A sample size of one
When Ebbinghaus began his systematic memory experiments in the early 1880s — work he would publish in 1885 as Über das Gedächtnis — he had no laboratory, no funding, and no subjects. What he had was himself.
He invented the nonsense syllable (consonant-vowel-consonant strings such as DAX or BUP) precisely to strip away the associations that make meaningful words easier to recall, then spent years memorising lists of them, testing his own retention at measured intervals, and recording how much effort re-learning took as a proxy for what memory had survived.
His forgetting curve — the steep initial drop in retention that flattens over days — was derived from a long series of trials on a single, obsessively disciplined observer. That it was him made the data coherent. He could hold procedure constant across years in a way that coordinating multiple subjects would have made almost impossible.
What the method was
- nonsense syllablea consonant-vowel-consonant string (e.g. DAX) with no prior associations, invented to isolate memory from meaning
- savings scoreEbbinghaus’s measure: how much less effort re-learning took compared with the first learning, used as a proxy for retained memory
- self-experimenta study in which the researcher is also the sole subject; controls for procedural consistency but eliminates generalisability across individuals
The liability is just as real. A sample of one is, statistically, no sample at all. Ebbinghaus could not know which of his findings were general laws and which were quirks of his own memory, his own rhythms, or his own tolerance for tedium.
The concepts of savings, the spacing effect, and the forgetting curve all emerged from those solitary sessions and have held up well under later replication with larger samples.
His results say nothing reliable about how age, anxiety, prior knowledge, or neurological variation shape forgetting — variables that would only become tractable once researchers at places like the University of Cambridge and, later, Washington University in St. Louis began working with broader populations. Frederic Bartlett’s contrasting approach at Cambridge — giving real stories to real, varied people — was partly a deliberate corrective to exactly this limitation.
Chronology
- Early 1880sEbbinghaus conducts his memory experiments, working alone
- 1885publishes Über das Gedächtnis (On Memory), introducing the forgetting curve and savings method
- 1932Frederic Bartlett publishes Remembering at Cambridge, explicitly challenging the artificial, asocial character of Ebbinghaus’s paradigm
There is also the problem of demand characteristics in an extreme form: the experimenter and the subject shared a skull. Ebbinghaus could not be blind to his own hypotheses, and the discipline required to test yourself honestly, session after session, is itself a personality variable that skews who can produce such data.
None of this erases what he built. But cognitive science has spent much of the century since broadening the base — moving from the one man at his desk to the populations whose memory actually needs explaining.