We focused on regions in early visual cortex (EVC; i.e., V1, V2, V3), the lateral occipital complex (LOC), comprising object-selective regions LO and pFs in the ventral stream, and on the posterior intraparietal sulcus (pIPS), comprising the regions IPS0 and IPS1 in the dorsal stream.
To define EVC, we first transformed the subject-specific t maps from the scrambled > objects contrast from the localizer GLM into MNI space. Based on these transformed t-maps, we computed a contrast comparing the group-level activation against zero, which resulted in one t-map across subjects. We then thresholded this t-map at the p < 0.001 level and calculated the overlap between the thresholded t-map and the combined anatomic definition of V1, V2, and V3 from the Glasser Brain Atlas (Glasser et al., 2016 (link)). Finally, we transformed this overlap image back into the native subject space for each subject, resulting in subject-specific EVC masks. A more fine-grained definition of the ROIs V1, V2, V3, and V4 based on the Wang et al. (2015) (link) atlas led to qualitatively similar results as the EVC definition.
To define object-selective cortex, we manually identified the peaks in the subject-specific t-maps of the objects > scrambled contrast from the localizer GLM, which corresponded anatomically to LO and pFS. We then defined spheres with a radius of 6 voxels around both peaks, including only those voxels in the spheres that had t values corresponding to p < 0.0001. This resulted in one ROI mask for LO and pFS, respectively. Initial exploratory analyses revealed that LO and pFS yielded highly comparable results. Therefore, we merged the two ROI masks into one combined LOC mask. This resulted in one object-selective cortex mask for each subject.
To define pIPS, we first combined the probability masks for IPS0 and IPS1 from the Wang et al. (2015) (link) atlas and then thresholded this combined IPS0-1 mask at a value of 20%. Next, we transformed the combined pIPS mask into the individual subject space. Finally, we computed the overlap between the individual pIPS mask and the subject-specific t-map of the contrast from the localizer GLM comparing all objects and scrambled objects against baseline, thresholded at p < 0.0001. This procedure resulted in one pIPS ROI mask for each subject. In case the EVC, object-selective cortex, or pIPS masks overlapped in a given subject, the overlapping voxels were discarded from all masks.
To define EVC, we first transformed the subject-specific t maps from the scrambled > objects contrast from the localizer GLM into MNI space. Based on these transformed t-maps, we computed a contrast comparing the group-level activation against zero, which resulted in one t-map across subjects. We then thresholded this t-map at the p < 0.001 level and calculated the overlap between the thresholded t-map and the combined anatomic definition of V1, V2, and V3 from the Glasser Brain Atlas (Glasser et al., 2016 (link)). Finally, we transformed this overlap image back into the native subject space for each subject, resulting in subject-specific EVC masks. A more fine-grained definition of the ROIs V1, V2, V3, and V4 based on the Wang et al. (2015) (link) atlas led to qualitatively similar results as the EVC definition.
To define object-selective cortex, we manually identified the peaks in the subject-specific t-maps of the objects > scrambled contrast from the localizer GLM, which corresponded anatomically to LO and pFS. We then defined spheres with a radius of 6 voxels around both peaks, including only those voxels in the spheres that had t values corresponding to p < 0.0001. This resulted in one ROI mask for LO and pFS, respectively. Initial exploratory analyses revealed that LO and pFS yielded highly comparable results. Therefore, we merged the two ROI masks into one combined LOC mask. This resulted in one object-selective cortex mask for each subject.
To define pIPS, we first combined the probability masks for IPS0 and IPS1 from the Wang et al. (2015) (link) atlas and then thresholded this combined IPS0-1 mask at a value of 20%. Next, we transformed the combined pIPS mask into the individual subject space. Finally, we computed the overlap between the individual pIPS mask and the subject-specific t-map of the contrast from the localizer GLM comparing all objects and scrambled objects against baseline, thresholded at p < 0.0001. This procedure resulted in one pIPS ROI mask for each subject. In case the EVC, object-selective cortex, or pIPS masks overlapped in a given subject, the overlapping voxels were discarded from all masks.
Free full text: Click here
