In may 2016. I tried to repeat a pioneering research published on NATURE (Kay et al., 2008) using a data-driven Voxel-Wise Model (Naselaris et al., 2011). I successfully constructed a liner receptive field model for every voxels using the Gabor wavelet filter according to the encoding theory of the primary visual cortex (Hubel and Wiesel, 1968). Without noise ceiling procedures, I also got remarkable accuracy in the identification task (about 48.8%, meanwhile the chance level is about 0.8%). My code is now open sccess on Github - Voxels-activity-modeling-using-fMRI-data.
Jingjie Li Jun 14, 2016