Education & Learning / Memory Systems
Predict when knowledge will decay and schedule timely reviews — decay prediction for proactive retention.
Why This Prompt Exists
Memory decays exponentially after learning. Without tracking, you don’t know when you’re about to forget. Most learners review too late — after memory has already decayed.
You get:
- reviewing after forgetting (have to relearn, not just reinforce)
- no prediction of when you’ll forget (reactive, not proactive)
- wasted reviews (too soon or too late)
- no visibility into retention strength
- inability to prioritize which items need review
But forgetting curves are predictable:
- Ebbinghaus curve: exponential decay, steepest initially
- retention after 1 day: ~50-80% (varies by material)
- retention after 2 days: ~40-70%
- retention after 6 days: ~30-60%
- retention after 30 days: ~20-40%
- each review flattens the curve
Without tracking, you don’t know what you’re about to forget.
This prompt predicts forgetting and schedules proactive reviews.
The Prompt
Assume the role of a memory researcher who tracks and predicts forgetting curves. Your task is to predict when knowledge will decay and schedule reviews before forgetting occurs. Generate: 1. LEARNING EVENT DATA - Material: [what was learned] - Initial learning date: [date] - Initial recall success: [100% / high / medium / low] - Material difficulty: [easy / medium / hard] - Reviews completed: [list of review dates and success rates] 2. FORGETTING CURVE PREDICTION | Days Since Learning | Predicted Retention | Review Needed? | Confidence | |--------------------|---------------------|----------------|------------| | Day 0 (initial) | 100% | No | High | | Day 1 | [75-85%] | Yes (if retention <90%) | High | | Day 3 | [60-75%] | Yes | Medium | | Day 7 | [50-65%] | Yes | Medium | | Day 14 | [40-55%] | Yes | Low | | Day 30 | [30-45%] | Yes | Low | 3. RETENTION THRESHOLDS | Review Type | Trigger Retention | Action | |-------------|-------------------|--------| | Immediate review | <90% | Same day | | Short-term review | <80% | Within 24 hours | | Medium-term review | <70% | Within 3 days | | Long-term review | <60% | Within 7 days | | Mastery check | <50% | Relearn, then resume spacing | 4. REVIEW SCHEDULE (proactive, before forgetting) | Review # | Optimal Timing | Your Schedule | Predicted Retention Before Review | |----------|----------------|---------------|----------------------------------| | 1 | 1 day after learning | [date] | ~50-80% | | 2 | 3 days after learning | [date] | ~60-75% | | 3 | 7 days after learning | [date] | ~50-65% | | 4 | 14 days after learning | [date] | ~40-55% | | 5 | 30 days after learning | [date] | ~30-45% | | 6 | 60 days after learning | [date] | ~25-35% | 5. RETENTION STRENGTH TRACKER | Item | Initial Strength | Current Strength (after reviews) | Decay Rate | Next Review Due | |------|------------------|----------------------------------|------------|-----------------| | [item 1] | [100%] | [X%] | [fast/med/slow] | [date] | | [item 2] | [100%] | [X%] | [fast/med/slow] | [date] | 6. ADAPTIVE INTERVALS BASED ON REVIEW PERFORMANCE | Review Score | Interval Adjustment | Next Interval | |--------------|--------------------|---------------| | 100% (perfect) | Multiply by 2 | 2x previous | | 80-99% | Multiply by 1.5 | 1.5x previous | | 60-79% | No change | Same interval | | 40-59% | Multiply by 0.75 | Shorter interval | | <40% | Reset to day 1 | Restart spacing | 7. FORGETTING CURVE PARAMETERS BY MATERIAL TYPE | Material Type | Initial Decay Rate | Stabilization Point | Review Sensitivity | |---------------|-------------------|---------------------|-------------------| | Facts (random) | Fastest | Lowest | High | | Vocabulary | Fast | Low | High | | Concepts (understood) | Medium | Medium | Medium | | Procedures (practiced) | Slow | High | Low | | Skills (physical) | Slowest | Highest | Lowest | 8. COMMON FORGETTING CURVE MISTAKES | Mistake | Why It Fails | Correct Approach | |---------|--------------|------------------| | Reviewing too late | Memory already decayed | Review at 80-90% retention | | Reviewing too soon | Wasted effort | Wait until just before forgetting | | Fixed intervals for all | Ignores individual differences | Adapt based on performance | | No retention measurement | Can't calibrate | Test before scheduling | | Ignoring material type | Wrong decay model | Match parameters to content | INPUTS: Material learned: [PASTE DESCRIPTION] Initial learning date: [PASTE DATE] Initial recall success: [HIGH (90-100%) / MEDIUM (70-90%) / LOW (<70%)] Material difficulty: [EASY / MEDIUM / HARD] Review history (if any): [PASTE REVIEW DATES AND SCORES] RULES: - First review within 24 hours of learning (critical window) - Review when predicted retention is 80-90% (optimal difficulty) - Double interval after perfect recall (100% score) - Shorten interval after poor recall (<60% score) - Track retention strength per item (not just overall) - Material type affects decay rate (facts decay faster than skills) - Test before scheduling next review (measure retention, don't assume)
How To Use It
- First review within 24 hours of learning — the critical window for consolidation.
- Review when predicted retention is 80-90% — optimal difficulty for strengthening memory.
- Double interval after perfect recall — 100% score means the memory is strong; space it out.
- Shorten interval after poor recall — less than 60% means review too late or material too hard.
- Track retention strength per item — not just overall; different items decay at different rates.
- Material type affects decay rate — facts decay faster than understood concepts; skills decay slowest.
- Test before scheduling the next review — measure retention, don't assume it.
Example Input
Material learned: "50 Spanish vocabulary words (basic nouns)"
Initial learning date: "June 1, 2026"
Initial recall success: "MEDIUM (80% on first test)"
Material difficulty: "MEDIUM"
Review history: "Reviewed June 2 (90% recall), June 5 (85% recall)"
Why It Works
Without tracking, learners review too late — after memory has already decayed, requiring relearning, not just reinforcement.
This framework improves outcomes by forcing:
- forgetting curve prediction (predicting when memory will decay)
- retention threshold setting (triggering review at optimal times)
- proactive scheduling (reviewing before forgetting, not after)
- adaptive interval adjustment (tailoring spacing to performance)
- material-type parameter matching (facts decay faster than skills)
Failure modes this prevents:
- reviewing after forgetting (relearning wasted time)
- no prediction of decay (reactive, not proactive)
- wasted reviews (too soon or too late)
- no visibility into retention strength
- inability to prioritize which items need review
This improves on: Fixed-interval review. Adaptive, predicted scheduling reviews at the optimal moment — just before forgetting.
Related to: MS-01 (Spaced Repetition) for intervals; MS-03 (Retrieval Practice) for recall measurement.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.