To what extent can brief, embedded self-explanation prompts inside interactive visuals (e.g., “Explain in words why this output changed when you moved that slider”) substitute for or augment more formal retention checks (quizzes, concept maps) in reducing illusion-of-understanding and improving long-term conceptual retention, relative to static worked examples that use the same self-explanation prompts without manipulation?

Answer

Embedded self-explanation prompts inside interactive visuals partially substitute for but more powerfully augment formal retention checks. By themselves, they reliably reduce illusion-of-understanding within the interactive session and modestly improve long-term conceptual retention relative to worked examples with the same prompts, but they do not fully replace spaced quizzes or cumulative concept-mapping when the goal is strong far transfer and very durable learning.

More concretely:

1. As a substitute for formal retention checks

• In interactive visuals, brief, tightly-timed self-explanation prompts ("Explain why the curve got steeper when you increased X") that directly follow a prediction–manipulation–feedback cycle:
  • Reduce illusion-of-understanding more than the same prompts in static worked examples, because learners must connect their own actions to outcomes when explaining, rather than paraphrasing a fixed solution path.
  • Produce moderate gains in delayed retention and near transfer over worked examples with identical prompts but no manipulation, especially for relations that are directly visualized.
• However, compared with designs that also use spaced single-concept quizzes or periodic concept maps, relying only on embedded self-explanation prompts:
  • Leaves more gaps in cumulative understanding across multiple concepts, because the prompts are highly local and tied to the immediate state of the visual.
  • Yields weaker very-long-delay retention and far transfer, where cross-topic coordination and retrieval from scratch matter.
• So, as a standalone replacement for quizzes/concept maps, embedded self-explanations are better than no checks and better than static self-explanation alone, but not sufficient if the target is robust far transfer and month-scale durability.

2. As an augmentation to formal retention checks

• When an interactive visual includes both embedded self-explanation prompts and sparse, well-designed retention checks (e.g., a short quiz after a block of related manipulations, or a small concept map summarizing the relations just explored):
  • The self-explanations clean up local illusions-of-understanding (outcome-chasing within the visual) by forcing learners to articulate why a particular manipulation produced a specific outcome.
  • The spaced quizzes or concept maps then test whether those local explanations have consolidated into a coherent schema that learners can retrieve without the crutch of the current visual state.
  • This combination typically outperforms both:
    • interactive + self-explanation without quizzes, and
    • static worked examples + self-explanation + quizzes, on delayed conceptual retention and far transfer tasks.

3. Relative advantage over static worked examples with the same prompts

• With static worked examples, self-explanation prompts mainly encourage verbal rehearsal of given steps or relations. Learners often explain what the example shows rather than what would happen under different conditions.
• In interactive visuals, the same kind of prompt, when triggered immediately after a learner-driven change, tends to:
  • Evoke counterfactual reasoning ("If I had moved X less, what would have happened?") and focus attention on functional relations, not just one instantiation.
  • Increase productive struggle by tying explanation opportunities to moments of prediction error or surprise.
• As a result, for the same total time-on-task and same literal prompt wording, interactive + embedded self-explanation usually yields:
  • Lower illusion-of-understanding rates (when measured by mismatch between in-session performance and delayed tests).
  • Higher near- and mid-delay conceptual retention, particularly for novice and intermediate learners.

4. Design conditions under which embedded self-explanation works best Embedded self-explanation prompts inside interactive visuals are most effective when:

• Each prompt is immediately tied to a small, interpretable change (ideally one-variable-at-a-time) rather than to a long, complex manipulation sequence.
• Prompts follow a prediction–action–feedback cycle, so learners must explain either why their prediction was correct or why it failed.
• Prompts are brief and infrequent (e.g., after key milestones or notable pattern changes), to avoid fragmenting exploration and overloading working memory.
• Some prompts explicitly require learners to generalize beyond the current state ("In general, when X increases, what tends to happen to Y and why?"), pushing them from local description to rule-level understanding.

5. Where formal retention checks still add unique value Even with strong embedded self-explanation, formal retention checks (quizzes, small concept maps) add critical capabilities that the prompts alone rarely provide:

• Context-free retrieval: Quizzes delivered outside the immediate manipulation context require learners to recall relations without perceptual cues, strengthening durable learning.
• Cross-concept integration: Concept maps or cumulative quizzes can connect multiple relations (e.g., how changing X affects both Y and Z over time), something local self-explanations rarely demand.
• Spacing and consolidation: Scheduled checks at increasing delays exploit spacing and retrieval practice effects more systematically than opportunistic in-the-moment prompts.

Putting it together: brief, embedded self-explanation prompts in interactive visuals are a powerful, low-friction tool for reducing illusion-of-understanding and boosting conceptual retention beyond what static worked examples with identical prompts usually achieve. However, they are best treated as a first line of defense and local amplifier, not as a full replacement for well-spaced, cumulative retention checks when the instructional goal is durable, far-transferable understanding.