Why Ebbinghaus' forgetting curve isn't useful for teaching
Updated: Aug 28
Ebbinghau's forgetting curve has seen a resurgence in popularity. While I agree that knowledge is essential for domain specific understanding and thinking, I argue here that Ebbinghaus' experiment, which has been generalised by many as a model of learning and forgetting is not a useful model for teaching. It is not an argument against retrieval practice.
What the Ebbinghaus curve shows
Ebbinghaus conducted an experiment on himself (n=1) to see how long it took to relearn absolutely nonsense words. The curve thus shows how much he was remembering, or forgetting, utterly isolated facts (see Gareth Bates' post for details of what the Y axis means).
What the Ebbinghaus curve does not show
The curve doesn't allows us to know much about forgetting anything other than isolated facts.
The distinction between facts and isolated facts is important. I cannot imagine any teachers ever teaching completely isolated facts. The discussion I see on Twitter seems often to be about contextualisation and the sequencing of content for understanding.
Not only do teachers go to great efforts to ensure new content is never isolated, students also bring a wealth of prior knowledge from other subjects and their lives to connect to and find meaning.
While we must always search for simplified (idealised) models to help our thinking, useful models are those that help us approximate reality. If we, as teachers, are working towards connected, networked, meaningful knowledge then we must question the popularity of a model that does not allow us to think precisely about these things. This is especially made worse when the graphs of precise timings and percentage of forgetting are shared with students.
Furthermore, while the Ebbinghaus curve remains popular as a go-to mental model of forgetting for both teachers—and for students when shared with them—it diverts attention away the more complex reality of learning. That complex reality of learning is actually our goal.
Ebbinghaus' forgetting curve may be intuitive and easy, but it takes us away from where we want to go and the type of learner we want to build. And we have to be careful with how a dominant model affects the learning culture of a classroom. Think about it this way (Hoverstadt 2022):
... to quote Einstein: “Whether you can observe a thing or not depends on the theory which you use. It is the theory which decides what can be observed.” Substitute ‘model’ for ‘theory’ and it is the models we use that determines what we can see, and critically what we will fail to spot, what we can understand and what we can learn.
I don't think we need to refer to Ebbinghaus in an effort to obtain authority over the idea. I think humans find the idea of forgetting quite intuitive—it happens all the time. How about a model that focuses on forgetting in more useful way for developing a culture of learning?
A useful model of learning and forgetting: Ausubel's obliterative subsumption
Ausubel's theory of learning, at its simplest, makes a distinction between rote learning and meaningful learning. Rote learning is the learning of isolated facts—they don't hold connections to other knowledge. Meaningful knowledge is the opposite, new knowledge being incorporated into a growing cognitive network (Figure 1).
Figure 1. Copied from: Novak, 2010. Learning, Creating, and Using Knowledge
With forgetting, again Ausubel made the distinction again between rote and meaningful. Rote learning, once forgotten, is lost forever. But when forgetting meaningful knowledge the cognitive structure—the network configuration we could say (or maybe even the new general way of seeing that knowledge)—remains.
The details go, but a general understanding persists. When we want to revisit the learning, the details are incorporated into the structure that remained—we can relearn details quickly. With revisiting rote learning, the learning is from scratch.
While this is also an idealised model, it diverts our attention to meaningful learning as the goal of knowledge, rather than just the retention of (any) knowledge. This in itself is much more useful as the models we hold closest affect how we and our students perceive the course and its objectives—something I discuss in detail in my book.
And it's just as simple to explain to students as a forgetting curve. A teacher could sketch an idealised concept map on the board, emphasising the connections, and then rub out some nodes to show forgetting while the structure remains (Figure 2).
Figure 2: A simple attempt at showing Ausubel's model of forgetting meaningful learning
This expresses more clearly that connected knowledge in meaningful mental models is the goal of learning, and thus, when teachers & students engage in retrieval practice they should have in mind that those questions should be building to something more bigger and connected.
Ebbinghaus: Focus on forgetting isolated knowledge.
Ausubel: Focus on learning and forgetting meaningful knowledge.
If you want to read more about meaningful learning—how to bring it about and how to see it—then check out my book. Download chapter 1 here—English edition—edición española—or check out my other posts.
Hoverstadt, P. 2022. The Grammar of Systems: From Order to Chaos & Back. SCIO Publications.
Novak, J. D. 2010. Learning, Creating, and Using Knowledge: Concept Maps as Facilitative Tools in Schools and Corporations. 2nd ed. Oxford: Routledge.