Fig. 3.

Procedure for making the TDC regressor (green curves) for PPI analysis. (a) At hemodynamic level: TDC regressor is simple componentwise multiplication ($\odot$) of the signal at seed location [TIC regressor, $y_s(t)$] and response model to task [TDR regressor, $X^{\mathrm{TDR}}(t)$]. (b) At neural level: TDC regressor is built by first deconvolving HRF from the signal at seed location [TIC regressor, $y_s(t)$] and multiplying the result [neural activity, $y_s^n(t)$] into task vector (vector of ones and zeros, one for when stimulus/task is on) and reconvolving the result with the same HRF. The floating scale on the right side shows the scale of signals in $x$ and $y$ axes.