Hi Dana,

Have you considered an F-test [1 0 0 0 ...; 0 1 0 0 ...]? Maybe there is a
reasonable response with negative, but not with neutral (or vice versa),
and averaging both may not reveal anything then.

That said, regarding the last scans, it isn't a big problem. I think you're
talking about a few volumes only. It shouldn't be an issue to have runs
with different lengths.

All the best,


On 29 October 2015 at 23:22, Dana Wagshal <[log in to unmask]> wrote:

> Hi FSL group,
> I have a declarative memory Encoding task where I ask subjects if the
> picture they see is negative or neutral. Right after each picture I have a
> jittered fixation cross. I have set up my first-level GLM with EV1 =
> Negative pictures, EV2 = Neutral pictures, EV3-8 = motion regressors, and
> EV9 = missed trials. I wanted to know what brain regions were involved in
> the task, (task-everything else), and I used a contrast of (1 1). But in
> looking at the results, I expected more activation.
> I looking back at my task, I realized that all the subjects finished the
> task early and had to wait until the scan was done. A lot of time, subjects
> were done with the task for over 1 minute. So there are quite a few TRs at
> the end of the scan that have nothing to do with my task. I assume this is
> the reason why I didn't get as much activation as expected, due to mind
> wandering etc. Sorry for the basic question, but is there a standard or
> acceptable method to deal with these TRs? What is typically done in
> event-related designs?
> It seems like it would be ok to remove them, and if so, is there a
> systematic way of removing the TRs? For example, would I remove the TRs
> after 8 seconds or so had gone by from the last trial? In this way each
> subject would have a different task length but there wouldn't be a lot of
> time when the subject is mind wandering. Or would I find the slowest person
> and remove the same number of TRs from all subjects? With this method at
> least the task length would be the same, but then you would still have
> subjects that have TRs when they're just mind wandering.
> Your help and insight would be greatly appreciated!!!
> Best,
> Dana