Mark ở vũ trụ khácVIMark học AI
Mark ở vũ trụ khác · Another Universe Common Myths · Memory · Tracing the source

Common Myths #9: Memory works like a video camera

? REC

"But I remember it clearly." Clearly, yes. Correctly, not necessarily.

The mytha memory is an exact recording; vivid means accurate
Who believes it63% of Americans surveyed in 2011
Realityevery act of remembering is a rebuild
Main sourcesLoftus & Palmer 1974, Neisser & Harsch 1992, Talarico & Rubin 2003

Source: the 63% figure is from Simons and Chabris, PLoS ONE 6(8): e22757, 2011, a survey of 1,500 Americans.

Reread in 1 minute

Essence

There is no tapeThe brain keeps no clip to rewind. It stores fragments and reassembles them each time.
One word changes a memoryAsk "smashed" instead of "hit" and twice as many people recall broken glass that was never there.
Vivid is not accurateMemories of shocking news stay vivid for months, even as the details drift like any everyday memory.
The gist is usually rightMemory is not useless: the big picture tends to hold, while who said what tends to go wrong.
Write it down, don't argueAfter a meeting, write the decisions within 5 minutes. Argue "who promised what" from memory and everyone feels right.

Sources: Loftus and Palmer (1974); Talarico and Rubin (2003); Neisser (1981). Details and figures in Part 3.

Part 1

The question

If memory is not a camera, why does a memory play like a film, sharp down to the details, and why do we trust it so much? The answer: the feeling of "clear" comes from the brain building the scene, not from the scene being right.

When it happensFragments storedOnly what we paid attention to: a few images, a few words, a feeling.
→
AfterwardsFragments fade, new ones creep inOther people's questions, things we read, each retelling.
→
When recallingAssembled into a sceneGaps get filled with what is "plausible". The new scene looks completely real.
Part 2

Where the myth came from

The "recording" metaphor dates back to ancient Greece and gets refreshed by each new technology: wax tablet, film reel, videotape, hard drive.

Timeline of the memory-as-recording metaphor. Around 369 BC, Plato compares memory to an imprint in a wax tablet. 1932, Bartlett shows remembering is reconstruction. 1963, Penfield summarises his brain stimulation work, likened to a film reel. 1974, Loftus and Palmer: one verb changes a memory. 1980, Loftus and Loftus: no evidence memories are permanent. 1992, Neisser and Harsch: memories of Challenger drift. 1995, Roediger and McDermott: people recall words never shown. 2011, survey: 63% of Americans still believe memory works like a video camera. ~369 BCPlato: wax tablet 1932Bartlett:memory rebuilds 1963 · Penfield"like a film"? 1974one verbchanges memory 1980not permanent 1992Challengermemories drift 1995words never shown 201163% still agree red: moments the "recording" metaphor was reinforced · teal: evidence against it
The evidence against it goes back to 1932, yet the metaphor keeps winning. Every new recording technology hands people an easy picture for talking about memory.Sources: Plato, Theaetetus 191c–d; Bartlett, Remembering (1932); Penfield and Perot, Brain 86 (1963); Loftus and Palmer (1974); Loftus and Loftus, American Psychologist 35 (1980); Neisser and Harsch (1992); Roediger and McDermott (1995); Simons and Chabris, PLoS ONE (2011).
~369 BC · Plato, TheaetetusMemory as an imprint in a wax tablet. Plato himself said imprints can be faint or misaligned, so misremembering is normal.
1932 · Frederic Bartlett, "Remembering"English participants retold an unfamiliar folk tale. With each retelling it got shorter and more like familiar stories. Bartlett called remembering "reconstruction", not "replay".
1950s–1963 · Wilder Penfield stimulates the brain during epilepsy surgerySome patients described reliving old scenes, like a film with a soundtrack being switched on. But only 40 of 520 patients stimulated in the temporal lobe had such experiences, and none could be verified as real events.
1980 · Elizabeth and Geoffrey LoftusReviewing Penfield, hypnosis and psychoanalysis: no evidence confirms that all memories are stored permanently and can be retrieved.
2011 · Survey of 1,500 Americans63% agreed that "human memory works like a video camera". All 16 memory experts asked disagreed.

