Dear Ray,
The idea seems to be that with shorter ISIs there will be signal within the faster frequency range, which would be prone to aliasing when going with the usual STC approaches (which is probably trivial / easy to determine for someone with a signal processing or related backgroudn, but rather difficult in case you just use those approaches as a preprocessing step to examine neuroscientific / psychological questions, with no special knowledge on signal processing).
With regard to the new method, as far as I can see the T value gain refers to a certain condition (visual stimulus). In case of rapid event-related designs any such contrasts would not be too meaningful though. I'm wondering about differential contrasts, if model fit were better for condition 1 and 2 when going with a certain STC the differential contrast might be unaffected then.
Anyway. I think it's very important to look at STC more closely and to come up with newer and better approaches where possible, as only few papers have addressed that issue so far. For me, it's more a practical issue right now in the sense of "would one want to use STC as implemented in SPM for rapid event-related designs with ISI < TR".
Best regards & please forgive my ignorance ;-)
Helmut
|