Distributed practice2 Distributed practiceDistributed practiceAugmenting Long-term Memory2020-09-24Journal
Ebbinghaus found that the probability of correctly recalling an item declined (roughly) exponentially with time. Today, this is called the Ebbinghaus forgetting curve:
What determines the steepness of the curve, i.e., how quickly memories decay? In fact, the steepness depends on many things. For instance, it may be steeper for more complex or less familiar concepts. You may find it easier to remember a name that sounds similar to names you've heard before: say, Richard Hamilton, rather than Suzuki Harunobu. So they'd have a shallower curve. Similarly, you may find it easier to remember something visual than verbal. Or something verbal rather than a motor skill. And if you use more elaborate ways of remembering – mnemonics, for instance, or just taking care to connect an idea to other things you already know – you may be able to flatten the curve out\\ Although this expansion is much studied, there is surprisingly little work building detailed predictive models of the expansion. An exception is: Burr Settles and Brendan Meeder, A Trainable Spaced Repetition Model for Language Learning (2016). This paper builds a regression model to predict the decay rate of student memory on Duolingo, the online language learning platform. The result was not only better prediction of decay rates, but also improved Duolingo student engagement..