Sources: Penfield and Perot (1963), Brain 86: 595–696 (40/520 via the 2023 review "Brain stimulation and elicited memories", secondary source); Loftus and Loftus (1980), American Psychologist 35: 409–420; Simons and Chabris (2011), PLoS ONE 6(8): e22757, results table.

Why the myth is easy to believe

Memories "look" like film scenes. When we recall, we see images, hear sounds, feel emotions. The brain builds the scene so well that we cannot tell which parts were filled in.

Technology metaphors. Camera, cassette, hard drive: anything that stores becomes a picture of memory.

Movies. A character is hypnotised and "replays" the crime exactly. In the 2011 survey, 55% of Americans believed hypnosis helps people remember accurately.

Source: Simons and Chabris (2011), PLoS ONE, hypnosis item.

Part 1 of 3

Where the myth came from

wax tabletfilm reel63% still agree"I remember clearly"
Part 2 of 3

What the evidence says

broken glassChallenger9/11words never shown
Part 3 of 3

Work lessons

5-minute minutesnon-leading questionsopen the notes, don't argue
Part 3

What the evidence says

Six studies, each removing one piece of the myth.

One verb changes both speed and broken glass

misinformation effectLoftus and Palmer, 1974
Mean speed estimates by the verb in the question: smashed 40.8 miles per hour, collided 39.3, bumped 38.1, hit 34.0, contacted 31.8. One week later, people saying they saw broken glass although the film had none: smashed group 32%, hit group 14%, group not asked about speed 12%. "About how fast were the cars going when they ___ each other?" smashed40.8 collided39.3 bumped38.1 hit34.0 contacted31.8 mph (40.8 ≈ 66 km/h) 1 week later: "Did you see broken glass?" 32%smashed 14%hit 12%not asked no glass in the film · 50 per group
A question asked after the event becomes part of the memory. People asked with "smashed" did not just guess faster; they also "remembered" a detail that fit the word.Source: Loftus and Palmer, Journal of Verbal Learning and Verbal Behavior 13: 585–589, 1974, experiment 1 (45 people, 5 verbs) and experiment 2 (150 people: 16/50, 7/50, 6/50 answered "yes"). Per-verb means taken from consistent summaries of the original paper (secondary sources).

"Flashbulb" memories of Challenger

flashbulb memoryNeisser and Harsch, 1992
The day after the 1986 Challenger disaster, students wrote down how they heard the news. About two and a half years later, 44 were asked again. Mean agreement with their original account was 2.95 out of 7. 11 scored zero. Mean confidence was 4.17 out of 5. 29 Jan 1986 · by hand where they heard, withwhom, from what, when autumn 198844 people asked again match with first account 2.95 / 7 confidence they were right 4.17 / 5 11/44 scored zero 7-point scale: where, doing what, who told them, how they heard...
Wrong, but very confident. The more shocking the event, the surer people were they remembered it "like it was yesterday". Most recalled a version different from the one they had written themselves the next day.Source: Neisser and Harsch, "Phantom flashbulbs", in Affect and Accuracy in Recall, Cambridge University Press, 1992, pp. 9–31. The figures 2.95/7, 11/44 and 4.17/5 are taken from several reviews citing the chapter (secondary sources); the stated interval varies between sources, from 2.5 to nearly 3 years.

Confidence, not consistency

confidence ≠ consistencyTalarico and Rubin, 2003
Trend sketch, not measurements. 54 students recorded how they heard about 9/11 and an everyday event. After 1, 6 or 32 weeks, consistency fell equally for both kinds of memory. But vividness and belief in accuracy fell only for the everyday memory; for 9/11 they held steady. consistency 1 → 6 → 32 weeks belief they remember correctly 1 → 6 → 32 weeks 9/11 memory everyday sketch of the trend reported in the paper, not measurements
Shocking memories are not more durable, they only feel more durable. The details drift like any other memory, but vividness and belief do not fade.Source: Talarico and Rubin, "Confidence, not consistency, characterizes flashbulb memories", Psychological Science 14(5): 455–461, 2003 (54 Duke students, retested after 1, 6 or 32 weeks).

