Week 3 — Drift-Diffusion: Reading Minds Off Reaction-Time Distributions

Week 3 of 8. The model that turned reaction times from a nuisance into data.
For decades, RT experiments reported mean response time and called it a day. The drift-diffusion model says that's throwing away almost everything: a decision is noisy evidence accumulating toward one of two boundaries, and the full shape of the RT distribution — including the errors, including the skew — pins down four separable parameters. Drift rate is how good you are at the task. Boundary separation is how cautious you're being. They dissociate. That is why the model matters: it separates ability from strategy, which mean RT can never do.
Agenda:

1. Recap and intros (10 min)
2. The generative process: Wiener diffusion, drift rate, boundary separation, starting point, non-decision time (25 min)
3. Why the RT distribution is right-skewed, and what fast errors vs. slow errors tell you (20 min)
4. The speed-accuracy tradeoff as a boundary shift, not an ability change (20 min)
5. Live demo: simulate a diffusion process and watch the distribution emerge (20 min)
   Reference: Ratcliff & McKoon (2008), "The Diffusion Decision Model."
   This is an online event. The video link appears on this page once you RSVP.