What the first list does to the second

The task

In 1958 John Brown in England, and in 1959 Lloyd Peterson and Margaret Peterson at Indiana, described versions of a procedure that became one of the most-run experiments in the field. A participant hears a three-consonant string — a trigram such as CHJ — then immediately begins counting backwards by threes from a given number. The counting occupies the interval and prevents rehearsal. After a delay of three to eighteen seconds the counting stops and the trigram is requested.

The published result was steep. Recall was near-complete after three seconds and fell to roughly one item in ten by eighteen seconds. For a field that had been arguing about the mechanism of forgetting for sixty years, this looked decisive: with rehearsal blocked, the trace disappeared on a timescale of seconds. Short-term memory appeared to have a half-life.

The reanalysis

Geoffrey Keppel and Benton Underwood published a paper in 1962 that asked a question the original reports had not addressed. Every participant in these experiments completed many trials in sequence — dozens of trigrams, one after another. Keppel and Underwood separated the data by trial position and looked at the first trial alone.

On the first trial there was very little forgetting across the eighteen-second interval. Recall was high at three seconds and remained high at eighteen. The steep decline appeared on the second trial, deepened on the third, and stabilised thereafter. The curve that had been read as decay was an artefact of aggregation: what accumulated across the session was not elapsed time but earlier trigrams, and those earlier trigrams were competing at retrieval.

This is proactive interference — disruption running forward from older material to newer, the mirror image of the retroactive case Müller and Pilzecker had described in 1900. Keppel and Underwood's contribution was to show that a laboratory task could manufacture it in minutes and then be misread as measuring something else entirely.

Release

A converging result arrived the following year. Delos Wickens, working with colleagues including Dallett Born and Charles Allen, ran the same task but changed the category of the material partway through. A participant who had been receiving consonant trigrams for three trials received digits on the fourth. Recall on that fourth trial recovered sharply, approaching first-trial levels.

Wickens named the effect release from proactive interference, and it carried an implication beyond the immediate finding. If a shift in category releases the buildup, then whatever accumulates must be organised by category — the interference is between similar things, not among all preceding things equally. Through the 1970s the release paradigm became a tool for asking what dimensions the memory system encodes, since any dimension that produces release is a dimension the system is sensitive to. Taxonomic category, semantic sense, grammatical class and even the language of presentation were tested this way.

What the account does not settle

The interference reading of the Brown–Peterson task is standard in textbooks, which tends to obscure how contested the underlying question remains.

Alan Baddeley and Barbara Scott reported in 1971 that with a large sample and a single trial per participant — removing any possibility of buildup — some forgetting still occurred across short intervals, though far less than the aggregated curves had shown. That result is hard to reconcile with a purely interference-based account, and it has been replicated and disputed in roughly equal measure since.

The modern version of the argument runs between two positions. Pierre Barrouillet and colleagues have defended a time-based resource-sharing model in which attention decays when it is diverted. Klaus Oberauer and Stephan Lewandowsky have argued that the evidence taken to support temporal decay can be reproduced by interference models without any decay parameter. The disagreement has proved unusually durable because the two accounts make similar predictions across most standard designs, and separating them requires experimental controls that are themselves contested.

What is not in dispute is the narrower methodological lesson. A curve averaged across trials can describe a process that occurs on no individual trial, and the aggregate can point confidently in a direction the data do not support. That lesson has outlived the specific controversy that produced it.

• Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experimental Psychology, 10, 12–21.
• Peterson, L. R. & Peterson, M. J. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58, 193–198.
• Keppel, G. & Underwood, B. J. (1962). Proactive inhibition in short-term retention of single items. Journal of Verbal Learning and Verbal Behavior, 1, 153–161.
• Wickens, D. D., Born, D. G. & Allen, C. K. (1963). Proactive inhibition and item similarity in short-term memory. Journal of Verbal Learning and Verbal Behavior, 2, 440–445.
• Baddeley, A. D. & Scott, D. (1971). Short term forgetting in the absence of proactive interference. Quarterly Journal of Experimental Psychology, 23, 275–283.
• Oberauer, K. & Lewandowsky, S. (2008). Forgetting in immediate serial recall. Psychological Review, 115, 544–576.

Editorial note: this article summarises published findings. It is not advice, and it makes no claim about any individual reader.