new language and repetition isn't working, you should try visualization, or creating a mental picture of the word you are trying to learn. There's plenty of improve your memory information out there that is helpful for language learners if you look for it. Good Luck!