Remembering a word that was never there

DRM paradigmRoediger and McDermott, 1995
Participants heard lists of words related to sleep, such as bed, rest, awake, tired, dream, snore, but not the word sleep. When writing down the words they heard, they wrote sleep 40% of the time. In a second experiment with longer lists the rate was 55%. bed rest awake tired dream snore blanket yawn ... 12 words, all circling one wordthat is never read out sleep not on the list 40%of the time "recalled"(experiment 1) 55%experiment 2,15-word lists
The brain stores the "theme" and fills in the most likely word. People also recognised the lure with high confidence. You can try it yourself in the game at the top of the page.Source: Roediger and McDermott, Journal of Experimental Psychology: Learning, Memory, and Cognition 21(4): 803–814, 1995 (40% in experiment 1, 55% in experiment 2); original idea by Deese (1959).

"Super-memory" people get fooled too

HSAMPatihis et al., 2013
20 people with highly superior autobiographical memory, who remember almost every day of their lives, and 38 ordinary people did the word-list task. False recognition of lure words: 70.3% for the super-memory group, 70.8% for controls. super-memory (20 people) 70.3% ordinary people (38) 70.8% rate of falsely recognising lure words in the word-list task
Remembering a lot does not mean remembering like a camera. People who can recall almost every day of their lives still reconstruct memories like everyone else.Source: Patihis et al., PNAS 110(52): 20947–20952, 2013; the 70.3% and 70.8% figures via BPS Research Digest (secondary source citing the paper).

John Dean: right on the gist, wrong on the details

gist vs. detailNeisser, 1981
John Dean, White House counsel, testified to the Senate in 1973 about meetings with President Nixon and was praised in the press for a tape-recorder memory. When the real tapes were released, Neisser compared them: who said what, the order, and Dean's own role were often wrong; but the core of what was going on, a cover-up and who was involved, was right. 1973 · Dean's testimony recounted each meetingwith Nixon in detail later: the real tapesare released details: often wrong who said whatorder of remarkshis own role(remembered as bigger) gist: right there was a cover-upwho was involvedwhat was going on
Just like a meeting at work. A week later we still remember "we agreed to do it", but who took the task and by when, everyone remembers differently. Neisser called this a "repisodic" memory: many meetings merged into one scene.Source: Neisser, "John Dean's memory: A case study", Cognition 9: 1–22, 1981.
Part 4

Fact check: the myth and its variants

10 common claims about memory, checked against the original research. Tap a line to see why.

SOLIDEvery act of remembering is a reconstruction

From Bartlett (1932) to the word-list experiments (Roediger and McDermott 1995): memories are assembled from fragments and general knowledge, so they can include details that never happened.

SOLIDLeading questions can change a memory

Loftus and Palmer (1974): asked with "smashed", 32% (16/50) reported broken glass that was not there; with "hit", 14%; not asked about speed, 12%.

SOLIDWe remember our own share of the work as bigger

Ross and Sicoly (1979) asked married couples, basketball teams and discussion groups: our own contributions come to mind more easily, so each person claims more responsibility than others give them. Dean also remembered his own role as bigger (Neisser 1981).

CAUTION"Memory can't be trusted at all"

Correcting a myth with the opposite myth is also wrong. The gist is usually right (Dean). Wixted and Wells (2017): under "pristine" identification conditions, a witness's INITIAL confidence tracks accuracy fairly well. What is unreliable is confidence that grows later, and small details.

CAUTION"It's easy to implant an entire false memory"

Loftus and Pickrell (1995): about a quarter (6/24) of participants partly or fully recalled a made-up "lost in the mall" event. A 2023 preregistered replication (123 people): 35%. But Andrews and Brewin (2025) reanalysed the data and argued false memories are much harder to implant than thought; many "memories" were people going along with the suggestion. The debate is still open.

