Bridging Continuous and Discrete Models of the Anterior Temporal Lobe via Cortical Gradients

The anterior temporal lobe (ATL) is critical for semantic cognition, yet its organisation remains debated. Some evidence supports a patchwork of discrete, domain-specific parcels, while the graded hub hypothesis proposes smooth, ordered transitions. We examined regions along ATL's curvature, testing whether connectivity profiles varied systematically or discontinuously along whole-brain dimensions of functional connectivity. The structural and functional connectivity of ATL varied largely systematically on the heteromodal-unimodal dimension, with intermediate ventral parcels showing stronger heteromodal connectivity; however, medial parcels deviated from this pattern. On the auditory-motor to visual dimension, there were partial dorsal-to-medial shifts in connectivity, but parcels at both ends did not follow this trend. There were also systematic dorsal-to-medial connectivity shifts between default and control networks. Individual differences in structural connectivity predicted activation in auditory tasks. These results largely support a graded hub model of ATL, while also revealing non-systematic functional variation that may enable flexible semantic processing.