Cognitive Connie
Understanding peterson & peterson — short-term memory
In 1959, Lloyd and Margaret Peterson posed a simple question: how long does a memory last without rehearsal? Their method was elegant. Participants heard a consonant trigram — three unrelated letters such as CHJ or QPW — followed immediately by a three-digit number. They then counted backward in threes from that number aloud (e.g., 397, 394, 391…) until a signal to recall. Counting prevented any form of rehearsal. The result was striking: after just 18 seconds, recall accuracy had plummeted from near-100% to roughly 5–10%.
Defining features
Key figures
Lloyd R. Peterson
1922–2007American psychologist who, with his wife Margaret, designed the consonant-trigram distractor paradigm that provided the first clean empirical estimate of STM duration without rehearsal.
Margaret J. Peterson
Co-designed and co-ran the 1959 study at Indiana University. The paper is the foundational source for STM duration and has been cited thousands of times in memory research.
Geoffrey Keppel & Benton Underwood
Re-analysed the Peterson & Peterson data in 1962, identifying proactive interference rather than decay as the likely explanation for the forgetting curve — sparking the decay vs. interference debate that continues today.
Methods
Design of the trigram study
Consonant trigrams
Each trial presented a consonant trigram (e.g., CHJ) — three unrelated consonants with no obvious chunking or word-formation. Trigrams were used rather than words precisely because they resist semantic organisation, controlling for the LTM benefit of meaningful material.
Distractor task (backward counting)
Immediately after hearing the trigram, participants heard a three-digit number and began counting backward in threes aloud. This articulatory suppression task occupied verbal rehearsal capacity, preventing participants from silently repeating the trigram and thereby measuring the natural persistence of the memory trace without active maintenance.
Retention intervals
The counting task continued for 3, 6, 9, 12, 15, or 18 seconds before the recall signal. By systematically varying the interval, the Petersons could map the complete decay function for unrehearsed STM within a single elegant design.
Within-participants design
Each participant completed many trials across all intervals. This maximised statistical power but also introduced a proactive interference confound — earlier trials' items could interfere with later recall — a limitation later identified as critically important by Keppel and Underwood (1962).
Findings
What the data showed
Rapid forgetting without rehearsal
Recall was ~90% at 3 seconds, declining to ~55% at 9 seconds, and to just ~5–10% at 18 seconds. The decay curve was steep and consistent: in the absence of rehearsal, unrehearsed traces in STM essentially vanished within 18–20 seconds.
Evidence for a separate STM store
The speed of forgetting was far faster than forgetting from long-term memory, which typically occurs over minutes to days. The Petersons interpreted their results as showing a distinct, fragile short-term store with a very limited natural duration — supporting Atkinson and Shiffrin's later distinction between STM and LTM.
Criticism
The interference alternative
Keppel & Underwood (1962): proactive interference, not decay
Keppel and Underwood reanalysed the data and showed that forgetting was minimal on the very first trial of each experimental session — the steep decay only appeared after several trials had accumulated. This proactive interference pattern suggests that earlier trigrams were competing with the to-be-recalled trigram, rather than the trace simply decaying over time. The decay vs. interference debate has never been fully settled.
Counting backward as a non-neutral distractor
Backward counting is not a neutral filler task — it is itself a verbal, phonologically demanding activity that may actively overwrite the phonological trace holding the trigram, rather than merely preventing rehearsal. This confounds decay with interference from the distractor itself, complicating interpretation.
Artificiality of consonant trigrams
The materials are deliberately unrelated to everyday memory. The results describe worst-case forgetting of maximally unstructured verbal material — they tell us little about how long we retain meaningful sentences, faces, or events, all of which benefit from semantic organisation and LTM support.
Relevance for today
Why it still matters
Foundation for the multi-store model
The Petersons' finding of a ~15–30 second natural STM duration was a key piece of evidence for Atkinson and Shiffrin's (1968) multi-store model, distinguishing the temporary short-term store from the more durable long-term store. Together with Miller's capacity findings and Conrad's coding findings, it defined the core properties of STM.
Rehearsal and attentional interference
The study makes vivid why unrehearsed information — a phone number, a name just heard — is lost almost immediately. Any task that occupies verbal attention during encoding will prevent rehearsal and cause forgetting within seconds. This has become a key paradigm in clinical neuropsychology for understanding prospective memory failures and the role of attentional resources in short-term retention.
Decay vs. interference debate
The controversy sparked by Keppel and Underwood remains alive in working memory research. Contemporary computational models of forgetting (e.g., Nairne, 1990; Oberauer & Lewandowsky, 2008) continue to test whether interference or time-based decay better explains the rapid loss of unrehearsed traces — one of the most productive empirical debates in cognitive psychology.
Test your knowledge
Frequently asked questions
Why did Peterson & Peterson use backward counting?+
To prevent rehearsal. If participants could silently repeat the trigram, STM traces would be actively refreshed and would not decay — so the study would measure rehearsal capacity, not natural STM duration. Backward counting in threes is demanding enough to occupy the verbal rehearsal system completely, giving a measure of what the trace does when left alone.
Did Peterson & Peterson prove that STM decays?+
They showed that unrehearsed material is forgotten very rapidly — but not definitively that decay is the cause. Keppel and Underwood (1962) showed the forgetting was largely absent on the first trial of a session, implicating proactive interference from earlier trials rather than passive time-based decay. Whether forgetting from STM reflects decay, interference, or both remains debated.
Sources
Last reviewed August 2026- 1.
Peterson, L. R., & Peterson, M. J. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58(3), 193–198. https://doi.org/10.1037/h0049234
+About this source
Original trigram study demonstrating ~15–30 second natural duration of unrehearsed STM across six retention intervals.
- 2.
Keppel, G., & Underwood, B. J. (1962). Proactive inhibition in short-term retention of single items. Journal of Verbal Learning and Verbal Behavior, 1(3), 153–161. https://doi.org/10.1016/S0022-5371(62)80023-1
+About this source
Reanalysis of Peterson & Peterson data showing forgetting is absent on trial 1, implicating proactive interference rather than decay.
- 3.
Baddeley, A. D. (1986). Working Memory. Oxford University Press.
+About this source
Places the Peterson & Peterson findings in the context of the working memory model, explaining how the phonological loop mediates short-term retention and how its disruption by articulatory suppression relates to the trigram paradigm.