> We have two separate scans of the same group of subjects. The
> subjects are engaged in different tasks during the two scans, and
> would like to compare them. Originally, we concatenated the two
> scans, and then ran a deconvolution analysis with follow-up t-tests
> on the concatenated dataset.
So it is an event-related design? How did you model the HRF with deconvolution analysis? And how did you run the t test of the contrast between the tasks?
> However, the results did not make any sense, leading us to
> suspect that our methods were off.
In what sense the results didn't make sense? Across all subjects? Can you provide a copy of the 3dDeconvolve script?
Gang