Hi,
In general it is best to model everything that you think might have a different brain activity associated with it.
Your contrasts can exclude things that you are not interested in, but you will still benefit from including other things by having small residuals and generally stronger stats (if there is an actual difference).
All the best,
Mark
On 4 Sep 2013, at 05:49, Mark <[log in to unmask]> wrote:
> Hi,
>
> My fmri experiment consisted of 3 types of trials. Within each trials, I have 6 different periods (regressors).
> However, I'm just interested in 4 regressors, but not the rest 2 (although I'm not sure but I don't expect differences between trial types for the rest 2 periods).
>
> In my 1st-level model, can I set up a model consisting of 14 EVs (4x3 + 2)? Or should I set up a 18-EV model (6x3)?
>
> Thanks in advance.
|