Mark ở vũ trụ khác · Another Universe Papers That Changed How We Learn #4 · Memory · Context

Divers learning underwater

Godden & Baddeley · British Journal of Psychology · 1975

The paper"Context-dependent memory in two natural environments: on land and underwater"
Findinglearn underwater and you recall better underwater; learn on land and you recall better on land
ReplicationMurre (2021) redid it with 16 divers: the old result did not appear
LabelCaution

Source: Godden and Baddeley (1975), British Journal of Psychology 66(3): 325–331; Murre (2021), Royal Society Open Science 8: 200724.

Recap in 1 minute

Essence

Learning on land and underwaterDivers heard word lists in one place, then recalled them in the same place or the other. Each did all 4 combinations.
1975: matching helpsWords learned underwater were best recalled underwater, and vice versa. The story made it into most textbooks.
2021: redone, not found16 divers, same design. Recall in the same place was no better than in the other place.
Real, but smallPooling many studies: context does help memory, modestly, and less when other cues are present.
A usable trickCan't go back to where you studied? Close your eyes and picture it. That makes up for most of the gap.

Source: Godden and Baddeley (1975); Murre (2021) (16 divers, remark on textbooks); Smith and Vela (2001), Psychonomic Bulletin & Review 8: 203–220 (meta-analysis, mental reinstatement of context).

Part 1

The paper's question

Does the place where you learn "stick" to the memory? If so, studying in the library and sitting the exam in a hall is making life hard for yourself. Two Scottish researchers picked the most different places possible to test it: on land and under the sea.

The question and framing are Mark's, based on the introductions of Godden and Baddeley (1975) and Murre (2021).

1975The originalMatching study and recall places helps.
→
1978–2001Tested in classroomsAn effect, but small; multiple choice barely affected.
→
2021–2022RedoneDivers: not replicated. Virtual reality: an effect.
Part 2

The story of the discovery

Who, where, how, and why divers.

Before 1975 · A chance observationIn studies of divers, people noticed that once back on land, divers remembered poorly what they had seen underwater. Could memory be tied to place?
The researchers · University of Stirling, ScotlandDuncan Godden and Alan Baddeley, two psychologists studying memory.
Four days of diving near ObanDivers listened to lists of unrelated words, on land or underwater, then freely recalled them in one of the two places. Each person did all 4 combinations.
Real conditions, not a labDone in the open sea. The authors themselves noted many distractions during the work.
1975 · The resultWords learned underwater were best recalled underwater, words learned on land best recalled on land. The recall place on its own did not decide it; matching did.
1980 · A second paper, on recognitionWhen divers only had to recognise words they had studied (rather than recall them), changing place did not hurt.

Source: Godden and Baddeley (1975) (4-combination design, result, no significant main effect of recall place); Murre (2021) (4 days diving near Oban, distractions noted by the original authors); Godden and Baddeley (1980), British Journal of Psychology 71: 99–104, per Simply Psychology (secondary source); the earlier chance observation per Wikipedia "Context-dependent memory" (secondary source).

How many divers, how many words?

Popular sources give 18 divers and a list of 36 two- and three-syllable words. Mark could only read the start of the original paper, not its data table, so these numbers are marked as secondary.

For the same reason this piece does not use the often-quoted "40% better" or "50% better": popular sources disagree, and Mark could not check them against the original table.

Source: Simply Psychology (18 divers, 36 words; secondary source).

Part 3

What the paper says, in 3 sentences

1. Divers learned word lists on land or underwater, then recalled them in the same place or the other.

2. They recalled more words when the recall place matched the learning place.

3. So the surroundings at learning become a "cue" that pulls the memory out; change them and you lose that cue.

Four-cell diagram showing the direction of the 1975 result, with no numbers. Rows are where words were learned: on land, underwater. Columns are where they were recalled: on land, underwater. The two matching cells are dark, more words. The two mismatched cells are pale, fewer words. recalled where? on landunderwater learned on landlearned underwater learned where? matchrecall more mismatchrecall less mismatchrecall less matchrecall more illustration of the 1975 direction, no numbers; every diver did all 4 cells
Not "underwater is worse", but "mismatch is worse". The clever part of the design: each person did all 4 cells, so they were compared with themselves.Source: Godden and Baddeley (1975), abstract: "lists learnt underwater were best recalled underwater, and vice versa"; 4-combination design per Murre (2021). Dark and pale cells are an illustration, not measured data.
Part 4

How this paper is rated

Every row has a source. Citation count: if it could not be found, we say so rather than guess.

