A recent paper by Kelley and Whatson [1] describes an approach to learning which uses short-term spaced repetition to better exploit the information encoding and LTM activation processes. They advocate a pattern of 20 minutes study, 10 mins break doing something very different (i.e. physical), then two sessions of 20 mins review separated by another 10 min break.
I found out about the work here [2].
Has anyone heard of this and/or experimented with an SRS like Mnemosyne, modified to review the same day/session's items in a scheme such as that described by Kelley and Whatson? It could probably be done via a plugin. A couple of spaced repetition systems do something almost like this by default; namely Pauker and Memrise.
Since it's very well accepted that spaced repetition is far superior to cramming, it'd be nice to see a study comparing the effectiveness of learning under three conditions: 1) control, 2) classic SRS (with a minimum resolution of 1 day) and 3) classic SRS supplemented by the study->distraction->review->distraction->review pattern described in the paper.