Through this guest blog at the Learning Scientists, I found a subtle but clever addition to the memory model that Daniel Willingham, among others, made popular in a simplified form. This is the version by Daniel Willingham:

This is obviously a lot older, going back to Atkinson & Shiffrin’s 1968 model, among others:

I found this new version at the Learning Scientists, created by Rob McEntarffer, Bethany Brunsman and Sarah Salem.

What I like about it is the arrows under the three memories describe, more specifically, what happens to the information when it goes wrong. Something that does not go from your sensory to your working memory is gone. Something not encoded in your working memory will not end up in your long-term memory. Finally, it can go wrong if the information is not correctly extracted from your long-term memory. As Casper Hulshof noted in a reaction to a Dutch version of this blog: don’t forget that forgetting can have important functions, which is true!

I also have some reservations about the model, for example, about those 7 chunks in working memory they describe. This is really too optimistic.