I'm bringing this thread back from the dead because I have the exact same question as Kim, but I would like to request a clarification.

Background: I performed a group analysis on the beta weights output by 3dD -stim_times, and based on one of the contrasts from that group analysis I created an ROI mask (using 3dclust -1noneg... so that the mask would only include the ROIs from the positive condition in the contrast). Now, I want to get average timecourses over each of those ROIs, perhaps using the -iresp files output from 3dD, or whatever else is reasonable (as Rick suggested above).

However, I am a bit unclear on what exactly the -iresp files are. I thought that each iresp file contains the voxelwise impulse response (i.e., a beta weight based on the amplitude of the scaled data and the model function) from the time series for a given subject for each stimulus type at each lag in the model I specified in 3dD. However, in my model, I specified a six parameter tent function (TENT(0,15,6)), but the iresp files each contain seven sub-bricks. My TRs are 2.5s. So I'm led to believe that the iresp files contain one sub-brick for each TR in the 15s window I specified. Is that right?

If so, where do the specific values in each (voxel of each) sub-brick of the iresp files come from? Are they still voxelwise impulse responses (in the sense that I described above)? If so, why are there seven rather than six sub-bricks in each iresp file?