How It Works
The science of remembering.
In 1885, Hermann Ebbinghaus proved that humans forget 80% of what they learn within 30 days. 140 years later, we built the infrastructure to fix it.
Based on Ebbinghaus (1885). Each dot represents a Cognition-timed intervention that resets the decay curve.
The Forgetting Curve
In 1885, German psychologist Hermann Ebbinghaus discovered that memory retention decays exponentially over time. Without reinforcement, humans forget approximately 80% of learned material within 30 days.
Weibull Decay Model
We model each user-concept pair with a Weibull distribution — capturing not just when they'll forget, but how their individual memory strength, difficulty, and learning pattern affect the decay rate.
Timed Interventions
By predicting the optimal review moment — just before recall drops below threshold — we turn the forgetting curve into a retention curve. Each intervention strengthens the memory trace, extending the next interval.
Under the Hood
From raw data to memory prediction.
We don't just track completions. We build a digital twin of each employee's memory, run decay simulations using patented algorithms, and predict exactly when they'll forget what they learned.
Ingest learning data
We connect to your LMS, CRM, or any learning platform and ingest every signal — completions, quiz scores, time-on-task, interaction patterns, and assessment results. This raw behavioral data becomes the foundation for each employee's memory profile.
- Quiz scores, pass/fail, time-on-task
- Interaction timestamps and session patterns
- SCORM data, xAPI statements, custom events
- Real-time sync via webhooks or batch import
Build a digital brain twin
For every employee, we construct a computational model of their memory — a digital twin of how their brain encodes, stores, and retrieves each concept. This isn't a population average. It's calibrated to their individual learning speed, difficulty sensitivity, and forgetting patterns.
- Per-concept memory stability (half-life)
- Individual difficulty calibration per user
- Lapse history for confusion/interference detection
- Uncertainty estimation for prediction confidence
Run decay simulations
Using our patented Weibull decay algorithm, we simulate memory degradation forward in time for every user-concept pair. The model predicts not just if they'll forget — but exactly when, with what confidence, and how much it will cost if you don't intervene.
- Weibull distribution modeling per concept
- Forward simulation: 7d, 14d, 30d, 90d projections
- Skip-cost analysis: what happens if you don't review
- Optimal review window calculation
Predict + intervene at the right moment
The system surfaces exactly who will forget what, and when — then triggers the optimal intervention. Not too early (wasted effort), not too late (already forgotten). Each review strengthens the memory trace, extending the next interval and building toward long-term mastery.
- Real-time alerts before threshold breach
- Auto-scheduled reviews at optimal timing
- Technique recommendations (active recall, spaced repetition, interleaving)
- Each intervention feeds back into the model, improving accuracy
See it in action.
Book a demo and we'll show you your team's forgetting curves in real time.
Book a Demo