Epistemic status as of 2021-11-22: sketchy, currently based only on initial reviews in QCVC
The data suggest that when we schedule a review too late, readers may be hit with two negative effects at once: they’re less likely to remember the question in that first review, but also, they’re less likely to recover from forgetting in the following session.
If true, this makes the Spacing effect something of a knife edge: you benefit by delaying practice, but only so long as you still remember the answer. This isn’t really a surprise: I’m sure there’s laboratory experimental data on this; but the matter is discussed in the models in Mozer, M. C., & Lindsey, R. V. (2016). Predicting and Improving Memory Retention: Psychological Theory Matters in the Big Data Era. In M. N. Jones (Ed.), Big data in cognitive science (pp. 34–64).
Using the data from 2021-04 Quantum Country schedule experiment, we can look at questions which are:
The second session will already feature a substantial forgetting curve in this scenario (see also Quantum Country users who forget in-essay exhibit sharp forgetting curves):
Third session accuracies, by second session latency, assuming forgotten in second session 20211123103658:
Another way to put this, which emphasizes the compounding nature of late scheduling: how often do people forget both in the delayed recall test and in the recovery session scheduled one day later?
And now, what about a fourth session, assuming forgotten in second and recalled in third?
Now let’s look at questions which are:
First session accuracies 20211123164928:
Second (“recovery”) session accuracies one day later, by first session latency 20211123163958:
Not an enormous effect, but it’s probably real. The difference between the effect size for initially-forgotten and initially-remembered traces is pretty large.
How often do people forget in both the first session and in the recovery session?
I’m not sure this is a large enough difference to be worth worrying much about.
If they succeed in the recovery session, how do they do in their third session, which will be delayed by the same period as the first session? 20211123165631
Not an obviously large effect.