This review scrutinizes several findings on human hierarchical processing within the prefrontal cortex (PFC) in diverse cognitive domains. Converging evidence from previous studies has shown that the PFC, specifically Brodmann area (BA) 44, may function as the essential region for hierarchical processing across the domains. In language fMRI studies, BA 44 was significantly activated for the hierarchical processing of center-embedded sentences and this pattern of activations was also observed in artificial grammar. The same pattern was observed in the visuo-spatial domain where BA44 was actively involved in the processing of hierarchy for the visual symbol. Musical syntax, which is the rule-based arrangement of musical sets, has also been construed as hierarchical processing as in the language domain such that the activation in BA44 was observed in a chord sequence paradigm. P600 ERP was also engendered during the processing of musical hierarchy. Along with a longstanding idea that a human’s number faculty is developed as a “by-product of language faculty”, BA44 was closely involved in hierarchical processing in mental arithmetic. This review extended its discussion of hierarchical “processing” to hierarchical “behavior”, that is, human action which has been referred to as being hierarchically composed. Several lesion and TMS studies supported the involvement of BA44 for hierarchical processing in the action domain. Lastly, the hierarchical organization of cognitive controls was discussed within the PFC, forming a cascade of top-down hierarchical processes operating along a posterior-to-anterior axis of the lateral PFC including BA44 within the network. It is proposed that PFC is actively involved in different forms of hierarchical processing and specifically BA44 may play an integral role in the process. Taking levels of proficiency and subcortical areas into consideration may provide further insight into the functional role of BA44 for hierarchical processing.

Despite myriads of studies on a parallel organization of cortico-striatal-thalamo-cortical loops, direct evidence of this has been lacking for the healthy human brain. Here, we scrutinize the functional specificity of the cortico-subcortical loops depending on varying levels of cognitive hierarchy as well as their structural connectivity with high-resolution fMRI and diffusion-weighted MRI (dMRI) at 7 tesla. Three levels of cognitive hierarchy were implemented in two domains: second language and nonlanguage. In fMRI, for the higher level, activations were found in the ventroanterior portion of the prefrontal cortex (PFC), the head of the caudate nucleus (CN), and the ventral anterior nucleus (VA) in the thalamus. Conversely, for the lower level, activations were located in the posterior region of the PFC, the body of the CN, and the medial dorsal nucleus (MD) in the thalamus. This gradient pattern of activations was furthermore shown to be tenable by the parallel connectivity in dMRI tractography connecting the anterior regions of the PFC with the head of the CN and the VA in the thalamus, whereas the posterior activations of the PFC were linked to the body of the CN and the MD in the thalamus. This is the first human in vivo study combining fMRI and dMRI showing that the functional specificity is mirrored within the cortico-subcortical loop substantiated by parallel networks.

Statistical learning is a powerful ability that extracts regularities from our environment and makes predictions about future events. Using functional magnetic resonance imaging, we aimed to probe how a wide range of brain areas are intertwined to support statistical learning, characterising its architecture in the whole-brain functional connectivity (FC). Participants performed a statistical learning task of temporally distributed regularities. We used refined behavioural learning scores to associate individuals' learning performances with the FC changed by statistical learning. As a result, the learning performance was mediated by the activation strength in the lateral occipital cortex, angular gyrus, precuneus, anterior cingulate cortex, and superior frontal gyrus. Through a group independent component analysis, activations of the superior frontal network showed the largest correlation with the statistical learning performances. Seed-to-voxel whole-brain and seed-to-ROI FC analyses revealed that the FC between the superior frontal gyrus and the salience, language, and dorsal attention networks were reduced during statistical learning. We suggest that the weakened functional connections between the superior frontal gyrus and brain regions involved in top-down control processes serve a pivotal role in statistical learning, supporting better processing of novel information such as the extraction of new patterns from the environment.