PublishedBritish Journal of Psychology, vol. 66, no. 3, 1975 (British Psychological Society).Source: doi 10.1111/j.2044-8295.1975.tb01468.x.
CitationsNot found. Google Scholar was blocked in the working environment, and web search turned up no sourced figure.Checked 15 Oct 2026.
Textbook statusThe researcher who redid it writes that nearly every textbook he has used describes this experiment, usually in some detail.Source: Murre (2021), Royal Society Open Science.
Replicated?Murre (2021): 16 divers heard word lists on land or underwater and recalled them 4 minutes later. Recall in the same place was no better (p = 0.215). Stranger still: words learned underwater were recalled better on land. In 2022 a reanalysis found an interaction (p = 0.017) but not in the original's direction, so the author kept his conclusion: not a successful replication.Source: Murre (2021), RSOS 8: 200724; Murre (2022), addendum, RSOS 9: 211924.
Meta-analysisSmith and Vela (2001): environmental context effects are real but small; they shrink when stronger cues are present and when learners mentally reinstate the old context. For free recall the effect size is around d = 0.25 (small).Source: Smith and Vela (2001), Psychonomic Bulletin & Review 8: 203–220; effect size per Murre (2021).
Main critiques(1) The effect shows in free recall, hardly at all in recognition (Smith, Glenberg and Bjork 1978; Godden and Baddeley 1980 themselves). (2) Taking a test in a different room from class did not lower students' scores (Saufley et al. 1985). (3) The two diver studies differ in some details (dive site, number of days), and it is unclear which matter.Source: Smith, Glenberg and Bjork (1978), Memory & Cognition 6: 342–353; Godden and Baddeley (1980) via Simply Psychology (secondary source); Saufley, Otaka and Bavaresco (1985), Memory & Cognition, via a review (secondary source); Collabra: Psychology 9(1) (2023).
VerdictCAUTION The idea "context is a memory cue" holds up across the meta-analysis. But the effect is far smaller than the story, and the exact diver experiment failed to replicate. Mark's assessment.
Part 5

Fact check: what is true and what is myth

13 things often said about the diver experiment and "study where you'll be tested". Tap a row to see why.

Two runs compared. In 1975, matching study and recall places clearly helped. In 2021, no difference between match and mismatch, p equals 0.215; words learned underwater were recalled better on land. 1975 · Godden and Baddeley match: recall more mismatch 2021 · Murre, 16 divers match mismatch no difference (p = 0.215) 2021: learned underwater, recalled better on land (p < 0.01), the opposite of the original bar lengths are an illustration, not measured data
Same design, two results. Murre calculated that if the effect were as large as the original, his design would almost certainly have detected it.Source: Godden and Baddeley (1975); Murre (2021), RSOS 8: 200724 (p = 0.215; p < 0.01; statistical power above 99.9% if the effect matched the original).
SOLIDIn 1975 divers recalled better in the place where they learned

Original abstract: lists learned underwater were best recalled underwater, and vice versa.

SOLIDEach diver did all 4 combinations

Learn on land or underwater, recall on land or underwater (Godden and Baddeley 1975, per Murre 2021).

SOLIDThe recall place alone did not matter; matching did

The original reports no significant main effect of recall place; the difference lay in match versus mismatch.

SOLIDEnvironmental context effects are real, but small

Smith and Vela (2001), meta-analysis; effect size around d = 0.25 for free recall (Murre 2021).

SOLIDRecognition tests (pick the right word) are barely affected

Smith, Glenberg and Bjork (1978); Godden and Baddeley (1980), per Simply Psychology (secondary source).

