Hank is there any fixation period in between your two tasks? Your EVs are closely correlated which makes your x matrix rank deficient. You may have to model only one of the two (say the visual task) and leave the other one unmodeled. Then the results for that regressor will be interpreted as vision - words and if you add a -1 contrast you can interpret it as words - vision.
Hope this helps
Martin
On Apr 6, 2011 6:05 AM, "Hank Owens" <
[log in to unmask]> wrote:
> Hi. I have a problem with setting up FEAT properly. I've read the FEAT manual, and went through FSL course on the FSL website, but I can't get it to work properly with my data. I have a simple block design study, in which the first block is a simple visual fixation for 28 seconds, and the second block is a visual presentation of various words for 28 seconds, one word each 2 seconds. The session I try to analyze has 6 blocks of each condition (12 blocks altogether). I used the model setup wizard in FEAT to set up the analysis, and the results are fine, but it only gave me the results for 1 EV (word presentation). I would like to compare them with the activations from fixation blocks, so I chose full model setup, and that is where the problems started. I chose 2 EVs, square types for basic shapes. For the first EV (fixation) I set up Off to 28 sec, On to 28 sec, and Phase to 0. For the second EV (words) I set up Off to 28 sec, On to 28 sec, and Phase to 28 sec. Then, when I choose to view design I get the warning that the model is deficient, and at least one EV is close to a linear combination of the others, and I should alter the design. Why? I mean, if I understand correctly, I should set Off and On to the length of the block, and Phase value to shift my second EV so that it starts after the first EV ends. What am I doing wrong? I couldn't really find that On/Off/Phase part described in detail in the manual.
>
> Please help.
>
> Hank Owens
