The man who tried to measure forgetting

Suppose you want to find out how much of a poem someone remembers tomorrow. You ask them to learn it today and recite it after a night's sleep.

The test sounds simple until you consider the poem. Some lines may already be familiar. Its rhythm might help one reader more than another. An image might bring back a personal memory. If the results differ, how much of that difference belongs to memory, and how much to the material?

Hermann Ebbinghaus needed material and conditions that could be compared fairly. In his 1885 book Über das Gedächtnis, known in English as Memory, he described experiments in which he used himself as the participant and tried to make the learning task easier to control.

Why a poem was hard to compare

Meaning gives us useful help when we learn. A familiar phrase or a striking image can give a new word somewhere to attach.

That help becomes a complication when you're trying to compare learning under different conditions. Two lists may take different amounts of effort because they bring different associations with them.

Ebbinghaus used sequences of syllables intended to carry little familiar meaning. His account of the method also acknowledges that the material was far from perfectly uniform.

He was trying to limit some of the differences between learning tasks. Learning languages still calls for meaningful words and situations; unfamiliar syllables served a particular experimental purpose.

What a second attempt can reveal

A recall test gives an obvious result: you can recall the answer or you can't. But that boundary may hide something useful.

After learning a sequence and waiting, Ebbinghaus learned it again. He compared the work needed the second time with the work needed initially. Less time or fewer repetitions suggested that the first learning was still doing some of the work. He explained this approach in his discussion of measuring memory.

The difference is often called savings in relearning. You might fail to recite a sequence and still learn it again more quickly. An unavailable answer doesn't necessarily mean that every benefit of earlier learning has disappeared.

The second attempt asks a different question. It looks at how much work is required to regain access, rather than only whether access is possible right now.

What the measurement leaves out

Savings describes a difference in learning effort. It isn't the same as the chance of answering correctly in a conversation tomorrow.

This matters when you see a memory experiment summarized as a percentage. Before using the number, you need to know what was measured. Time saved, items recalled, and probability of recall describe different things.

Ebbinghaus's work also had limits. A researcher studying his own performance with artificial lists cannot represent every learner or every kind of knowledge. His results don't give one timetable that fits every learner using a language app.

For your own learning, the useful point is narrower. If a word doesn't come, a later attempt may still benefit from earlier practice. A single missed answer cannot tell you how much work remains.

How a delay became a variable

Once learning and relearning could be compared, the interval between them became something to investigate. Ebbinghaus could vary the delay and examine how the work changed.

That gave forgetting a repeatable procedure: learn, wait, learn again, compare. The underlying memory remained out of sight, but part of its effect could be measured.

Comparing different delays raised the next question. Did the loss proceed at a steady rate, or did the pattern change as time passed? The answer became a curve, which we'll examine in the next article.