To get the delay \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau $$\end{document}τ (the delay between the peak of fMRI and the peak of neural signal), we searched all integers between [0 8] based on a previous study that reported 4–8 s latencies in the gray matter [87]. Then the optimal integer was chosen as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau $$\end{document}τ for which the covariance of noise \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$cov\left( {y\left( t \right) - \hat{x}\left( t \right) \otimes h\left( t \right)} \right)$$\end{document}covyt-x^t⊗ht was smallest, to obtain the set of onsets. Subsequently, the vertex-by-vertex HRF was fitted and extracted with these pseudo-events. The readers are referred to Tagliazucchi et al. [86] and Wu et al. [58] for further details on the deconvolution method.