SOLIDPicturing the place you studied makes up for a change of place

Meta-analysis by Smith and Vela (2001): the effect weakens when learners mentally reinstate the context.

SOLIDThe experiment appears in most textbooks

Murre's (2021) remark about the textbooks he has used.

CAUTION"It didn't replicate, so context effects are made up"

One failed redo does not erase a meta-analysis. Virtual reality experiments in 2021 and 2022 still found an effect (Shin et al. 2021; Essoe et al. 2022).

CAUTION"Divers remembered 40% (or 50%) more"

The circulating figure differs between popular sources; Mark could not check it against the original data table, so it is not used.

CAUTION"The 1975 experiment was tightly controlled"

Done at sea over 4 days; the authors themselves recorded many distractions (per Murre 2021).

CAUTIONLearning two easily confused things in two separate settings reduces mix-ups

Essoe et al. (2022): about 92% of words remembered after a week when each language was learned in its own setting, versus 76% in a shared setting. One study, in virtual reality, and only when learners felt the setting was "real".

FALSE"The 2021 redo confirmed the result"

The opposite: Murre (2021) found no effect, and the 2022 addendum kept the conclusion "not a successful replication".

FALSE"To remember, you must take the test in the room where you studied"

The effect is small, near zero for multiple choice, and offset by mental reinstatement. In real classrooms, testing in another room did not lower scores (Saufley et al. 1985, via a review; secondary source).

Source: as noted in each row; full list in Part 10.

Part 6

What the latest research says

Real divers did not replicate, but a "virtual sea" did.

2001 · Meta-analysisSmith and Vela pooled studies of environmental context: a reliable but small effect; picturing the context nearly removes it.
2021 · Divers againJaap Murre (University of Amsterdam), 16 divers, recall after 4 minutes. No advantage for matching. Data made public, and in 2022 he published his own addendum fixing the analysis: the conclusion stood.
2021 · Mars and the seabed in virtual realityShin et al. (Princeton): recall in the same virtual world was better, especially for items that fit that world's "script".
2022 · Two languages, two settingsEssoe et al. (UCLA): 80 words from two similar languages. One setting per language: about 92% remembered after a week; one shared setting: 76%. It only worked when learners felt the setting was "real".

Source: Smith and Vela (2001); Murre (2021), RSOS 8: 200724 and addendum (2022), RSOS 9: 211924; Shin, Masís-Obando, Keshavarzian, Dáve and Norman (2021), Psychonomic Bulletin & Review 28: 574–582; Essoe, Reggente, Ohno, Baek, Dell'Italia and Rissman (2022), npj Science of Learning 7: 31.

Part 7

Verdict

Scores are Mark's assessment based on the sources in Part 10, not data from a study.

Idea and design
5/5
Strength of the 1975 result
2/5
General idea still standing
4/5
Directly useful to learners
3/5
Strengths
  • Two extremely different, real-world settings
  • Every person did all 4 combinations
  • Opened a whole line of research
Weaknesses
  • Small sample, sea conditions hard to control
  • The 2021 redo found no effect
  • Story inflated into "you must test where you study"

An experiment so neat that everyone remembers it: right in its big idea, but its size has been told bigger than the truth.

More: why is this story so easy to overstate?

Divers studying on the seabed is a vivid, easy-to-tell image. Strong images get stretched beyond the evidence: from "matching helps a little" to "study in the exam hall".

The best lesson may be how science corrects itself: someone redid the exact experiment, shared the data, and published his own addendum when a flaw in the analysis was pointed out.

The "More" box is Mark's commentary, based on sourced facts in Parts 2 to 6.

Part 8

Lessons: don't let your memory depend on your seat

Context is a small memory cue. Use it when you can, and don't come to need it.

