Learning as a Generative Process
Jonathan Firth's Memory & Metacognition Updates #31
I hope things are going well for you. And for those who have a mid-term break right now, enjoy!
Last time I talked a little about generative learning, and today I’d like to say more about this important theory. As I mentioned before, it is a model of learning which emphasises certain processes, each of which involves using existing knowledge to create something new. These include:
summarising a text
generating an example
predicting something
producing a new creative output based on course material
One thing that is really interesting about generative learning theory is that it neatly combines several ideas discussed previously in these updates. It focuses on how we process new information into memory, but it also places a great emphasis on a learner’s understanding, and their existing schemas. It focuses on how new information can be integrated and creatively used. It tries to explain transfer.
In doing so, the originator of generative learning (the marvellously named Merlin C. Wittrock) combined two important perspectives on learning that are often seen as separate, but which both inform modern cognitive science.
Classic theories
The two major ideas I’m thinking of are as follows:
Cognitivism, a view of learning that emphasises processing in working memory as a means of getting things into long-term memory. This perspective therefore sees learning as mainly about input, and focuses on things like repetition and memory.
Constructivism, which sees learning as a more active process, based on prior knowledge. This perspective focuses on how a learner makes sense of what they are learning, and focuses on the role of schemas.
I’ve always agreed partially but not entirely with both of the above ideas! After all, most of what I say about memory fits well with a cognitivist perspective. It’s about how learners get ideas into memory. But at the same time, I can see that simplistic versions of this theory focus too much on processing in working memory, and tend to ignore the important issue of what the learner already knows and understands.
Here is something that Wittrock said when he first proposed generative learning in 1974 (p. 89):
“the generative model predicts that learning is a function of the abstract and distinctive, concrete associations which the learner generates between [their] prior experience, as it is stored in long-term memory, and the stimuli.”
Although Wittrock’s theory could be described as constructivist, it can be seen that he was very much focused on how this fits with existing cognitive research, including our understanding of long-term memory. Indeed, he describes the theory as “a cognitive model of human learning”, and praises the previous work done in experimental and educational psychology.
However, he (rightly, in my view) expresses the importance of combining those prior cognitive theories with advances from developmental psychology, and of applying them to realistic learning situations.
You can read Wittrock’s 1974 paper here:
Wittrock (1974) “Learning as a Generative Process”.
There are two key ideas, as I see it:
Viewing learning as an active, constructive process, where new information is connected with existing knowledge.
Viewing learning as a generative process, where meaningful creation is valuable to learning.
The model today
I’m sure you’ll have noticed that many people who talk about applying cognitive science to education focus heavily on the role of memory, including working memory.
However, such theories are over-simplistic if they ignore what the learner already knows, as Wittrock argued. It’s too simplistic to view working memory as a ‘conveyor belt’ into long-term memory (which is why pure repetition does not work as a learning strategy).
And it’s over-simplistic to view schemas just as knowledge storage devices. Schemas also affect our later understanding and processing, too. We use our knowledge to make sense of the world and to think, as Willingham explains here.
Fortunately, we do see good examples of the idea that learning need to be generative in the following model of CPD, and similar ones:
As can be seen, this and similar models emphasis prior knowledge. If learners don’t know enough, they struggle to learn. Another effect of schemas is that learners may bring misconceptions to the classroom.
In terms of pedagogy that focuses on generation, I mentioned last time that the research by Dr Dietsje Jolles on generating predictions also fits this model well.
I’d be interested to know your views on this, and especially on whether you think that an emphasis on both memory and prior knowledge are being captured well enough in CPD and in broader discussions about learning.
All the best,
Jonathan
Last week: Generation, Creativity, and Predictions
Website: www.jonathanfirth.co.uk
Did someone forward this to you? Subscribe to Memory & Metacognition Updates below, so that you don't miss anything!