on fast and slow errors

I try really hard, but I still struggle when it comes to convolving distributions entirely in my head. Apparently, the presence of slow errors is a common prediction for the drift diffusion model. Here’s a code-adventure showing it is indeed the case.