Context-Dependent Memory: Why Where You Learn Changes What You Remember

In 1975, Duncan Godden and Alan Baddeley took a group of divers off the Scottish coast and asked them to learn lists of unrelated words. Some learned the lists on dry land, some learned them roughly six metres underwater. Then each group was tested in both places. The result is one of the tidiest findings in memory research: recall was substantially better when the environment at test matched the environment at learning. Words learned underwater came back underwater. Words learned on the beach came back on the beach. Switching contexts cost a meaningful chunk of what had been learned.

The experiment is memorable because of the diving suits, but the mechanism is ordinary. When you encode a memory, you do not store the target alone. You store it entangled with whatever else your mind happened to be processing: the room, the light, the noise, your posture, your mood, the smell of the coffee. Later, those incidental details can act as retrieval cues, small handholds that help you find the thing you were actually trying to remember.

Two kinds of context

Researchers usually separate the external from the internal, because they behave slightly differently.

• Environmental context: the physical setting. Room, background sound, visual surroundings, even the device you were reading on. This is what the diver study manipulated, and it is the version most people mean by context-dependent memory.
• State-dependent context: your internal condition. Mood, arousal, caffeine, fatigue, and in some studies substances. Material learned in a particular internal state tends to be somewhat easier to retrieve in that same state, which is one reason revising while calm and testing while panicking feels so unfair.

There is a third relative worth knowing about. Context effects are much larger for free recall, where you have to generate the answer with no support, than for recognition, where the material is in front of you and you only have to identify it. If the cue is already supplied by the question, the room matters less. This is a useful caveat: context dependence is real, but it is not equally strong for every kind of remembering.

Context is why forgetting sometimes has a doorway in it

The same machinery explains a very common domestic experience. You walk into the kitchen with a clear purpose and arrive with none, because crossing a boundary shifted the context your intention was attached to. That is covered properly in the doorway effect explainer, and it is essentially context-dependent memory operating on a timescale of seconds rather than days.

Deliberate use of the same principle is what makes the memory palace technique work. Instead of hoping helpful context attaches itself by accident, you choose a familiar place and hang the material on specific locations within it, then walk the route to retrieve. It is context dependence turned from a bug into a tool.

The counterintuitive practical advice

Here is where people usually draw the wrong conclusion. If matching contexts helps recall, the obvious advice is to reinstate the learning context at test: study in the exam hall, revise at the desk where you will be assessed, drink the same coffee. Sometimes this is possible and mildly useful. Usually it is not, and more importantly it is the wrong goal.

What you actually want is knowledge that does not depend on any particular room, because real use happens in unpredictable places. A conversation at work, a question from a child, a decision in a shop. Knowledge welded to one armchair is fragile knowledge. So the better strategy is contextual variation: study the same material in several different settings, so that no single set of cues owns it. Each new context adds another retrieval route, and the material gradually becomes portable.

This is the same family of ideas as interleaving and spacing. Varying conditions makes each individual session feel slightly harder and less fluent, and produces better long-term retention anyway, which is why it belongs alongside the schedule effects described in the spaced repetition explainer. Difficulty during practice is a poor guide to durability, and this is one of the clearest cases of that mismatch.

What this means for phone-based learning

Anyone learning on a phone is already running the varied-context experiment without meaning to. You read something on a train, review it on a sofa, get tested on it standing in a queue, revisit it in a waiting room. That scattering is often described as a weakness of mobile learning, and for concentration it can be. For context dependence it is an advantage: the material never gets a chance to bind to one setting.

One extra trick worth knowing, because it comes from the same literature: if you cannot physically return to a context, you can partly reinstate it mentally. Deliberately recalling where you were, what was around you and what you were doing when you learned something often unsticks a memory that felt gone. Cognitive interviews used with witnesses rely on this, and it works reasonably well for the everyday version of forgetting a name.

Practical rules

1. 1Rotate your study locations rather than optimising one perfect desk. Three ordinary settings beat one ideal one for durability.
2. 2Vary time of day and internal state a little. Learning only in one narrow state ties the material to that state.
3. 3Test yourself somewhere you did not learn. If recall collapses when the room changes, the knowledge was leaning on context rather than standing on its own.
4. 4When you are stuck, reinstate the context in your head before giving up: where you were, what you were doing, what came just before.
5. 5Do not confuse fluency with learning. A session that feels harder because the setting is unfamiliar is frequently the more valuable one.

Disclosure: MindSnap is our app. Its short story-driven lessons, across five flagship collections plus unlimited topics, happen to travel well between contexts because they are short enough to fit anywhere, and the Turning Points scenarios ask you to retrieve and decide rather than reread. That is a design consequence of this research rather than a claim of superiority over any other format.