Paul A. Kirschner & Mirjam Neelen
In the coming weeks, we’ll be publishing several blogs on generative and productive learning strategies. In short, generative learning isn’t just ‘being actively involved with the subject matter’, but rather doing something with what you have to or want to learn. It’s generative when you produce new things like making a concept map or a drawing about the subject matter. In this introductory blog, we discuss two studies examining the effectiveness of two generative strategies – explaining and producing questions.
In traditional learn-by-explaining research, explaining as a learning activity is usually about learners giving “instructional explanations” to their peers, with the explicit intention of giving them some sort of interactive lesson. Sometimes it involves learners at the same ‘learning’ level and other times it’s done with learners who are at different levels. If the latter is the case, older pupils who are one or two classes higher or who are more advanced from the same class teach their younger or less advanced peers (so-called peer teaching or peer tutoring). The idea is that the learners who receive the instruction from their peers benefit from that lesson and that the peers who teach benefit from that because they have to process the material (more deeply) in order to be able to give the lesson.
Peer tutoring (Source)
Building on this, it was thought that even without real interaction with peers, ‘giving explanations’ – as in merely producing/generating them – could be useful in and of itself. In this situation, learners are asked to give oral explanations to an imaginary/fictional (i.e., not a real) peer, for example by explaining aloud to themselves at home or recording explanations as a video. In general, the results were mixed; sometimes this approach worked, sometimes it didn’t. The benefit of explaining something to yourself or via a video to a fictional other is easy to organise; you don’t need anyone and almost every child in secondary education at least, has a mobile phone with a camera (actually it’s more a mobile camera with a phone integrated into it J). Explaining to a ‘real’ other student, on the other hand, often isn’t easy to organise in a classroom from a practical point of view (e.g., there’s not enough time, it makes the classroom too noisy, the lessons of older and younger pupils don’t run parallel…).
A question that arose by others is whether the often positive findings of actually giving an explanation to another would also work if the explanation was written and not oral. A positive aspect of giving written explanations is that by having to write the explanation down, the learners are forced to list their ideas neatly and in an orderly manner. The down side to this approach is that writing down an explanation makes additional cognitive demands, as learners have to add a certain (rhetorical) structure to the explanation while writing and also use specific grammar and syntax which can hinder learning (as the explainer needs to focus on something else than what needs to be learned; kind of like the problem with discovery!). Spoken language is easier than written language!
Lachner, Jacob, and Hoogerheide (2021) investigated what they call learning-by-writing. It involved:
- giving written explanations to an imaginary peer (instructional explaining),
- writing down explanations for themselves, (self-explaining) and
- written retrieval (retrieval practice).
All three are considered generative or productive learning strategies. They conducted both a lab and a field experiment in the classroom. There was also a fourth group that had to make a puzzle and therefore did nothing with what had to be learned (in other words a real control condition).
In the lab, they found no differences between the groups when it came to learning concepts (i.e., knowledge), making predictions, or applying the learned concepts in other situations (i.e., transfer). This suggests that engaging learners in additional ‘explaining activities’ isn’t more effective than retrieval practice or – in this case – making a puzzle.
In class, again they found no difference for knowledge but a big difference in transfer; the self-explainers did much better than the imaginary interpreters, the practice testers, and the puzzlers.
This difference is quite remarkable. Why would there be a difference in results between a lab and a classroom experiment? The authors suggest it might be because in the context of the first (lab) experiment, the participants didn’t have any medical prior knowledge. Perhaps, so they suggest, the learners were overwhelmed by the task of learning from a medical text. This is speculation of course, but it’s plausible, also given the fact that learners gave a high ‘mental effort’ rating on the task and they also showed low test performance.
In another study, Ebersbach, Feierabend and Nazari (2020) compared learners who produced questions themselves with learners who (1) answered questions asked by another (practice / self-assessment) and (2) re-examined/reviewed the material.
Self-explanation example (Source)
After a week, they took at a surprise test testing both remembering facts and transferring what they learned to other situations. In other words, in the period between learning and testing, the material was neither practiced nor studied. The results revealed that generating questions and self-tests both had a positive effect, but studying the material again didn’t; a result in line with most other retrieval effect experiments. This applied to both types of knowledge (factual knowledge and transfer). In other words, the generative strategies led both to better memorisation and application of the material.
Overall, we can conclude that generative learning generates learning!
PS. In the coming weeks, we discuss several generative and productive learning strategies in more detail…
Ebersbach, M., Feierabend, M. & Nazari, K. B. B. (2020). Comparing the effects of generating questions, testing, and restudying on students’ long‐term recall in university learning. Applied Cognitive Psychology, 34, 724–736. https://doi.org/1 0.1002/acp.3639
Lachner, A., Jacob, L., & Hoogerheide, V. (2021, online). Learning by writing explanations: Is explaining to a fictitious student more effective than self-explaining? Learning and Instruction, 74. https://doi.org/10.1016/j.learninstruc.2020.101438
10 thoughts on “Generative Learning Generates Learning”
Thank you for once again providing very welcome clarity on a range of topics. With regard the difference between oral and written self-explanations, I’d like to introduce the work of Merlin Donald (The Making of the Modern Mind, 1991) who said that, in cognitive evolutionary terms, when the Greeks wrote down their thoughts, they introduced a new phenomenon that changed our cognitive architecture. Being now (semi)permanent, these thoughts are no longer transient — as are verbal accounts — and therefore set up a loop enabling recursive thoughts. The initial thoughts are improved upon. This capture of information and laid out before one’s eyes (Donald called the External Memory Field) was the launch pad for subsequent intellectual growth. So, I am thinking, Donald’s neuroanthropological model might well have resonance for individuals too. It’s a current musing of mine, so thank you for allowing me to air it with you.
Oliver – I wouldn’t go so far as to say that it changed their cognitive architecture. This would imply that it somehow, magically, added a memory store or the like. I assume that even the Greeks had multiple memory stores (phonological loop, visuo-spatial sketchpad + stores for tastes, smells, and feel if we look at more recent work od |Baddeley and Hitch). See figure 6 in: The phonological loop as a buffer store: An update – https://www.sciencedirect.com/science/article/abs/pii/S0010945218301710
Reblogged this on kadir kozan.
Thanks Paul, I think what I clumsily tried to convey was Donald’s notion that “Humans did not simply evolve a larger brain, an expanded memory, a lexicon, or a special speech apparatus; we evolved new systems for representing reality”. Which, I am assuming, is quite in keeping with Paivio’s writing’s in his 1990 Mental Representations. Thanks for the link too.
Reblogged this on From experience to meaning….