Dear AFNI experts,
We ran an AFNI proc.py for task-based analysis and excluded incorrect trials from stim files. As far as we understand, 3dDeconvolve does not model the incorrect trials because we did not include in the stim files. However, the TRs for the incorrect trials still included in the analysis. Could you please clarify if we are understanding correctly?
By not including them in the model, such times are consider to be baseline, with no subsequent BOLD response. If you DO expect a response, but not one of interest say, then it might be good to put such events into a different stim class. In that case, perhaps some subjects would have no misses, and then there would be an empty stim times file that 3dDeconvolve would complain about. That would lead to using options like -GOFORIT.
But it would be very reasonable to do.
Thanks for the response. This is really helpful. Could you please clarify one more thing? You mentioned that ‘by not including them in the model, such times are consider to be baseline’. Does this mean they were completely removed and censored from the model? If so, could you also please let me know how can I check these in the afni proc.py output folder?
If no stim times are given, it is the same (with respect to the model) as having fixation (or whatever your “baseline” condition is) during that period. That period of time is in the data and is part of the model, but you have not told 3dDeconvolve about any events to model above baseline. This time is not censored out. There will be nothing special in the results directory. The regression model has all time points, except for those censored due to (presumably) motion.
But BOLD responses to “incorrect” trials are still in the data, and as such, people generally model them, but as different conditions than they would use for the “correct” trials.
Consider reviewing the (fairly old, but still useful) tutorial on a simple case of single subject analysis under AFNI_data6/FT_analysis/tutorial.
In particular, focus and t16_X_matrix.txt, though it might be good to review more than that. Please feel free to post questions about it.
Thank you for the explanations and suggestion. This really helps in understanding the correct and incorrect trials. I’ll check with the posting you suggested.
Can I have one more question that is seemingly a little bit off from this but still related issue. I am checking with number of TRs for each stim files and @ss_reivew_basic says num TRs per stim (orig) : 190 199 202. We had three types of stim files and does this mean each number indicate the number of TRs for each stim file?