When using the attempt–worked‑example cycle in adult online training with spaced sessions, does dynamically adjusting task difficulty between sessions based on each learner’s recent success on unguided attempts (to keep them in a productive struggle band) yield larger gains in long‑term retention and transfer than dynamically adjusting only hint‑gating while holding difficulty fixed, and under what conditions does combining both adaptations produce diminishing or negative returns due to excessive cognitive load?

Answer

Dynamically adjusting task difficulty between spaced sessions, based on each learner’s recent success on unguided attempts, is likely to yield somewhat larger gains in long‑term retention and transfer than adjusting only hint‑gating while holding difficulty fixed—provided that (a) difficulty adaptation is smooth and keeps most items in a productive‑struggle band, and (b) spacing between sessions is long enough for some forgetting.

Dynamic hint‑gating (with fixed difficulty) mainly improves calibration and prevents over‑reliance on support; dynamic difficulty mainly improves the quality and distribution of retrieval and problem solving. For long‑term retention and transfer, that latter effect is typically more powerful.

Combining both adaptations (difficulty + hint‑gating) can be beneficial when:

• the interface keeps the adaptation simple and predictable to the learner, and
• the underlying item pool spans a reasonable range of difficulties without huge jumps.

However, combining both starts to show diminishing or even negative returns—largely via excessive cognitive load and confusion—when:

• difficulty is adjusted aggressively between sessions (big jumps up after modest success, or big drops after one bad session), and
• hint‑gating is also tightened or loosened sharply based on recent behavior, and
• learners have low prior knowledge or low metacognitive skill, so they must simultaneously cope with harder problems and reduced access to hints without understanding why.

Under those conditions, the joint system can:

• push learners outside the productive‑struggle band (too difficult, too little support),
• increase extraneous cognitive load as they try to understand the rules of the system rather than the content, and
• reduce engagement, which undermines both retention and transfer.

Practical synthesis:

1. For typical adult online training with spaced attempt–worked‑example cycles and a mixed‑ability audience:

   • Use primary difficulty adaptation across sessions to keep most learners in productive struggle.
   • Add mild, transparent hint‑gating adaptation (e.g., nudge away from constant hint use, but never fully lock out hints for long) rather than aggressive gating.

2. Avoid combining strong difficulty jumps with strong hint restrictions for:

   • low‑prior‑knowledge learners,
   • conceptually dense materials,
   • or platforms that provide little explanatory feedback about why conditions changed.

3. Expect diminishing returns from stacking both adaptations when:

   • baseline items are already well‑targeted to productive struggle, or
   • learners self‑regulate hint use reasonably well; in such cases, added dynamic hint‑gating adds complexity more than it adds learning.

So, if you must choose one lever for spaced attempt–worked‑example training, prioritize dynamic difficulty; use dynamic hint‑gating as a light secondary control and avoid complex, opaque combinations that substantially increase cognitive load, especially for less prepared learners.