2008/10/28 Yael Shani <[log in to unmask]>

Hi Eugene

Thank you for your reply.

I think I wasn't very clear in my second question: What am I supposed to write in the scond column? Should I write the time until the screen changes?

I would say the time until the button is pressed. But it really depends on the task and the expected responses that you are trying to model. Do you expect the stimulus associated activity to continue after the button press? You do also want to model, probably with a separate ev, the activity associated with the pressing of the button.

Eugene

In addition. I just want to make sure I understand what to write in the first column- I need to write the onset time from the beginning of the Run, and not the time between 2 different events. Did I get this right?

Yes, that's right.

Many Thanks

Yael

On Sun, Oct 26, 2008 at 3:30 PM, Eugene Duff <[log in to unmask]> wrote:

Hi Yael,

2008/10/26 Yael Shani <[log in to unmask]>

Dear fsl users

I need some help with an event related design I have.

In my experiment, the subject sees a display and has to choose between 2 options: Up or Down.

The time lag for this choice is up to 15 seconds. When the subject chooses any option the screen changes to a fixation.

My questions are:

1. In the text file (for the custom 3), what should I write in the third column, if the subject's response is imporatant for my analysis?

If you want to compare responses, use a separate regressor for each of the two response types, both with 1 in the third column.

2. In the text file (for the custom 3), in the second column, should I write the whole time lag of the event displayed or only the time until the subject has pressed a buttom?

You could probably use multiple regressors to separately model activity associated the decision period and the motor(?) response. If average response times are short you may need to orthogonalise the motor regressors w.r.t. the decision regressors.

2. I want to see the difference in the activation when the suject chose Up or Down. How do I define a contrast for that?

Set the respective regressors (evs) to +1 and -1 to test for greater activation associated with the first compared to the second. I guess you will be more interested in the decision period than the responses.

Eugene