CAUTION"One confident eyewitness is enough to convict"

37% of Americans agreed (Simons and Chabris 2011). Innocence Project: 69% of the first 375 DNA exonerations involved eyewitness misidentification. Yet initial confidence, recorded at once, under pristine conditions is still useful information (Wixted and Wells 2017). The problem is how and when the question is asked.

FALSEMemory records events exactly, like a camera

63% of Americans agreed in 2011; all 16 experts asked rejected it. No experiment shows the brain keeping a continuous recording to play back.

FALSEA vivid memory is an accurate memory

Neisser and Harsch (1992): match score 2.95/7 but confidence 4.17/5. Talarico and Rubin (2003): 9/11 memories lost detail like everyday memories, while vividness and belief stayed put.

FALSEEverything is stored forever and just needs the right cue (hypnosis, brain stimulation)

Loftus and Loftus (1980) reviewed all the evidence: none confirms it. Penfield's "film-like" scenes occurred in only 40 of 520 patients and could not be verified. On hypnosis, the American Medical Association's Council on Scientific Affairs (1985) concluded it increases the number of details recalled, including wrong ones, and increases confidence.

FALSEPeople with superior memory are immune to false memories

Patihis et al. (2013): false recognition of lure words was 70.3% in the super-memory group and 70.8% in controls. For details in a photo slideshow, the super-memory group made even more errors.

Sources: as given in each line; full list in Part 8. Loftus and Pickrell's 6/24 and the 2023 replication's 35% are via secondary sources.

Part 5

Assessing the myth

Scores are Mark's assessment based on the sources in Part 8, not figures from a study.

How wrong
4/5
How persistent
5/5
How harmful
4/5
The grain of truth
  • The gist of events is usually remembered correctly
  • Emotional memories really are more vivid
  • Initial confidence, recorded at once, is informative
What's wrong
  • There is no exact recording to rewind
  • Feeling "clear" says nothing about right or wrong
  • Later questions and retellings get written into memory

A very stubborn myth, because our own memory makes it feel true. The evidence against it dates from 1932, yet in 2011 nearly two thirds of those surveyed still believed it.

More: the cost outside the lab

In court, this myth makes confident eyewitnesses more believable than they should be. In the first 375 DNA exonerations in the US, 69% involved eyewitness misidentification (Innocence Project).

At work the cost is smaller but daily: "I remember you took this task", "No, I only said I'd look into it". Both are sincere, and both may be misremembering.

Source: Innocence Project, "DNA Exonerations in the United States (1989–2020)".

Part 6

Work lesson: 5-minute minutes

Don't rely on "I remember it clearly". Write the decisions down right after the meeting, while the memory has been retold little and nobody's questions have crept in.

The 5-minute minutes routine. Last minute of the meeting: read the decisions aloud. Within 5 minutes after: send 4 lines with decisions, who does what, deadline, still open. Same day: everyone replies agree or correct. Start of the next meeting: open the notes and read them back. 1 · last minuteread decisions aloud"so we've agreed..." 2 · within 5 minsend 4 linesdecision · who does whatdeadline · still open 3 · same dayeveryone confirms"agree" or "correct" 4 · next meetingopen the notesnot your memory "I remember it clearly" is not evidence. Notes the whole team has confirmed are.
Each step blocks one kind of misremembering from Part 3. Write early before the details drift (Neisser and Harsch), send to the whole team because everyone remembers their own share as bigger (Ross and Sicoly), and keep a record so you don't argue from a merged "gist" (Neisser on John Dean). This routine is Mark's suggested application; no study has tested "5-minute minutes" as such.
The 4-line templateDecision: … · Who does what: … · Deadline: … · Still open: … It doesn't need to be pretty, it needs to be fast.
Record how sure you are"Decided" is different from "tentatively agreed, pending numbers". Write it at the time, before confidence quietly grows.
Ask open, non-leading questions"What did we agree about the deadline?" rather than "We said Friday, right?". A question with the answer built in is the word "smashed".
Disagree? Open the notesTwo people who remember differently are usually both sincere. Don't settle it by who sounds surer.
Confidence, not consistency, characterizes flashbulb memories.Title of the paper by Jennifer Talarico and David Rubin, Psychological Science, 2003.
Part 7

