In interactive visual explanations that already include prediction-before-manipulation and manipulation budgets, does temporarily hiding direct numeric feedback (e.g., exact values, error magnitudes) during key variable sweeps reduce illusion-of-understanding and improve delayed, out-of-context retention and far transfer more than adding additional embedded comparative prompts, and for which concept types does withholding numbers tip productive struggle into opaque guessing?

Answer

Hiding numeric feedback during selected variable sweeps probably reduces illusion-of-understanding and can slightly improve delayed retention and far transfer over just adding more embedded comparative prompts, but only for concepts where qualitative patterns are visually clear and where learners already have minimal schemas. For opaque or heavily quantitative concepts, or for fragile novices, hiding numbers is likely to turn productive struggle into guessing and harm learning.

Use hiding-then-reveal sparingly: brief runs where learners must predict direction/relative size, then see the exact numbers and a short comparison prompt. For multivariate, visualizable relations (e.g., curves, trends, regime changes), this can push attention to structure rather than outcome-matching. For topics where exact magnitudes, thresholds, or formulas are primary, numbers should stay visible and prompts are safer than withholding.

Net: treat temporary number-hiding as a niche, pattern-focused desirable difficulty layered on top of prediction and budgets, not a default replacement for numeric feedback or for well-written comparative prompts.