On 15 Oct 2009, at 20:03, David Soto wrote:
-hello; I am new to FSL and am trying to figure out what is the better way
of including RT in the FEAT when event-related designs are used
My paradigm involves: A) memory cue(1s) -B)delay-200ms- C) search
display-200ms, followed of a response within a time window of one second- D)
memory test for 100 ms followed by a memory response within one second.
I've thought about these options
1)In the model specification, thought of using a Custom 1 entry per volume
for each the search response and also for the memory response, but since my
design is event-related with a jittered inter-trial interval, there is not
always a reaction time per scan volume
so I think the Custom 1 entry may only work well for block designs with a
fixed phase and one reaction time per scan?
Yes - the timing precision with the custom-1 entry is crude so custom-3 gives you more temporal precision, which it sounds like you need here.
2) or using a Custom 3 with the onset of the critical display in the first
column, and the reaction time in the second column; again doing 2 entrys one
for search RT and for memory RT
but not sure this is the right approach here?
Not quite right - see the FEAT manual: "The first number in each triplet is the onset (in seconds) of the period, the second number is the duration (in seconds) of the period, and the third number is the value of the input during that period".
So in your case if you want the height of the regressor to be the reaction time, that should be the third column.
For such cases where you have events that each have a varying 'strength' in some sense, a common approach is to have one EV of constant height to model the mean response, and a second EV which includes the height/strength variations, and that second EV is orthogolised wrt the first.
Cheers.
3) could the reaction time be included as an additional confound EVs, one
for each of the responses..
many thanks and best wishes!