Further reading

Daniel Schacter, The Seven Sins of Memory (2001)Seven ways memory "betrays" us, from forgetting to suggestibility, by a leading memory researcher.
Christopher Chabris and Daniel Simons, The Invisible Gorilla (2010)Includes a chapter on the illusion of memory; the same authors ran the 2011 survey.
Frederic Bartlett, Remembering (1932)The original work on memory as reconstruction. Old, but surprisingly right.
Famous Experiments: The invisible gorillaSame authors, on how we miss things right in front of us.
Common Myths #10: Sugar makes children hyperactiveThe last in the series: believe it first, and your eyes will see it.
Part 8

Sources

Original papers first. Where a figure could only be checked through a secondary source, this is stated next to it.

  1. D. J. Simons and C. F. Chabris, "What people believe about how memory works: A representative survey of the U.S. population", PLoS ONE 6(8): e22757, 2011. journals.plos.org
  2. E. F. Loftus and J. C. Palmer, "Reconstruction of automobile destruction: An example of the interaction between language and memory", Journal of Verbal Learning and Verbal Behavior 13: 585–589, 1974.
  3. U. Neisser and N. Harsch, "Phantom flashbulbs: False recollections of hearing the news about Challenger", in E. Winograd and U. Neisser (eds.), Affect and Accuracy in Recall, Cambridge University Press, 1992, pp. 9–31.
  4. J. M. Talarico and D. C. Rubin, "Confidence, not consistency, characterizes flashbulb memories", Psychological Science 14(5): 455–461, 2003. dukespace.lib.duke.edu
  5. H. L. Roediger III and K. B. McDermott, "Creating false memories: Remembering words not presented in lists", JEP: Learning, Memory, and Cognition 21(4): 803–814, 1995; and J. Deese, Journal of Experimental Psychology 58: 17–22, 1959.
  6. D. A. Gallo, M. J. Roberts and J. G. Seamon, "Remembering words not presented in lists: Can we avoid creating false memories?", Psychonomic Bulletin & Review 4(2): 271–276, 1997. link.springer.com
  7. L. Patihis et al., "False memories in highly superior autobiographical memory individuals", PNAS 110(52): 20947–20952, 2013.
  8. U. Neisser, "John Dean's memory: A case study", Cognition 9: 1–22, 1981.
  9. M. Ross and F. Sicoly, "Egocentric biases in availability and attribution", Journal of Personality and Social Psychology 37(3): 322–336, 1979.
  10. E. F. Loftus and G. R. Loftus, "On the permanence of stored information in the human brain", American Psychologist 35: 409–420, 1980.
  11. W. Penfield and P. Perot, "The brain's record of auditory and visual experience", Brain 86: 595–696, 1963; the 40/520 figure via the 2023 review "Brain stimulation and elicited memories" (secondary source).
  12. J. T. Wixted and G. L. Wells, "The relationship between eyewitness confidence and identification accuracy", Psychological Science in the Public Interest 18(1): 10–65, 2017.
  13. E. F. Loftus and J. E. Pickrell, "The formation of false memories", Psychiatric Annals 25: 720–725, 1995; G. Murphy et al., Memory, 2023 (replication); B. Andrews and C. R. Brewin, Scientific American, 2025 (reanalysis).
  14. Innocence Project, "DNA Exonerations in the United States (1989–2020)". innocenceproject.org
  15. Council on Scientific Affairs, AMA, "Scientific status of refreshing recollection by the use of hypnosis", JAMA 253: 1918–1923, 1985.
  16. F. C. Bartlett, Remembering: A Study in Experimental and Social Psychology, Cambridge University Press, 1932; Plato, Theaetetus 191c–d.