This study re-examined the VAN and the DAN neuroanatomy by co-registering individual network maps in a common functional space. We propose a new comprehensive model of these networks based on the convergence of functional, structural, and neurochemical findings. First, we confirmed the initial hypothesis that subcortical structures, namely the pulvinar, the superior colliculi, the head of caudate nuclei, and a group of brainstem nuclei, are constituent elements of the attentional networks. Second, we characterized the structural connections underlying functional connectivity. Deep brain nuclei are densely connected and structural network hubs. Third, we showed that the identified brainstem nuclei projections are spatially correlated with the acetylcholine α4β2 nicotinic receptors and serotonin and dopamine transporters.
Pulvinar is a high-order thalamic relay nucleus participating in cortical-thalamocortical circuits that modulate information processing (Sherman 2007). Cytoarchitectonically, the pulvinar is divided into four regions: the anterior pulvinar, the inferior pulvinar, the medial pulvinar, and the lateral pulvinar (Stepniewska and Kaas 1997). The medial pulvinar is particularly important in establishing connections with heteromodal association areas, such as the superior and inferior temporal, the inferior parietal, the dorsolateral prefrontal, and the orbitofrontal cortices (Bridge et al. 2016). In our model, the pulvinar regions with the highest statistical level were medial, and we demonstrated that they were structurally connected with VAN cortical areas, through fronto-pulvinar projections, and with DAN cortical areas, by fronto-pulvinar and parieto-pulvinar projections (Bos and Benevento 1975; DeVito 1978; Lemaire et al. 2011). Pulvinar lesions may provoke hemispatial neglect (Karnath et al. 2002). Decades ago, Sprague impressively found that hemispherectomy prompted symptoms of hemispatial neglect in cats which were attenuated by removing the contralesional superior colliculus (Sprague 1966; Krauzlis et al. 2013). This effect was later observed in humans (Weddell 2004). In our model, the pulvinar connects with the superior colliculi through the tecto-pulvinar fibers (Luppino et al. 1988), demonstrating the importance of pulvinar - superior colliculi interactions in attention processes. Therefore, in the context of the so-called Sprague effect, removing the contralesional superior colliculus in cats with hemispatial neglect would damage the spared attentional network and might partially compensate for the imbalance in the attentional processing (Vuilleumier et al. 1996; Bartolomeo et al. 2007). Recently, hemispatial neglect was linked to lesions of the human superior colliculus (Nyffeler et al. 2021). The Sprague effect is also mediated by the pedunculopontine nuclei (Durmer and Rosenquist 2001; Valero-Cabré et al. 2020), which is one of the brainstem nuclei included in our model. The pedunculopontine nuclei possess a population of cholinergic neurons in their caudal portion, giving rise to a distinct network that regulates attentional states and enhances the processing of salient stimuli (Mena-Segovia and Bolam 2017). The descending projections from these cholinergic neurons innervate the nucleus pontis oralis (Mena-Segovia et al. 2008) and the gigantocellular nuclei (Martinez-Gonzalez et al. 2014), while their dorsal ascending projections innervate the colliculi (Jeon et al. 1993; Motts and Schofield 2009) and several nuclei of the thalamus, including the pulvinar and the mediodorsal nuclei (Steriade et al. 1988). The pattern of the pedunculopontine projections closely matches the brainstem and thalamic map evidenced in our analysis. Hence, lesion analyses and axonal tracings studies confirm the validity of our subcortical model of the VAN and the DAN.
The neurotransmitter system correlation analysis reinforced the proposed relationship between the subcortical nuclei of the attention networks. The highest spatial correlation of both networks was with the acetylcholine α4β2 nicotinic receptors. The acetylcholine α4β2 nicotinic receptors have a well-established relationship with sustained attention. Acetylcholine α4β2 nicotinic receptors agonists reduce distractibility of adult monkeys during the performance of matching-to-sample tasks with distractors (Prendergast et al. 1998) and increase the firing rate of dorsolateral prefrontal neurons during sustained attention tasks, an effect that is reversed by the co-administration of receptor antagonists (Sun et al. 2017). In humans, transdermal nicotine administration improves attentiveness (Levin et al. 1998; Valentine and Sofuoglu 2017). All these observations in animals and humans support the critical role of the subcortical acetylcholinergic system in attentional processes.
