“…Finally, the choice of a hard‐segmentation approach could represent an inherent limitation since it attributes voxels to the clusters exhibiting higher connectivity density (Behrens et al, 2003). Although widely used in literature (Cacciola, Milardi, Basile, et al, 2019; Cacciola, Milardi, Basile, et al, 2019; Middlebrooks, Tuna, Almeida, et al, 2018; Middlebrooks, Tuna, Grewal, et al, 2018), the use of such approach may introduce a bias selecting only the most connected voxels as part of each connectivity map therefore imposing a stricter parcellation in respect to the anatomical reality. This is particularly true when brain structures are thought to be composed of partially overlapping functional territories rather than well‐delineated subdivisions (Alkemade, Schnitzler, & Forstmann, 2015; Lambert et al, 2015; Plantinga et al, 2018).…”