Due to the rather slow change over time of the HRF there's probably no real benefit of splitting one TR into more than the standard 16 time bins. It might be different with long TRs though.
In case you have a short TR of ~ 2 s, increasing the microtime resolution according to the number of scans, e.g. 30, probably leads to very similar results compared to the default microtime resolution of 16.
Recently I reanalyzed some data set with an event-related design and very brief stimuli. Relevant scanning parameters were TR = 3 s, 27 slices. Setting microtime resolution to 27 lead to almost identical results in terms of p-/T-values when compared to the standard 16.
See also an older thread:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;2ecc5b1e.0904
|