Dear Torben,
Yes, definitely, a slow response and/or a slow sampling rate does not mean we can gain no high-res temporal information. It is going to depend on response variability and no. of trials, scanner & physiological noise, noise introduced during preprocessing, ... whether it can be extracted or not. And shorter TRs would certainly be useful.
We had a little discussion since then in this thread https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=SPM;2e65f180.1508 on microtime settings reflecting the slice parameters vs. e.g. temporal resolution of behavioral data. I guess it's also related to the fact that the exact settings during model set-up tend to be ignored / are often treated as a black box and/or some settings are established in some way although there's actually rather weak support or only just a few studies with partly contradictory findings (like the slice timing you've mentioned, high-pass filter & detrending issues to mention another topic.
Most of the time it doesn't matter, but sometimes the code/settings are suboptimal (or could be at least). To cite myself, https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;f525092b.1508 on orthogonalization of time bin regressors within conditions - when stumbling across this I started wondering whether this could explain some rather weird time courses at the beginning / end of the time window in some papers ;-)
Best
Helmut
|