Since Ebbinghaus first described his famous ‘Forgetting Curve’ in 1885, the validity of his findings has frequently been disputed. His methodology, as well as some of the assumptions underpinning his research, have been called into question. So has Ebbinghaus been debunked? Should we place him and his theories in that fast-filling file of once trusted and even cherished ideas that have turned out to be flawed – along with Learning Styles, Myers-Briggs personality types and NLP?
Let’s look at the two principal areas of criticism:
- Ebbinghaus experimented only on himself, opening the door to bias and skewed data
- His use of ‘nonsense syllables’ in the experiments makes a nonsense of the results
Before we explore these objections in more detail it’s probably best to briefly describe how his experiments for the forgetting curve were conducted.
Ebbinghaus’s basic memory experiment was fairly simple. He began by creating a set of 2,300 single-syllable nonsense words or trigrams (e.g. DAX, BOK, YAT), which would have no associative value to aid memorability. At each session, he would pull a number of these out from a box at random and write them down in a notebook. Next, to the beat of a metronome, he would read out the syllables and then try to recall them at the end of the session. The experiments involved tens of thousands of recitations, and by plotting the data from his results he was able to draw a curve to show how memory fails over time.
Though rigorous in its own way, the experiment had only one experimental subject: Ebbinghaus himself. Surely this should limit the generalizability of his results and open the door to bias? Does this aspect of his method make the results invalid?
Objection 1: Self-experimentation
It should be recognized that self-experimentation was not at all uncommon in the era when Ebbinghaus was working. We owe many scientific discoveries, including that of vaccination, to some often very risky examples of self-experimentation. And it continues down to our own era, in fact, with a number of scientists (including at least five Nobel laureates) using self-experimentation to advance progress in fields such as infectious diseases, vaccine research, cancer, and pharmacology.
More significantly, perhaps, Ebbinghaus’s experiment has frequently been replicated down the years, not only by individuals but by whole cohorts of learners. The fundamental result has held firm.
To put this in context, in the last ten years or more, science has been going through a ‘replication crisis’, with extremely high numbers of key findings in the literature proving to be incapable of replication by modern experimenters. This crisis has been particularly severe in the field of psychology. Ebbinghaus’s forgetting curve, it seems, is a highly replicable finding; making it a particularly robust piece of science.
But what about the other major challenge to Ebbinghaus’s forgetting curve: the fact that he used nonsense syllables?
Objection 2: Nonsense in, nonsense out
This objection is most lucidly and compellingly expressed by Nick Shackleton-Jones in his 2019 book ‘How People Learn’. Shackleton-Jones has a theory of his own, the Affective Context Theory, that emphasizes the importance of emotion in imprinting memories. Put simply, it says that you can’t learn anything you don’t care about.
Ebbinghaus’s nonsensical trigrams, according to this theory, were inherently unmemorable. It would be impossible for any subject to form an emotional connection with a word which is hardly even a word composed of three random letters, because making an emotional connection is, under the Affective Context Theory, the key to imprinting anything at all in memory. Ebbinghaus’s experiments, therefore, were too disconnected from the normal situation in which learning takes place to tell us anything useful. ‘Led astray by his assumptions,’ Shackleton-Jones writes, ‘he went on to investigate means by which the mind might be forced to retain nonsense… He discovered a kind of psychological force-feeding method.’
The Affective Context Theory is well backed up by other findings in cognitive psychology, however, this objection is probably not the body blow to Ebbinghaus’s relevance that might be supposed. In order to truly blow Ebbinghaus out of the water, surely, you would have to create an experiment that told us what the curve looked like when inherently memorable information is to be learned.
In fact, Ebbinghaus had been down this route himself before arriving at the final form of his experiment, as described by Jaap Murre and Joeri Dros in their paper about their own replication of his findings:
‘After having tested himself with tones, numbers, and poem stanzas, he decided that none of these served his purposes. Tones were too cumbersome to handle and too difficult to reproduce for him, he did not find digits zero to nine suitable as basic units for the long-running experiments he envisioned, and the poem fragments he tried to learn (from Byron’s Don Juan) were deemed too variable in the meanings they evoked and therefore likely to cause measurement error… He, therefore, introduced nonsense syllables, which had more uniform characteristics than existing words or other verbal material. In his later experiments on learning, however, he did verify his results with the Don Juan verses, confirming both his main results on learning and his intuition that the latter stimuli did indeed yield much more variance in the data.’
So what’s the verdict?
It’s not unusual nowadays to read blog posts on learning by authors who take it as read that Ebbinghaus has been debunked.
Wrongly, as it turns out. In fact, Ebbinghaus’s discovery of the spacing effect in learning is one of the most robust findings in the whole of experimental psychology. And not only is the necessity for learning professionals to know about this effect more pressing than ever, but it is also now easier to implement in training programs than ever before, due to advances in technology and the advent of learning systems such as LXPs and LMSs with LXP-like features.
At the same time, we should recognize limits to Ebbinghaus’s research and what we can draw from it – even if the prevailing culture of debate in learning makes it hard to take a balanced view.
As an area of professional practice, learning tends towards essentialism – the belief that there are, for example, exactly 16 personality types (Myers-Briggs), nine significant team roles (Belbin), and only four levels of evaluation (Kirkpatrick). Useful as such models might be, they don’t necessarily tell the whole story about their respective areas of personality type, team working and evaluation. Nevertheless, such models often get very rigidly applied and can get cemented into the process. While this might save time and effort for a training manager, it will necessarily tend to shut off a more nuanced appreciation of the problem at hand.
The same tendency towards essentialism can be seen at work in the expectations we tend to have of canonical theories such as the Forgetting Curve. We expect that they should explain everything there is to know about learning, ignoring the partial and incremental nature of progress in the science of learning.
Ebbinghaus told us something of fundamental importance about memory – that learning is a process, not an event. Something we should indeed make efforts to remember.
This blog is the first in a six-part series to follow the recent publication of our, ‘The Spacing Effect: Harnessing the Power of Spaced Practice for Learning That Sticks’ whitepaper. Discover the spacing effect and how it can be harnessed using learning technologies. Download it now.
John Helmer FLPI is a writer, speaker, podcaster and communications consultant. He runs and presents the acclaimed fortnightly podcast The Learning Hack, which features conversations with leading figures in learning in the US and UK.