As part of my fMRI analysis, I performed a regression using the beta weights from individual trials to predict behavioral accuracy. I found that higher activity in one of my ROIs predicts errors. As a visual for this result, I would like to plot an average time series of the error versus the correct trials in this region.

However, since task accuracy was quite high, my original model did not separate correct and error trials. I used the -stim_times_IM option to get the individual beta weights and matched these up with the trial and accuracy for the regression. I was wondering if (rather than creating a new model that puts correct and error trials in as separate regressors) I could use the original model to get individual IRFs for each trial and then average the correct/error trial IRFs together to get mean IRFs for each.

Would this approach work to look at correct versus error trials? And if so, is there a way I can output the IRFs for each trial during my regression? And if not, what is the best approach I could use to get separate correct and error time series with AFNI?

~Katelyn Wills