x_true is not uncertain. It’s a value set by the experimenter, so always deterministic, and it always appears as a conditioning variable. Which part is misleading?
ah. i’m confused about the additional application of the gaussian centered around the x_true on each trial.
for example, in the first tutorial, we do not need to do this–> we just generate a likelihood around x_true and modulate by a prior
Hey everybody, since this is already a megathread, I’ve renamed the thread to focus on W2D1 and changed tags accordingly. I think it would benefit students to distill this into a FAQ about W2D1, because this thread is a bit unwieldy. Somebody can start a gdoc and post when it’s ready. Just remember: NMA W2D1 starts in less than 1 hour (timezones!) - it’s too late to make last minute changes to the tutorials.
3 Likes
I agree there is an inconsistency… however I can’t possibly fix all tutorials. I’ll update my notes. A cover story is that here x_true is a very weak signals, so induces noisy signal x_tilde.
Hey, not sure I understand the issue. Likelihood functions and (arguably) priors are generally about hypothesized positions (because the subject can not know the real position)
i hope I understand now, also after your answers on zoom 
to restate: you’re essentially just generating / simulating brain data, or x_tilde, or what the subject has access to, on each trial
ManyThanks! Helps a lot!
Hey all, Konrad is making live edits to T3 right now - I am going to lock and archive this thread temporarily, since it’s likely going to confuse students and it will be stale by the time they see this. ETA 2 hours. We can clean up and move non-stale content into a new thread later.