For the preparation of the synthesis, the reported peak voxel coordinates and the sample sizes of the individual studies were manually extracted into a text file and served as input data for performing an ALE meta-analysis. In one case, where the peak voxel coordinates were not published and we requested them via personal correspondence, we received a NIfTI output file created by FMRIB Software Library (FSL) [40] Randomise tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Randomise). For extraction of the most meaningful peak voxels, we used FSL v6.0 (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSL) on Windows 10 and the Cluster tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Cluster), setting the threshold for the cluster level to p < 0.01. In three other studies, where the coordinates were specified in Talairach space, we used the Lancaster icbm2tal transform implemented in GingerALE v3.0.2 to transform to MNI space (http://www.brainmap.org/ale/) [41, 42]. The same version of GingerALE was used to perform the meta-analysis itself [37, 38, 43]. Prior to the actual meta-analytical calculations, four mask outliers were identified in the input data and were subjected to a plausibility check. In one case, for a coordinate reported from Yeh et al.
