This post covers the theory behind Anki and follows the previous post introducing Anki. The last post in this series describes best practices for using Anki.
One of the best meta-analyses reviewing best approaches to learning is Improving Students’ Learning With Effective Learning Techniques by Dunlosky et al. If I were to summarize its 50+ pages in one paragraph, it would be: The most utility for improved learning is provided by distributed practice and practice testing, followed by interleaved practice, elaborative interrogation, and self-explanation. In contrast, popular techniques such as highlighting and rereading provide only low utility for learning. The meta-analysis was done in the context of students’ learning, but the top techniques were shown to benefit learners across varying ages and abilities. This makes these techniques applicable to lifetime learning.
Anki implements the above mentioned top two techniques directly and provides a way for the third most useful practice. Let’s examine these three techniques in more detail:
- practice testing (a.k.a. active recall): memory needs to be actively stimulated during the learning process, which is what Anki does by showing questions that the user must answer.
- distributed practice (a.k.a. spaced learning): learning is broken up into several short sessions over a long time. Anki schedules flashcards for review based on an algorithm over a long period.
- interleaved practice: mixing multiple subjects or topics while studying improves learning. Anki cards are organized in decks and Anki encourages users to limit the number of decks being reviewed. Since decks are reviewed in succession, ideally all the Anki cards are within a single deck to ensure that knowledge from different subjects is reviewed at the same time.
Another great source of the theory behind Anki is a lengthy article by Michael Nielsen titled “Augmenting Long-Term Memory,” where the author lists many historical underpinnings for memory augmentation. It is surprising that as early as in 1885, some scientists studied memory decay, see “Memory: A Contribution to Experimental Psychology” by Hermann Ebbinghaus. For a modern take on memory decay, read “A Trainable Spaced Repetition Model for Language Learning” that describes an algorithm used in Duolingo, an excellent app for learning foreign languages.
At the end of his article, Michael Nielsen goes through a rough estimate of the effort needed to remember one piece of information. It assumes that the user takes just a few seconds to recall a fact listed on an Anki card when it is scheduled. If her effort is spaced over twenty years in ever increasing intervals, the total time spent on the card is about 5 minutes. This sounds like a decent trade-off: spending a total of 5 minutes over twenty years will result in remembering one particular flash card. In other words, it will take an average of only 7 minutes daily to remember 10,000 flashcards for twenty years!