/ Library/ What Should We Remember/ Context-Dependent Memory
Recall Bias № 031 · Last updated 13 May 2026

Context-Dependent Memory.

"We recall things best in the environment we learned them in."

01Overview

Context-dependent memory is the phenomenon where recall is significantly better when the retrieval environment matches the encoding environment. Scuba divers who learned word lists underwater recalled them better underwater; divers who learned on land recalled better on land. The context in which information is learned becomes a cue for retrieving it later.

For designers, this means that a feature learned in a tutorial modal may not transfer to the live interface — the context has changed, and the memory may not fire in the new environment.

02Detailed explanation

Godden and Baddeley's 1975 underwater scuba study is the canonical demonstration: divers learned word lists in two environments (underwater and on land) and were tested in both. Words learned underwater were recalled approximately 50% better in the matching underwater context than in the mismatched on-land context.

  • The mechanism is encoding specificity: memory traces include contextual information as part of their storage structure. Context is not just decoration on a memory — it is part of the retrieval key.
  • State-dependent memory extends the principle: mood, arousal, even blood glucose at the time of learning affect recall under matching vs. mismatched states.
  • Interface context applies: the visual environment, the URL structure, the surrounding UI elements, and even the device type all form part of the learning context.

03Why it exists

The brain stores memories with their context as part of the trace. Context is not just background information — it is included in the memory encoding because it was present and potentially relevant when the memory formed. When the same context recurs, it provides additional retrieval cues that make the memory easier to access.

The short version

If the context changes between learning and doing, the memory may not transfer. Teach in the environment where the action will be taken — not in a special tutorial layer that looks nothing like the real product.

04Effects on users

  • Users who learn a workflow in a product tour may not perform it in the real interface — the tutorial environment was the learning context, not the live product.
  • Mobile users who learn a feature on desktop may not find or recall it on mobile — the interface context has changed significantly, even if the feature is present on both.
  • Help documentation read outside the product context is less effectively recalled when the user returns to the product than equivalent in-product contextual help.
  • The "I set this up once and can never remember how" problem in settings is partly a context issue: setup happened in a high-attention state; recall happens in the middle of trying to do something else entirely.

05Effects on designers & teams

  • Usability testing: participants perform better in familiar environments (their own workspace, their own device) than in lab settings. Results from lab studies may systematically underestimate real-world performance.
  • Design system adoption: training on a design system works better in the tools where the system will be used than in a separate training session with slides.
  • Remote research: participants joining from home offices often recall interaction details more accurately than equivalent participants in structured testing environments, because the home environment more closely matches their typical usage context.

06Practical takeaways

  • Teach features in the live product context, not in a demo or tutorial overlay with a different visual environment.
  • Progressive disclosure at point-of-use leverages context-dependent memory: the teaching moment happens in the same context as the doing moment.
  • For cross-device products: assume learning on one device does not fully transfer to another without re-education — the context has changed enough to matter.
  • Help content embedded in the product (contextual tooltips, inline documentation) is recalled better than equivalent help opened in a separate browser tab.
  • Test in conditions as close to the natural use environment as possible — this is both a research validity issue and a product design signal.

07Design examples

Onboarding

In-product tooltips vs. product tours

A product tour in a dedicated modal is a different visual context from the live product. Users often "forget" what the tour taught them when they return to the real interface. In-product tooltips shown in the live context are better recalled because the context matches the action.

Cross-device

Mobile vs. desktop learning

A user who sets up integrations on desktop will likely need re-guidance when accessing the same settings on mobile. The interface context has changed enough that context-dependent memory means the learning may not fully transfer.

Help systems

Embedded vs. external help

A help article opened in a new browser tab is in a different context from the product UI where the problem occurred. In-product help panels and contextual tooltips are better recalled and applied because they share the problem context.

Research

Lab vs. natural environment

Remote moderated research, where users work in their own environment on their own devices, captures more ecologically valid behaviour than lab studies. Context-dependent memory is one reason: familiar environments cue familiar behaviours.

08Ethical risks

Context-dependent memory is relevant to consent flows: a consent screen that appears in a distinctly different visual context from the product UI may be less connected to the user's mental model of what they're agreeing to. This can work in either direction — the distinct context may increase attention, or it may reduce the user's ability to relate the consent to their actual product experience.

Informed consent in design means ensuring the context supports genuine understanding of what's being agreed to — not just legibility of the text.

10Suggested reading