Hi, I wanted to ask what will be the right use of 3dDeconvolve in a block design experiment where the blocks’ durations are not fixed (for example, I have one block 58 sec long and the other (same stimuli type) 46 sec long).
Thank you,
Michal
Hi Michal,
Those are pretty long blocks, for which you might have to worry about being too closely modeled with slow baseline fluctuations.
Ignoring that, when event durations vary, we generally suggest the “duration modulation” approach, which uses a BLOCK-shaped basis function.
For a quick overview, the stimulus timing files should not just have onsets, but should have corresponding durations attached, e.g.,
12.3:58 89:46
If the durations were shorter, we would suggest you peruse statistics/deconvolve_block.html. But since those events seem generally long (15s or more), you can probably just go with dmBLOCK, without any parameters.
For example, see “afni_proc.py -show_example NARPS”.
If there are no amplitude modulators (only duration modulators), use AM1.
Please feel free to post part of or all of a timing file to discuss the details.
- rick