Supplementary Figure 4: Comparison of cortical depth with cytoarchitectonic cortical layers.

(A) Grayscale Nissl-stained images of the dlPFC from the Ding Atlas (ref. ⁶⁶ in Methods) were converted into single-slice NIFTI format. Borderlines of the GM-CSF surface and GM-WM surface are manually drawn in the same space. Layers are extracted with LAYNII as described in Methods subsection “Layering for sagittal protocol”. The position of the internal pyramidal layer is visible as a small bump (blue arrow). Just superior to it is the outer band of Bailager (black arrow), which is taken as a landmark to identify the approximate transition area between layer III (green arrow) and layer V (red arrow). This layer has an elevated myelin content, which comes along with reduced T₁ values in MRI contrast (ref. ⁶⁷ in Methods); as such, this landmark is also visible in in-vivo T₁ and T₂* weighted profiles (see black arrows in panel B-D). The reason to show the histological slice in panel A) is to confirm that the transition area between later III and V is approximately 40-60% of the cortical thickness away from WM as visible in myelin-sensitive MRI data. This confirmation is helpful because the relative depth of cytoarchitectonic cortical layers may differ across brain areas. Based on this depth calibration, we believe it is appropriate to pool fMRI signal (E-F) and time courses from two independent subsets of voxels, superficial voxels (superficial ≈ 30-40% of the cortex) and deeper voxels (deeper ≈ 30-40%) as shown in Supplementary Fig. 3. This dip between cytoarchitectonic layers III and IV is used here as the border between so-called “superficial layers” and “deeper layers” which show different responses to the different functional task contrasts (E-F). (Data in panels A-D come from single subjects: data in panel A are from the Ding Atlas, which is based on a single adult female brain, and data in panels B-F come from the same subject in the present study. Error bars in panel A-D refer to the standard error across columns within the area of interest [without layer smoothing]. In panel A this represents the variability across a distance of 2 mm sampled at 7270-8949 points. In panel B, this represents the variability of across a distance of 17 mm sampled at 29-45 points.) The data in panel E-F) refer to the manipulation (alphabetize) versus. remember runs. 10 trial of each condition where acquired per run. The data shown here are extracted after averaging across 4 runs with (20 trials each). This totals to 80 trials. The green contrast refers to the time average across the delay period, whereas the red contrast refers to the time average across the response period (Fig. 1b). As in Fig. 4, the green contrast refers to the signal difference of alphabetizing minus remember trials ((S_alpha-S_remember)/S_rest) and the red refers to the signal difference of the response period minus the rest period ((S_response-S_rest)/S_rest) taken from the inter stimulus interval. For more subjects, see Fig. 4. The uncertainty area in panel E-F refers to the standard error of the mean across the ten respective trials (upon averaging across four runs).