Hi Alberto,
> we have three tasks (only the cue stimulus changes), each performed by a
> different group (subj number: 15, 15, 12). each task has 6 conditions
> (55 trials each): 2 congruent, 2 incongruent, 2 neutral.
> i'm trying to see why, in task1 and 2, one congruent condition facilitate the
> response and the other not and why this effect disappears completely in task3.
So, it seems that you are interested in a contrast between congruent conditions, that is, you want (Task1.cong1)>(Task1.cong2) and (Task2.cong1)>(Task2.cong2) to identify brain regions that are correlated to facilitation in Task1 and Task2. The same contrast for Task3 will presumably be empty, since facilitation is not present.
The way you detect facilitation is through changes in RT. So differences between congruent conditions (i.e. cong1>cong2) could be explained by greater time on task or by the presence of a neural process that does facilitation. By constructing your congruent regressors as boxcars equal to RT, you should be able to control for time on task. Then, any differences between conditions will be due to the presence of the facilitation, and not due to non-specific processes. This line of reasoning requires the linearity assumptions I mentioned earlier.
However, you also have events that occur very close to each other:
cue(200ms)-blank(100)-target(until subj. resp)-target2(until subj resp)
The ISI is not jittered, the ITI is very small, the ITI has only a small jitter, the trials are fixed relative to the TR, and the response durations are very similar to each other. All this means that it will be very difficult, if not impossible, to disambiguate activation from the different events within a trial. So, creating boxcar regressors for response1 and response2 may not work and you may need to treat the entire trial as a single event.
> intertrial interval varies randomly between 1000-1500ms after second button press.
So, does this mean that you have a new trial occurring every 3 s? If so, this is probably the biggest factor for why you are getting empty contrasts. The signal is probably saturated.
Hope this helps.
jack
