Short term memory spaced repetition

I have came up with the idea of why you are only able to hold in your working memory few word, and the answer I suppose is simple: You forget the word if you don’t review it. For example: Today I was reviewing one word I didn’t remember, and in less than 10 seconds I guess I forgot it. I guess that is because the harder a word is for working memory, the shorter is the span in which you got to review it. For example nesoverchenoletni (I do not want to search the Russian word), that means minor. Let’s suppose that after you see it, the threshold for forgetting it is 10 seconds, then in 10 seconds you review it and the span becomes 15, then 30, then 50, 140, etc… Now a word like tak (“that” in Russian) may have a span of 30 seconds of even more, so now I have one question and one insight.
Question: Do you have any comments on that?
Insight: Making spaced repetition systems that are based on short/working memory. The minimal span that Anki has, for example, is “10 minutes”.


It’s a good insight - a word like nesoverchenoletni (7 syllables) takes up almost all of the phonological loop (the part of working memory that you’ll typically use for remembering this task) and so it is slower to handle.

More relevant perhaps is that the word simply contains more information than “tak”, and so you have to make more links in your brain to remember it (the rhythm of saying it, relation to other similar words in Russian, familiar substrings like “over”, or whatever comes to mind).

Modern spaced repetition software algorithms try to determine how difficult an item is, and this changes the frequency at which they show you the item.

For example if you can’t recall the item confidently on the first few tries, it will increase the “difficulty” rating of the word. On Mnemosyne you can even show a bar charts showing these difficulties, although it isn’t a feature I have paid much attention to.