Four habits: picture your study place before the test; do the last review in exam-like conditions; study two easily confused things in two different places; quiz yourself in unfamiliar places. 1 · picture it30 seconds, eyes shutyour desk, the sounds 2 · practise like the testquiet, timedbook closed on the last review 3 · split placestwo confusable subjectstwo different places 4 · quiz somewhere oddcafé, bus, yardrecall anywhere The goal is not to study in the right place, but to remember in any place.
Each habit rests on a sourced result. Mental reinstatement of context (Smith and Vela 2001); separate settings for confusable material (Essoe et al. 2022, still caution); the effect is small, so the goal is recall anywhere.Source: as noted; habits 2 and 4 are Mark's suggestions.
Picture it before the testShut your eyes for 30 seconds: the desk, the light, the noise, the page. Then read the questions.
Last review like the testQuiet, timed, book closed, at a desk like the exam desk. The more alike, the fewer surprises.
Two confusables, two placesFrench at the desk, Spanish in the library. There is only one study so far, so test whether it suits you.
Don't worry about the exam roomThe effect is small and multiple choice is barely affected. Reviewing well matters more than where you sit.
Self-check: the "change places" test in 10 minutes

1. Learn 15 words or 5 key ideas in your usual spot (your desk).

2. The next day, quiz yourself somewhere completely different (balcony, café). Note how many you recall.

3. Quiz again at your usual spot. Big gap? You are leaning on cues from your seat.

4. If there is a gap: review in 2–3 different places, and practise "picturing it" before answering.

The test is Mark's suggestion, not a procedure from a study.

Part 9

Read next

Murre (2021), "The Godden and Baddeley (1975) experiment on context-dependent memory on land and underwater: a replication"Royal Society Open Science. Free to read, and a clear account of redoing a classic.
Smith and Vela (2001), "Environmental context-dependent memory: A review and meta-analysis"Psychonomic Bulletin & Review. The big picture on context effects.
Decoding Super Memory (book #4 on Vũ Trụ Khác)Another way to use places for memory: the memory palace. Read it
Papers That Changed How We Learn #05: Cognitive load (Sweller 1988)Why solving lots of problems doesn't always make you better.
Part 10

Sources

The main sources are the original paper, the replication and a peer-reviewed meta-analysis. The working environment blocked journal sites, PubMed Central and Google Scholar, so figures were checked against those sites' own excerpts in search results, without opening the full text; the original paper's data table could not be read. Secondary sources are marked.

  1. Godden DR, Baddeley AD (1975), "Context-dependent memory in two natural environments: on land and underwater", British Journal of Psychology 66(3): 325–331. Wiley
  2. Murre JMJ (2021), "The Godden and Baddeley (1975) experiment on context-dependent memory on land and underwater: a replication", Royal Society Open Science 8: 200724. royalsocietypublishing.org · PMC
  3. Murre JMJ (2022), "Addendum to: Murre (2021)...", Royal Society Open Science 9: 211924. PMC
  4. Smith SM, Vela E (2001), "Environmental context-dependent memory: A review and meta-analysis", Psychonomic Bulletin & Review 8: 203–220. Springer
  5. Smith SM, Glenberg A, Bjork RA (1978), "Environmental context and human memory", Memory & Cognition 6(4): 342–353.
  6. Godden DR, Baddeley AD (1980), "When does context influence recognition memory?", British Journal of Psychology 71: 99–104 (results per secondary source).
  7. Saufley WH, Otaka SR, Bavaresco JL (1985), "Context effects: Classroom tests and context independence", Memory & Cognition 13 (results per a review, secondary source).
  8. Shin YS, Masís-Obando R, Keshavarzian N, Dáve R, Norman KA (2021), "Context-dependent memory effects in two immersive virtual reality environments: On Mars and underwater", Psychonomic Bulletin & Review 28: 574–582. PMC
  9. Essoe JKY, Reggente N, Ohno AA, Baek YH, Dell'Italia J, Rissman J (2022), "Enhancing learning and retention with distinctive virtual reality environments and mental context reinstatement", npj Science of Learning 7: 31. nature.com
  10. Collabra: Psychology 9(1) (2023), "Assessing the replicability of cognitive psychology during remote experiential learning via mobile phone technology" (auxiliary assumptions of the diver replication).
  11. Secondary sources: Simply Psychology, "Context and state-dependent memory" (18 divers, 36 words; the 1980 result); Wikipedia "Context-dependent memory" (the chance observation before 1975).