Hello FSLers,
I am analyzing fMRI data from an n-back memory task, where on each trial the subject is asked to remember what letter happened n trials before. I have 3 conditions:
0 back
1 back
2 back
My question is regarding how best to model these data. In reading up, it seems that I can either:
1. model each condition independently. Then, at the group level, use the linear contrast of [-1 0 1] to model 0 back < 1 back < 2 back;
or
2. model each condition as one, and weight individual condition types differently, depending on the working memory load. So for example, weight the 2-back trials as 1, 2-back trials as .66 and 0-back trials as .33.
Can you please advise?
Many thanks in advance,
Lara
|