The VAN and DAN brainstem nuclei projections were also spatially correlated with the distribution of dopamine and serotonin transporters. This finding is consistent with the psychopharmacological knowledge about attention. Methylphenidate is the first-line treatment for attention deficit hyperactivity disorder (Cortese et al. 2018). Pharmacologically, it is a noradrenaline-dopamine reuptake inhibitor with higher potency for dopamine transporters (Gatley et al. 1996; Faraone 2018). Modafinil is a selective inhibitor of dopamine transporters (Zolkowska et al. 2009) and produces attention enhancement effects (Turner et al. 2004; Repantis et al. 2010). Further studies are needed to understand how the interplay between the nicotinic acetylcholine and the dopamine systems occurs in attention networks, but it might be mediated by their interaction at the levels of the striatum (Zoli et al. 2002; Exley and Cragg 2008) and midbrain (Blaha and Winn 1993; Forster and Blaha 2003). Serotonin reuptake inhibitors also modulate attentional processes (Harmer and Cowen 2013). They increase the perceptual bias towards emotional stimuli (Harmer et al. 2004; Browning et al. 2007) by regulating the activity of visual processing circuits (Harmer and Cowen 2013). Therefore, our improved model of the DAN and VAN functional neuroanatomy appears to reconcile previous neuroimaging and pharmacological findings. As previsouly suggested [INSERT CORBETTA 2008] additional pharmacological studies will be required to understand the preferential association of VAN with acetylcholine α4β2 nicotinic receptors. Similarly pharmacological studies are required to shed light on the effect of serotonin transporter on the VAN and to reveal the relationship between dopamine transporters and the DAN. Finally, understanding the relationship between the neurochemical signature and hemispheric functional dominance still require more research in animals and humans (Corbetta et al. 2008).
Characterizing the human brain's subcortical anatomy of attention networks fosters the exploration of a common structural-functional attentional framework across species. Attention is far from being a specific cognitive ability of human beings (Washburn and Taglialatela 2006). Species with either close or distant common ancestors in the phylogenetic tree, such as monkeys, rats, and pigeons, can scan, select and maintain attention to surrounding environmental stimuli (Mackintosh 1965; Blough 1977; Washburn and Taglialatela 2006; Wasserman and Castro 2021). A common subcortical attention framework may surpass the challenge of finding the cortical homologs of the human VAN and DAN in other species (Patel et al. 2015). Accordingly, future studies might use the subcortical areas we highlighted to explore comparatively the organization of the VAN and the DAN in non-human species.
In our analysis, VAN and DAN structural connectivity maps were right-lateralized. Regarding functional connectivity, the distribution of VAN was not different between hemispheres, and DAN was slightly left-lateralized. Task-based fMRI studies indicate right lateralization of the VAN (Downar et al. 2000; Fox et al. 2006), but the asymmetry might vary according to the nature of the task (Doricchi et al. 2010). Accordingly, while functional asymmetry is expected for some task-related activations (Shulman et al. 2010), resting-state functional connectivity may not capture function-specific asymmetries due to its global nature.
A limitation of our study is the inability to untangle the different roles and dynamic interactions between the proposed subcortical structures. While the cortical regions of the DAN and the VAN are quite neatly segregated (Vossel et al. 2014), the subcortical nuclei described in our model probably contributed to both the VAN and the DAN. Future investigations using our model to explore the BOLD signal during task-related fMRI in humans or direct electrical recordings in animals might better dissociate the hierarchical organization and functional role of subcortical regions than resting-state fMRI.
In conclusion, this work proposes an improved neuroanatomical model of the VAN and the DAN that includes the pulvinar, the superior colliculi, the head of caudate nuclei, and a group of brainstem nuclei interrelated with the acetylcholine nicotinic and the dopamine and serotonin transporter systems. This novel framework reconciles behavioural, electrophysiological, and psychopharmacological data and provides a shared fundation to explore the neural basis of attention across different species and brain pathologies.