MC · The Lure

A three-minute memory experiment

You are about to remember words that were never there.

Thirty-six cards, dealt once. A thirty-second break. Eighteen judgments. Three of the words you will swear you saw were never in the deck — and this page will prove it, with your own hand.

  • 36 dealt
  • 30 s of sums
  • 18 judged
  • ~3 min total

Your answers never leave this page. No account, no server, no leaderboard — the only mind measured here is yours. Built on the DRM paradigm (Deese 1959; Roediger & McDermott 1995).

The deck is shuffled

Thirty-six cards, shown once, one and a half seconds each.
Then sums. Then judgment.

about three minutes · eyes on the cards · answers are final, like memory

Why the ghosts get in

Read the deck and your mind does two things at once. First it spreads: bed lights up rest, rest lights up tired, and the whole lit neighbourhood points at one word that never appears — the lure. By the end of the list that word is glowing, activated by everything around it, though your eyes never met it.

Then the tag washes off. Every memory carries a small label — seen versus thought — and labels smudge. At test time the lure feels familiar (it was activated, after all) and its label reads seen. Psychologists call this a source-monitoring failure: you remember the word correctly and its origin wrongly. A cousin theory, fuzzy-trace, says you keep the gist of the list rather than its print — and the lure is the gist.

The lab behind the trick

In 1959 James Deese noticed that word lists kept producing intrusions — people recalled words he had deliberately left out. Thirty-six years later Henry Roediger and Kathleen McDermott turned the observation into a paradigm: lists of associates, each converging on one absent lure, then a recall or recognition test. The effect has been replicated hundreds of times since; norms for 36 lists were published by Stadler, Roediger and McDermott in 1999.

What the lab finds, on average

84%
of unseen lure words are falsely recognized as “old” — against 86% correct recognition for words actually studied. Unrelated control words: 2%. (Roediger & McDermott 1995, Experiment 1)
58%
of those false alarms are marked “sure old” — the highest confidence on offer. Certainty is no disinfectant. (Same experiment)
4 in 10
lure words are falsely recalled outright, unprompted, in free recall. (Roediger & McDermott 1995, Experiment 1)

Group averages, not prophecies: the lab ran longer sessions than this page (the norms study covered 36 lists; you run three families of twelve). The English lists here are the standard published associates; the Romanian lists are careful adaptations, not normed data. Your run is one data point — compare its shape, not its decimals.

Why it matters outside this page

Word lists are toys; the machinery is not. The same reconstruction that printed sleep into your deck operates on faces, conversations and childhoods. Its heaviest bill arrives in courtrooms: more than 60% of the Innocence Project's clients were wrongfully convicted based on eyewitness misidentification — which the organisation calls a contributor to an overwhelming majority of convictions overturned by post-conviction DNA testing. And confidence misleads twice — a confirming word after an identification (“good, you picked the suspect”) measurably inflates a witness's later certainty (Wells & Bradfield 1998).

Closer to home, it operates in every quarrel about what was “definitely” said. The quarrel is usually sincere on both sides — that is the disquieting part. On planted memories with darker histories — the recovered-memory therapies of the 1980s–90s — see the companion investigation What Remains of Psychology.

The same machinery, pointed forward

There is a constructive flip. If retrieving builds memory — even false ones — then deliberate retrieval builds true ones: testing yourself beats re-reading, by a large margin (Roediger & Karpicke 2006; see the testing-effect entry in the companion piece). You have just demonstrated the engine on yourself. Aim it at something you actually want to keep, and test before you feel ready.

What this test is not

  • It is not a measure of your memory ability, and certainly not a diagnosis of anything. It measures one session, on one device, with three word families.
  • Word lists are not life events. False alarms for sleep do not imply false memories of things that happened to you.
  • The Romanian lists are translations, not normed instruments — expect the effect's shape, not laboratory decimals.
  • Skipping the sums, peeking, or replaying changes the result honestly: the page shows what your hand did under the conditions you chose.

Sources

  • Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of Experimental Psychology, 58, 17–22.
  • Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21(4), 803–814.
  • Stadler, M. A., Roediger, H. L., & McDermott, K. B. (1999). Norms for word lists that create false memories. Memory & Cognition, 27, 494–500.
  • Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114(1), 3–28.
  • Wells, G. L., & Bradfield, A. L. (1998). “Good, you identified the suspect!”: Feedback to eyewitnesses distorts their reports of the witnessing experience. Journal of Applied Psychology, 83(3), 360–376.
  • Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning. Psychological Science, 17(3), 249–255.
  • Innocence Project — Eyewitness misidentification (“more than 60% of our clients”; accessed September 2026).

Method, list provenance and claim checks: SOURCES · Full protocol: 3 × 12 associates, 30-second arithmetic distractor, 18-item old/new test with 3-level confidence.