Protocol
Change the surface before claiming transfer
Transfer starts when the costume changes and the rule still works.
When it fits
- When a learner succeeds on practice cases that closely resemble the examples.
When to avoid it
- Near transfer is not proof of broad expertise. Increase variation gradually and keep domain limits explicit.
Why it matters
Keep the underlying principle but alter names, numbers, order, surface features or surrounding context. Ask the learner to identify the same mechanism and produce the correct action. Use failure to see whether practice encoded a reusable rule or a template match.
Steps
- Choose a case the learner can already solve.
- Change surface details while preserving the governing principle.
- Ask for the decision and the reason.
- If performance collapses, compare the stable principle with the changed surface cues.
An example
After learning a join error on customer data, the next case uses product data with different field names but the same key-grain problem.
Check your result
The learner succeeds after meaningful surface change and can name the invariant that transferred.
Keep this limit in mind
- Near transfer is not proof of broad expertise. Increase variation gradually and keep domain limits explicit.
Evidence and sources
A 2021 systematic review of 50 classroom experiments found retrieval practice benefited learning across varied educational settings, with most included effects in the medium-or-large range.
Retrieval is not guaranteed to improve every kind of inference or transfer, and the review had limited non-WEIRD representation.
A meta-analysis of interleaving found a moderate overall benefit (g = 0.42) but strong moderation by material type and category similarity, including conditions where blocking performed better.
Interleaving should be selected for a discrimination problem rather than used as a universal mixing rule.
Similarity matters: A meta-analysis of interleaved learning and its moderators · Abstract