Dear Dr. Alfonso,

A quick but naive question following the above: after getting the data_reduced and w_reduced time variables for each subject, if we wish to calculate the weighted correlations, then we simply multiply data_reduced with w_reduced (weights) over all ROIs. Then we calculate pairwise correlation between these weighted time series, followed by taking the average (to get the second level group measure). Computing one sample or two sample t test at this pint will give us the group statistics and I can call the conn_fdr function to perform the FDR correction.

Summarizing,
1. For all ROIs, get the w_reduced and data_reduced
2. weighted_ts = data_reduced*w_reduced;
3. Calculate correlation between all the weighted_ts (followed by Fisher's transform)
4. Take average of Z values across all subjects to get the second level ROI-to-ROI beta values

Is the above correct or have I misunderstood the hrf-weighting of the time series to calculate the weighted correlations?


Thanks and Regards
Pravesh