Dear SPM users,
I am designing a fMRI language paradigm using 2 different language tasks, each containing 2 types of stimuli (2 levels of difficulty per task).
The goal is to obtain a single run :
- that is non monotonous, but clear regarding the task to be performed, alternating between the 2 tasks in a 'A/B' block design;
- that jitters the onsets of the 2 types of stimuli (easy vs difficult) within the blocks to allow the estimation of the effect of difficulty;
- that allows to estimate the 4 contrasts separately, in an event-related manner;
- and thus containing all the 1st-level contrasts to feed to the 2nd level analysis.
I used optseq2 to find the ideal sequence of events and null events for the event-related presentation.
For each task, I obtained the schedule to effectively present 40 easy and 40 difficult events. These 2 presentations are therefore scheduled as if they were a continuous and separate run of each task.
To fit my paradigm, I segmented each task sequence by blocks of 8 items (mix of easy and difficult stimuli), preserving the order of the type of stimuli to be presented.
Because the block length is predetermined (30 s), the null events were decreased in each block in the following manner if the sum of the duration of the 8 stimuli and their respective subsequent null events exceeded the block length : shorten the longest null event in the block to the minimum (2s), then the second and so on until the block length = 30 s.
I thereby obtained a sequence which may look like :
A1 A1 A2 A1 A1 A2 A2 A1 B1 B1 B1 B1 B2 B2 B2 B1 ...
A pilot study showed that the signal of interest remained in a time domain that was no filtered out and the contrasts showed plausible activations for each condition A1, A2, B1,B2.
However I know that the timing modifications to fit the block may render the design suboptimal.
My question to you is how valid it is to consider 2 separate event-related schedules and to segment them in blocks A/B?
|