In 1975, members of a university diving club learned lists of words, either sitting beside the water or about 20 feet under it. Then they tried to recall the words, either in the same place or in the other one. Results like this sit behind a common piece of advice: study in conditions like the exam’s, same silence, same time of day, ideally the same room. The full body of evidence says the room matters less than that advice assumes, and that studying in more than one place may serve you better than matching one.
The previous post was about mnemonics, cues you build on purpose. This one is about cues you never chose. The room, the background noise and other features of your surroundings count as context in this research (Van Hoof et al., 2025), and context can attach itself to what you learn. Psychologists call the result context-dependent memory. Context belongs to a different category from the strategies earlier in this series: a strategy builds knowledge, while context changes whether you can reach that knowledge later.
Divers and Word Lists
The diving study was run by Godden and Baddeley (1975, as cited in Van Hoof et al., 2025). When the place of learning and the place of recall matched, the divers recalled about a third of the words. When they differed, they recalled about a quarter. The surroundings had nothing to do with the task, yet changing them cost the divers words.
The usual explanation is the encoding specificity principle of Tulving and Thomson (1973): how a memory can be retrieved depends on how it was stored. The divers stored the setting along with the words, so the setting became one of the cues that could bring the words back.
There is a detail the popular retellings leave out. A later diving study found that the change of setting affected recall, where you produce the words from memory, but not recognition, where you pick studied words out from new ones (Godden & Baddeley, 1980, as cited in Smith & Vela, 2001). It was a first hint that the effect depends on how memory is tested.
A Real Effect, Mostly Measured in the Lab
To see how general the room effect is, Smith and Vela (2001) ran a meta-analysis, a statistical method that pools results across many studies. They combined 93 estimates of the effect (separate room-change comparisons) from 75 studies published between 1935 and 1997. Most compared people tested in the room where they learned with people tested in a different room; internal states such as mood were left out. Their verdict was that context “has a modest … but reliable” effect on memory (p. 213). The effect is real, and on average it is modest.
The average hides a wide spread. Under some conditions the effect was close to zero; under others it was considerably larger. Effects were also larger when a different experimenter ran the test along with the different room, a factor Smith and Vela single out as important.
The recall and recognition split needs a qualification. Across all studies, recall and recognition showed effects of similar size. Smith and Vela trace this to the materials: the diving studies used word lists, which are easy to link together, while the clear recognition effects came mostly from less associable material such as faces. With word lists on both tests, recall still depended on the setting more than recognition did.
The most useful caveat for students is about where this evidence comes from. Smith and Vela left classroom studies out of the meta-analysis, mainly because students in a course are free to study in other places between the lesson and the test. Early classroom field experiments had found no context effect (Saufley et al., 1985, as cited in Smith & Vela, 2001). So the question every exam-taker asks, whether a strange testing room will hurt their score, has a laboratory answer more than a classroom one.
When the Room Stops Mattering
One of the firmest patterns in Smith and Vela’s data concerns what learners did while studying. When the study task had people relate the items to one another, for example by sorting them or linking them in pairs, the room effect rarely rose above zero.
Smith and Vela offer two explanations. In overshadowing, attention to the links between items crowds out the encoding of the room. In outshining, learners later use those links as cues during the test, and those cues outshine the weaker ones the room supplies. The data gave each explanation only partial support, so the mechanism is still open.
A second finding is more practical. When people tested in an unfamiliar room were asked to picture the room where they had studied, their memory improved. Smith and Vela call the support for this mental reinstatement clear. It rests on ten estimates, some from eyewitness studies with more elaborate procedures. In Smith’s own list-learning studies, participants simply visualized the study room and listed objects in it (Smith, 1979, 1984, as cited in Smith & Vela, 2001).
The first finding suggests a prediction. Elaboration, elaborative interrogation and self-explanation, covered earlier in this series, all ask you to connect ideas to one another. If relating items loosens a memory’s tie to the room, these strategies might do the same, though that has only been shown with laboratory word lists and not with the explanations and principles students actually learn.
Varying Instead of Matching
Start with a study closer to medical school. In a single study, first-year medical students attended one lesson in scrubs and another in their own clothes, then took a retention test five weeks later in their own clothes (Finn et al., 2010, as cited in Van Hoof et al., 2025). Performance fell more than twice as much for the content learned in scrubs.
James Nairne (2002) explains why physically matching the room is a weak strategy. He argues that the link between encoding–retrieval match (how much the test situation resembles the study situation) and good memory is “essentially correlational” (p. 389): the two tend to go together, but the match is not what does the work. What decides recall is how well a cue points to one memory and not to others. His example of a poor cue is contextual features, because they are shared by many events (p. 391). Your usual desk is linked to everything you have ever studied there, so it does little to single out the one fact you need.
