Dear all,
I am sorry that this is actually a re-post but I really need some advice and comments about my design.
I came across some questions about the timing issues of my event-related design and I have been reading Dr Rik Henson's writing in Human Brain Function 2nd edition, Ch 10 Analysis of fMRI timeseries, in seek of the optimal timing schemes in order to get best results, but I still have not acquired a clear figure.
The timing details of my design are as follows:
There are altogether 3 event types A B and C of which I include 40 trials respectively. The procedure for a single trial is 0.5s of fixation, 1.5s of presentation of a stimulus of no interest (say a picture of a player whose sole purpose is to enhance the contextual authenticity), and 4s of presentation of target stimulus during which the subjects make their decisions. The inter-trial-interval(ITI) ranges from 2.75s to 3.25s with a centered mean of 3s and the TR is 2.5s.
However, I am confused by these following problems after comparing my design to what was suggested in Dr Henson's writing:
1) According to Dr Henson, short SOAs (of a few seconds) are more desirable in terms of statistical efficiency [pg.12 Section IIIB]. However, Dr Henson also indicated that because of nonlinearilty short SOAs will result in saturation, i.e. the response to a run of events would be smaller than would be summation of responses to individual events.
So I wonder whether my SOAs are too long to be statistically efficient because I suppose in my case, the SOAs ranges from 8.75s to 9.25s ( 4s after the onset of target stimulus + ITIs of 2.75-3.25s + 2s before the onset of target stimulus of the next trial).
Howover, reducing the length of 4-s period seems unresaonable because in practice subjects require enough time to think and respond. In addition, I assume adequate time interval must be inserted between two subsequent trials to allow for the BOLD's signal's return to resting baseline. So how should I solve this problem?
2) Dr Henson also mentioned the relationship between the ratio of minimal SOA to TR, i.e. peristimulus sampling rate. So what is the optimal range of the effective PST sampling rate? In my case, I supppose my SOAmin/TR is 3.5Hz. Is this rate acceptable?
Any comment or advice will be greatly appreciated. And again, I apologize for any inconvenience and redundancy I might have caused by reposting my problems.
Many thanks and best regards
Ce
