Hi Patrick, On 30 January 2010 01:44, Patrick Thompson <[log in to unmask]> wrote: > > > Hi, > > I have an experiment in which events of interest are of different > durations 1 s to 15 s. I want to look for brain regions that show increase > or decrease in activity at the onset, during, and end of these events. Since > events are of different duration and numbers, I am wondering what is the > best way to model onset transient, sustained, and end transient activity for > this experiment. My approach is to have three different regressors: > > The first regressor will be an impulse regressor which models transient > activity at the onset of all events. > > The second regressor will be a variable epoch regressor which models > sustained activity at longer events (> 2.5 s) . > > The third regressor will be an impulse regressor which models transient > activity at the end of longer events (>2.5 s). > > Three contrast, [1 0 0], [0 1 0], and [0 0 1] will give me brain regions > active at onset, during, and end of these events. > > Is this a valid approach? > > Yes, in general that's sensible, and has been done before. However, your events need to be long enough for the different components to be distinguished. You mention your events vary from 1 to 15s. Do you mean you are trying to model events of a range of durations with the three regressors? If so, only the longer duration events will drive your results. Eugene > Regards, > Pat > > > >