Varying the setting has some direct support, all of it from retention tests. In a study by Smith, Glenberg, and Bjork (1978, as described in Soderstrom & Bjork, 2015), participants studied a list of words twice, three hours apart, either in the same room both times or in two different rooms. Tested later in a third, neutral room, those who had studied in two rooms recalled more. Soderstrom and Bjork report that the result was later replicated with the same or similar materials.
A second study used more complex material. People retained statistical concepts better after five days when four lectures had been given in four different places instead of one (Smith & Rothkopf, 1984, as cited in Soderstrom & Bjork, 2015). Tests given right after each lecture showed similar results for both groups, so the advantage appeared only at the delayed test.
Van Hoof et al. (2025) go further and argue against matching altogether. Citing a chapter by Roediger and colleagues, they write that “simply matching learning and retrieval contexts is not sufficient to improve memory” (p. 179). They also call a consistent setting undesirable, because health professionals have to use what they learned in many settings. A narrower reading fits the evidence better. Matching and varying serve different needs. In laboratory studies, picturing the study setting helped people tested in an unfamiliar room. Studying in more than one place suits knowledge you will need in many places.
All of these studies measure retention: recalling the same material in a new setting. None of them tested whether varied settings help you apply knowledge to new problems, which is transfer.
Using Context on Purpose
For practice, Van Hoof et al. (2025) name three dimensions they believe are likely the most actionable: how, where and when you engage with the material. For a student, that could mean reading a topic one day and explaining it aloud or testing yourself on it the next, in a different room, at a different hour. And when an exam will be in an unfamiliar place, take a minute at the start to picture where and how you studied.
Some of this variety may come free. Van Hoof et al. assert, citing other work but without a direct comparison, that spacing, retrieval practice, interleaving and elaboration make learners less reliant on any one context. Spacing sessions across days varies the time at least, and the place too if you let it.
If you teach or design a course, the same three dimensions apply to your material. Van Hoof et al. suggest pairing readings and lectures with discussion and reflection, offering activities both inside and outside the usual setting, and mixing live and self-paced options. Smith and Rothkopf’s four-lecture study suggests that even moving a lecture to a different room is worth trying when it costs little, though one study is a thin basis for a rule.
What This Looked Like for Me
For a long stretch of medical school I had a fixed study ritual: same library carrel, same headphones, same playlist, same hours. It felt productive, and I did cover material. On exam days, though, sitting in an unfamiliar testing center without any of those cues, the material sometimes felt slightly out of reach.
Over time I broke the ritual on purpose. I wrote out mechanisms at the kitchen table, talked through cases on walks, quizzed myself on the wards between patients and read at odd hours instead of in one protected block.
I changed several things in that period, spacing, retrieval practice and interleaving among them, so I can’t credit the change of scenery alone. What I trusted by the end was material I had gone over in more than one place.
The Core Lesson
The setting you study in becomes part of the memory, and a change of setting can cost you some recall. Across laboratory studies that effect is modest on average, it shrinks when learners relate items to one another, and picturing where you studied can partly restore it. Classroom evidence is scarce, and the evidence for varied study covers retention only. Instead of trying to recreate the exam room, vary how, where and when you study, and use mental reinstatement when the test is somewhere new.
The next post turns to Bloom’s taxonomy, a framework for asking what kind of knowing a learning goal requires.
This post is part of a series on the science of learning. Each post covers one evidence-based principle and how to apply it to your own studying. Follow us on Instagram @acemeded to keep up with future blog posts and related content.
References
Nairne, J. S. (2002). The myth of the encoding–retrieval match. Memory, 10(5–6), 389–395. https://doi.org/10.1080/09658210244000216
Smith, S. M., & Vela, E. (2001). Environmental context-dependent memory: A review and meta-analysis. Psychonomic Bulletin & Review, 8(2), 203–220. https://doi.org/10.3758/BF03196157
Soderstrom, N. C., & Bjork, R. A. (2015). Learning versus performance: An integrative review. Perspectives on Psychological Science, 10(2), 176–199. https://doi.org/10.1177/1745691615569000
Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80(5), 352–373. https://doi.org/10.1037/h0020071
Van Hoof, T. J., Sumeracki, M. A., & Madan, C. R. (2025). Science of learning strategy series: Article 7, the role of context in learning. Journal of Continuing Education in the Health Professions, 45(3), 178–181. https://doi.org/10.1097/CEH.0000000000000601
