Dear Alfonso,

Thanks for your clarification. Although I understand why Conn toolbox implement weighted correlation, i.e., the hrf convolved blocks for calculation ROI-ROI correlation, I am still a bit confused on if this step is necessary. As by default, the Conn toolbox, in the Preprocessing step, has regressed out task effects convolved with the hrf as well as other covariates like with matter and CSF, is it still necessary to convolve the hrf or hanning window function with the stimulus blocks to capture the effect of hrf on the pseudo "rest" time courses? Or anywhere I was misunderstanding? 

Thanks.

Best Regards,

Fang