Week 2 — ACT-R Activation: Why Forgetting Follows a Power Law

Week 2 of 8. Our first real model.
ACT-R's declarative memory assigns every chunk a base-level activation that decays as a power function of time since each rehearsal. From that one equation you get the forgetting curve, the spacing effect, the fan effect, and a prediction for retrieval latency — all falling out of the same three parameters. It is one of the most quietly impressive results in cognitive modeling, and almost nobody outside the field has seen it.
Agenda:

1. Recap and intros (10 min)
2. The base-level learning equation, term by term. Why a power law and not an exponential? (25 min)
3. Deriving the spacing effect and the fan effect from it (25 min)
4. Retrieval latency, the threshold parameter, and what "I know I know it" looks like in the math (20 min)
5. Open floor: where does this fail? (15 min)
   Bring: nothing required. A calculator or a Python REPL if you like following along numerically.
   Reference: Anderson & Schooler (1991), "Reflections of the Environment in Memory."
   This is an online event. The video link appears on this page once you RSVP.