Surface files for cifti-smoothing in task analysis scripts

[droediger]

In the TaskfMRIAnalysis scripts, if one asks for additional smoothing, it runs wb_command -cifti-smoothing using the midthickness surfaces from MNINonLinear/fsaverage_LR32k (see here).

Elsewhere, for example on the mailing list here, it is recommeneded to smooth using the midthickness surfaces from T1w/fsaverage_LR32k.

Is it a mistake that the task scripts use the midthickness files from MNINonLinear/fsaverage_LR32k instead of T1w/fsaverage_LR32k?

If it's not a mistake, would you mind helping me understand under which conditions it is preferred to use the surfs from MNINonLinear/fsaverage_LR32k rather than T1w/fsaverage_LR32k with -cifti-smoothing? I'd assumed you should always use the T1w versions since as the distances between vertices would better reflect the actual anatomy.

Thanks!