When interactive visual explanations are used to pre-train learners before ordinary tutoring and worked examples versus to follow up after those explanations, how does this ordering change productive struggle patterns (e.g., outcome-chasing vs systematic one-variable-at-a-time tests) and long-delay retention and far transfer, holding total time and content constant?

Answer

Using interactive visual explanations as follow-up after tutoring/worked examples more reliably produces systematic, relation-focused productive struggle and stronger long-delay retention and far transfer than using the same interactive visuals as pre-training before tutoring, when total time and content are held constant.

1. Productive struggle patterns

• Pre-training with interactive visuals → more outcome-chasing risk
  • Novices enter the visual with weak schemas; even with good intentions, they tend to explore via rapid multi-variable sweeping and target-matching ("make it look right").
  • This pre-tutoring phase amplifies outcome-chasing patterns already known to predict illusions-of-understanding, unless the pre-training is very tightly constrained (OVAT, contrasting cases, prediction gates).
  • When later tutoring and worked examples arrive, they often function as repair for misconceptions and messy mental models formed during unguided or loosely guided pre-training.
• Follow-up with interactive visuals → more systematic OVAT testing
  • Tutoring/worked examples first supply a rough causal schema and highlight key variables and relations.
  • Subsequent interaction then more often shows the systematic OVAT, reuse-of-informative-cases, and prediction–test–explanation cycles that prior work links to durable learning, because learners now have hypotheses to test instead of searching blindly.
  • Productive struggle shifts from “what matters?” exploration to “how exactly does this relation behave?” refinement, with more focused one-variable changes and longer dwell times per configuration.

2. Long-delay retention and far transfer

• Matched time/content, typical learners
  • With realistic levels of constraint and prompting, tutor/works → interactive follow-up yields higher delayed retention and far transfer than interactive pre-training → tutor/works, because:
    • Less time is spent in high-load, poorly structured interaction.
    • More of the interactive time is used to instantiate and refine an existing model rather than building one from scratch via trial-and-error.
  • Pre-training can match follow-up only when interaction is strongly scaffolded to prevent outcome-chasing (e.g., forced contrasting-case toggling and prediction-before-manipulation gates) and the later tutoring explicitly re-organizes what was seen.
• Illusion-of-understanding
  • Pre-training with rich visuals often inflates early performance and confidence, especially on in-visual retention checks, yet yields weaker performance on long-delay, far-transfer tasks than the follow-up ordering.
  • Follow-up use, in contrast, tends to expose gaps (by confronting the learner’s verbal/analytic model with concrete manipulation) and, when paired with sparse embedded checks, reduces illusions-of-understanding more consistently.

3. Edge cases

• For higher-prior-knowledge learners, pre-training with tightly constrained, prediction-focused visuals can be beneficial, but these same learners also gain substantially from follow-up use; the net ordering advantage shrinks but generally still favors tutor/works-first.
• For very low-prior-knowledge learners, interactive pre-training without heavy constraints is especially likely to generate unproductive struggle and fragile understanding; for them, pre-training is rarely superior and is often worse than follow-up use.

Net expectation (holding time and content constant):

• Best default: Brief, focused tutoring/worked examples first, then a constrained interactive visual phase designed around OVAT changes, informative contrasting cases, and prediction–explanation prompts.
• Higher risk: Interactive visual pre-training before tutoring, unless the pre-training interaction is heavily structured to suppress outcome-chasing and the subsequent tutoring explicitly restructures what learners saw.