Skip to content

Core scientific guidelines

Here are the key scientific assumptions (and associated references) that are guiding the development of SARAR:


  • Why practice memorizing things?
    • "Working memory — which relates to the amount of information that can be retained in short-term working memory when engaged in processing, learning, comprehension, problem solving or goal-directed thinking. Working memory is strongly related to a person’s ability to reason with novel information (i.e., general fluid intelligence)." (see Learning strategies: a synthesis and conceptual model)

This justifies why SARAR uses the spaced repetition technique, and why the initial default time interval between learning sessions is one day long.


I did not found a lot more evidences about this (help would be appreciated here). Therefore, in SARAR, active recall is recommended and proposed by default. But passive recall is also optionally available.


  • A flashcard system is a very convenient way to experience spaced-repetited learning sessions with a software. I did not find any meta-analysis revealing the superiority or inferiority of this system compared to another one. Unless someone find some indisputable arguments (meta-analysis based) against it, this system will remain the one used in SARAR.

This justifies why a maximum interval time between sessions is configurable in SARAR (per deck of cards).


This justifies why the evolution of the interval time between sessions (fixed, expanding or even contracting) is configurable in SARAR (per deck of cards).

However, while longer intervals are not necessarily better for memorization, they are much more convenient for learning more material per session. Therefore, this will be the default approach.


I am not aware of any meta-analysis arguing the superiority of any equation in order to approximate how information is lost over time when there is no attempt to retain it. Knowing how fast information is forgotten would be interesting in order to find when to schedule reviews. But, without any consensus about a model/equation describing this process, SARAR will consider that it depends on the individual and on the complexity/subject of the learning material.

Therefore, the evolution of the time interval is configurable in SARAR (per deck of cards) with the equation you want.


If you think this section does not properly reflect the current scientific consensus, then, based on the best meta-analysis you have found, let's discuss it (please open an issue) and let's improve SARAR together.


References

Learning strategies: a synthesis and conceptual model

This article from Nature (from august 2016) is a synthesis of 228 meta-analyses about learning strategies. It addresses topics well beyond the scope of this project (not just "remembering things"), but it really is a must-read if you are interested in learning strategies.

Visible learning: A synthesis of over 800 meta-analyses relating to achievement

This book (from november 2009), as its name declares, is a synthesis of 800 meta-analyses about learning. The scope of this book is beyond the one of this project (not just "remembering things"), but it is a major reference if you are interested in learning.

A Meta-Analysis of the Spacing Effect in Verbal Learning : Implications for Research on Advertising Repetition and Consumer Memory

Distributed Practice in Verbal Recall Tasks : A Review and Quantitative Synthesis

A Meta-Analytic Review of the Distribution of Practice Effect : Now You See It, Now You Don't