Anatomical and functional connectivity support the existence of a salience network node within the caudal ventrolateral prefrontal cortex

Three large-scale brain networks are considered essential to cognitive flexibility: the ventral and dorsal attention (VAN and DAN) and salience (SN) networks. The ventrolateral prefrontal cortex (vlPFC) is a known component of the VAN and DAN, but there is an important gap in the current knowledge regarding its involvement in the SN. In this study, we used a translational and multimodal approach to fulfill this gap and demonstrate the existence of a SN node within the vlPFC. First, we used tract-tracing methods in non-human primates (NHP) to quantify the anatomic connectivity strength between the different vlPFC areas and the frontal and insular cortices. The strongest connections with the dorsal anterior cingulate cortex (dACC) and anterior insula (AI) locations comprising the two main cortical SN nodes were derived from the caudal area 47/12. This location also has strong axonal projections to subcortical structures of the salience network, including the dorsomedial thalamus, hypothalamus, sublenticular extended amygdala, and periaqueductal gray. Second, we used a seed-based functional connectivity analysis in NHP resting-state functional MRI (rsfMRI) data to validate the caudal area 47/12 as an SN node. Third, we used the same approach in human rsfMRI data to identify a homologous structure in caudal area 47/12, also showing strong connections with the SN cortical nodes, thus confirming the caudal area 47/12 as the SN node in the vlPFC. Taken together, the vlPFC contains nodes for all three cognitive networks, the VAN, DAN, and SN. Thus, the vlPFC is in a position to switch between these three cognitive networks, pointing to a key role as an attentional hub. Its tight additional connections to the orbitofrontal, dorsolateral, and ventral premotor cortices, places the vlPFC at the center for switching behaviors based on environmental stimuli, computing value and cognitive control.