Averaging of single runs (task-based design)

Dear all,

I have a theoretical question concerning the averaging of multiple runs in a task-based design. Imagine that I would like to calculate variables such as functional connectivity, fast fourier transform (and then the power-law exponent), etc. after the preprocessing with AFNI_proc is done.


  1. Should I first use 3dmean to average all runs (for each subject). E.g., there are 6 runs for each subject and I would create one “final” average run for each subject via 3dmean; or
  2. Should I calculate things like FFT, PLE, FC, etc. first for each single subject to subsequently calculate their mean via 1deval, 3dcalc, etc.; or
  3. Should I use the all_runs file that is created by AFNI_proc automatically and merges all runs.

Is there a recommendation that most people in fMRI (or in AFNI) use these days?


Philipp, I don’t see any strong rationale for the averaging process. So I would consider dealing with all runs together (your option 3).

Dear Gang,

thanks for the response and information.