← The Casebook

Contaminated record · genuine or implanted?

Leveling and Sharpening

Leveling and sharpening are complementary memory distortions in which peripheral details are omitted or simplified (leveling) while central, schema-consistent, or emotionally salient details are exaggerated or added (sharpening) during recall and retelling.

Exhibit · the account as remembered

Bartlett (1932) demonstrated in his serial reproduction experiments with 'The War of the Ghosts' that memories systematically distort over retellings—shortening by up to 45% while becoming more culturally conventional. Allport and Postman (1947) coined the terms 'leveling' and 'sharpening' in their rumor transmission studies. Klein and Holzman (1950s) operationalized these as individual cognitive control differences via the Schematizing Test. Modern research (Ganske & Hebl, 2001) confirms these processes reinforce stereotypes during memory recall. ← reconstructed, not recorded

How the contamination enters

Leveling and sharpening arise from the reconstructive nature of human memory. During encoding, attention is selective and limited (approximately 4–7 chunks). During retrieval, memory traces are reconstructed using existing schemas—pre-existing knowledge structures that guide what is recalled. Leveling occurs through trace decay of weakly encoded peripheral details and assimilation of similar memories into broader categories. Sharpening occurs through selective rehearsal of central or emotionally charged details, confabulation to fill gaps with schema-consistent information, and source monitoring errors where imagined embellishments are misattributed to the original event.

Competing account — Some researchers distinguish between leveling/sharpening as general memory processes versus individual cognitive styles (Klein & Holzman's 'levelers' and 'sharpeners'). The cognitive style tradition emphasizes stable individual differences, while Bartlett's tradition treats these as universal reconstructive tendencies.

Seams the examination found

seam 01War of the Ghosts serial reproductionStory length reduced from ~330 to ~180 words (~45% reduction)

After about seven serial reproductions, the story shortened from approximately 330 words to 180 words. Unfamiliar details were omitted (leveling)—ghosts, supernatural elements, and culturally strange features disappeared. Details were sharpened to fit British schemas: 'canoes' became 'boats,' 'paddling' became 'rowing,' and 'hunting seals' became 'fishing.' The story became more coherent and conventional over transmissions.

Bartlett · 1932 · n = 20 participants in serial reproduction; additional participants in repeated reproduction condition

seam 02Rumor transmission study40–50% loss of original details after 10 transmissions

Rumors lost 40–50% of original details over 10 transmissions (leveling), while key emotional elements and stereotype-consistent features were amplified (sharpening). For example, a rumor about racial conflict became increasingly assimilated to racial stereotypes as it propagated.

Allport & Postman · 1947 · n not reported in available sources

seam 03Gender stereotype memory distortionConsistent themes across 9 groups (John) and 8 groups (Sylvia); occurrence rates of errors tabulated

For John's story, participants leveled feminine details (baking cookies, crying, talking to a friend) while sharpening competitive and aggressive elements. For Sylvia's story, they sharpened emotional and shopping details while leveling her competitive and aggressive behaviors. Participants also assumed the unspecified competitor 'Terry Browning' was male.

Ganske & Hebl · 2001 · 51 participants evaluating the demonstration; serial chains of 5 participants each

Where provenance breaks

Watch for stories that become shorter and 'cleaner' with each retelling while key dramatic details grow more vivid. When reviewing eyewitness accounts, note whether descriptions have become more stereotypical over time. In organizational contexts, be alert to project post-mortems where failures are leveled into vague generalities while successes are sharpened into heroic narratives.

alteration confirmed

Authenticity — robust. The basic finding that serial reproduction causes progressive distortion of memories replicates robustly—virtually every experiment using Bartlett's technique confirms that social transmission of information is error-prone and more links in the chain increase distortion probability. However, the specific pattern of distortion (what gets leveled vs. sharpened) depends heavily on cultural schemas and stimulus materials. Wheeler and Roediger (1992) found that with long intervals between tests, forgetting occurs but distortion is less pronounced than Bartlett reported. The core phenomenon is well-established, though some quantitative details vary.

Records later found altered

  • law · 1974–2023

    Eyewitness misidentification in DNA exonerations

    As of 2023, eyewitness misidentification has been implicated in approximately 70% of over 3,000 DNA exonerations in the United States. Leveling and sharpening contribute: witnesses level out complex environmental details while sharpening facial features to match prototypical criminal stereotypes. Loftus and Palmer's (1974) studies showed that leading questions can sharpen memories of event intensity, causing false memories of non-existent details like broken glass.

  • law · 1974

    Loftus and Palmer car crash study

    Loftus and Palmer (1974) showed 45 participants films of car accidents and asked about vehicle speed using different verbs ('smashed,' 'collided,' 'bumped,' 'hit,' or 'contacted'). Mean speed estimates varied significantly by verb: 40.8 mph for 'smashed' versus 31.8 mph for 'contacted.' A follow-up with 150 participants found those exposed to 'smashed' were approximately 70% likely (16/23) to falsely recall non-existent broken glass, compared to 12% in controls.

To authenticate the memory

Record information in writing at the time of the event to reduce reconstructive distortion. When evaluating memories or retold accounts, compare current versions against original records. For important decisions, seek multiple independent sources rather than relying on serially transmitted information.

Bartlett found that written reproductions or immediate recall produce more accurate memory than oral serial transmission. Loftus's research shows that warning participants about potential misinformation can reduce but not eliminate the misinformation effect.

Easily confused with

File your own case

Open the same case on your own draft.

Paste a memo, a research draft, or a strategy argument. It is scored against all 175 cards, and the strongest two or three risks come back with the evidence quoted and one practical next check.

Open a case on your draft →