Retrieval Practice and Generative Learning: Classroom Implications
Jonathan Firth's Memory & Metacognition Updates #122
Hello!
Today’s post follows on from last week’s, in which I discussed research by Obergassel et al. (2025) on retrieval practice and generative learning.
Today I’d like to flag up a couple of other interesting issues arising from the paper, as well as considering what the findings mean for educators.
Prior knowledge
As has been mentioned a few times in these updates, prior knowledge makes a difference to new learning. Students in the Obergassel et al. (2025) experiment were given a pre-test of their knowledge of the textbook content, and also asked to report their high school graduation grade. High school grades correlated with task performance, which can mainly be put down to stronger students doing better in both contexts.
In terms of prior knowledge, it appeared that almost none of the participants were familiar with the concepts studied in the experiment. This is fairly typical for lab research (researchers often want to present previously-unseen tasks), but is worth bearing in mind when interpreting the results, given that teachers and students often use evidence-based techniques on material which they have seen before.
Can we implement the findings?
It’s worth highlighting that this is just one study, and normally I would be wary about applying such findings immediately.
However, the individual techniques are both well established. This study suggests that we should make space for both, rather than focusing too much on just one or the other.
This doesn’t feel like a risky proposition. I’d have been more wary if, for example, the researchers had insisted that generative learning always needs to be done before retrieval practice (or vice versa). But it appears that the order does not matter very much.
I’d therefore suggest that it’s a fairly safe bet to incorporate both understanding-focused and memory-focused strategies into classroom learning.
Conflicting findings?
The researchers do highlight another study – O’Day and Karpicke (2021) – which did not find a benefit of generative learning over and above retrieval alone.
I actually mentioned that study way back in update #23, in a post on study skills. At the time, I connected the finding with a concern that having students engage in mind mapping is too often ‘busy work’; the learning is more likely to stick if they have to retrieve ideas from memory.
I largely stand by that view. Still, it should be noted that mapping is one of the eight generative strategies recommended by Fiorella and Mayer (2015), who say (p. 722):
“generative learning is fostered when learners translate from linear text to spatially arranged text that highlights the structure of the lesson.”
So, what’s the difference compared to the more recent study? One explanation could be that the task used by Obergassel and colleagues – generating examples of concepts – is superior to concept mapping. Perhaps not all generative tasks are equal! Generating examples could plausibly require deeper or more active processing. Another explanation, favoured by Obergassel et al – is that the O’Day study didn’t implement the generative task effectively enough for it to help.
No doubt further research will explore this issue further, but it is a reminder that whatever evidence-based strategy we choose, we have to do it in a sufficiently effective way to reap the benefits.
Putting it to use
Despite their differences, the studies mentioned above both broadly support the view that retrieval practice is important for consolidating long-term memory, while generative tasks focus on understanding.
While educators rightly pay a lot of attention to how well students understand concepts in class (for example via formative assessment strategies), we should not focus only on this. After all, students could understand a lesson well at the time, but later forget everything that was covered.
I’ve actually said quite a lot about this in CPD sessions, talks and blog posts.
No matter how important it is to check for understanding, learning will not succeed if we don’t work to ensure that concepts and skills are retained over the long term. Although well-understood material will be forgotten more slowly, it is not immune to forgetting.
A good example is our retention of the plot of a movie or novel. These are certainly well understood at the time, but can easily be forgotten after a few weeks (or years).
You can see the image above as an analogy – focusing only on understanding is like only thinking about building and never about maintenance. To achieve the goals of any lesson or course, we need to ensure that learning lasts.
That is all for now. Have a great week!
Jonathan
Last time: Combining Retrieval Practice and Generative Learning
Please note that my slides and similar materials are shared under a CC BY-NC 4.0 license. This means you can use or adapt them with attribution for non-commercial purposes. If you wish to use my materials for other purposes, feel free to get in touch.