Attention is conceived by the majority of definitions as a mechanism for the selection of information (Posner, 1994). Given that we have a limited amount of processing resources and live in a very complex world, this function is of extreme relevance. Despite being grouped under the umbrella term of attention, diverse mechanisms are employed for the selection of information. One of the most relevant models (proposed by Petersen & Posner, 2012; Posner & Petersen, 1990) suggested that the human attentional system can be divided into three attentional networks, namely alerting, orienting, and executive attention; with each of them representing a different set of attentional functions.
The alerting network (Posner & Petersen, 1990; Sturm & Willmes, 2001) is in charge of preparing and maintaining a state of vigilance that allows high priority signals to be processed more efficiently (faster and better). Following an external warning signal, this system can boost a transient increase in the preparation of the cognitive system (i.e., phasic alertness). The orienting network (Posner, 1980; Posner & Petersen, 1990), on the other hand, supports the ability to select information from a specific spatial location (spatial orienting), or from different features of objects. Spatial attention can be oriented to the location where a salient or relevant stimulus occurs, which is known as exogenous or bottom-up orienting of attention. It can also be allocated according to our plans, intentions, or task goals, which is known as endogenous or top-down orienting of attention. The executive attention network is required in situations for which we do not have a learned schema of action, or our schemas are not appropriate (Petersen & Posner, 2012). These are typically novel, difficult, dangerous, or changing situations, ones that involve planning or decision making, or when there is a conflict or error that has to be detected and solved (Norman & Shallice, 1986).
Anatomically, these attentional systems are supported by distinct neural networks (Petersen & Posner, 2012; Posner & Petersen, 1990). The anterior alerting system comprises a network of midbrain and thalamic areas, as well as frontal (the anterior cingulate cortex [ACC] and the dorsolateral prefrontal cortex [dlPFC]) and inferior parietal areas (Clemens et al., 2011; Sturm et al., 1999; Sturm & Willmes, 2001). In addition, tasks including warning signals activate left fronto-parietal areas (Coull et al., 2001; Fan et al., 2005; Yanaka et al., 2010) and bilateral extrastriate regions (Thiel et al., 2004). The model of Corbetta and Shulman (2002) proposes two separate but interacting networks for spatial attention: a bilateral dorsal fronto-parietal system, including the superior parietal lobe (SPL), the intraparietal sulcus (IPS), and the frontal eye field (FEF), involved in orienting and maintaining spatial attention; and a right-lateralized ventral fronto-parietal system, including the bilateral temporoparietal junction and the right middle and inferior frontal gyri (MFG/IFG), involved in attentional re-orienting to unexpected or salient events (Corbetta et al., 2008; see also Thiel et al., 2004 and most recently Alves et al., 2022). Finally, a broad network of regions has been related with executive attention, including the ACC, the dlPFC, the supplementary motor area (SMA), the anterior insula, the premotor cortex, the posterior parietal cortex, and the inferior frontal junction (Braver, 2012; Cole & Schneider, 2007; Macdonald et al., 2000; Miller & Cohen, 2001; Posner & Petersen, 1990).
As mentioned above, our current knowledge about the contribution of grey matter brain areas to healthy attentional functions is quite extensive, thanks to the use of functional magnetic resonance imaging (fMRI), among other neuroimaging techniques. However, evidence concerning the anatomical pathways connecting and functionally linking the attentional brain networks is much more recent and scarce. Over the last years, the development of methods such as diffusion-weighted imaging (DWI) has allowed us to investigate in vivo the structural brain connectivity and the micro- and macrostructural properties of white matter tissue. Nevertheless, to the best of our knowledge, only a few studies have examined the brain connectivity underlying the healthy human attentional networks. Moreover, the evidence from these studies has not always been consistent. The alerting network has been linked to the left posterior limb of the internal capsule (Niogi et al., 2010), the right dlPFC-caudate tract, the splenium of the corpus callosum (Luna et al., 2021), the cerebellothalamic tract (Ge et al., 2013) and the superior longitudinal fasciculus (SLF, Chica et al., 2018). Some studies have associated the orienting network with the splenium of the corpus callosum (Niogi et al., 2010) and the SLF (Carretié et al., 2012; Chica et al., 2018; Ge et al., 2013; Thiebaut de Schotten et al., 2011). Finally, the executive attention network has been related with the anterior corona radiata (Ge et al., 2013; Niogi et al., 2010) and the SLF (Crespi et al., 2018; Sasson et al., 2012, 2013; Smolker et al., 2018).
Most of the results of these DWI studies are based on correlations between different measures of white matter properties (e.g., fractional anisotropy, Le Bihan et al., 2001, or hindrance modulated orientational anisotropy, Dell’Acqua et al., 2013) and behavioral indices of attentional functions. Therefore, these investigations are of relevance to understand how relatively stable characteristics such as the microstructural properties of the relevant anatomical connections correlate with cognitive functions (tract-function correlations). However, they do not reflect the functional involvement of white matter during attentional tasks, as the existing neuroimaging methods did not allow such relationships to be established. In this vein, some recent developments have allowed the detection of task-related and resting-state white matter fMRI signals (Gawryluk et al., 2014; Gore et al., 2019). However, while very promising, this approach needs more time and practice to be fully understood and generalized. Recently, new proposals have also tried to integrate functional and structural data. The functionnectome approach (Nozais et al., 2021) projects the fMRI signal from grey matter voxels to the white matter, and weights the signal by the probability of structural connections, which are derived from 100 high-resolution DWI datasets. These new methods hold much promise to integrate function and structure to further understand core questions in the field of cognitive neuroscience beyond localization of function, such as how do function and structure interact? Or how does structure support cognitive function? Indeed, since the very first post-mortem anatomical dissections, the study of white matter has allowed the scientific community to move from a more localizationist approach to a more associationist approach. The development of DWI enriched the comprehension of brain lesions and neurological syndromes, allowing a better understanding of the functional models of the brain and enabling the discovery of individual variability and its relationship with healthy cognition, recovery after brain lesions, and response to neuromodulation (Assaf et al., 2019; Forkel et al., 2022). We consider the integration of function and structure to be vital to inform current models of human cognition and, in particular, to shed further light on theoretical accounts of human attentional networks.
Thus, the main aims of the present work are to examine the involvement of anatomical circuits in the functional correlates associated with the alerting, orienting, and executive attentional networks, as well as to contribute to the scarce, and sometimes inconsistent, knowledge of white matter contributions to healthy attentional functions. To this end, we have utilized a new method with high sensitivity, specificity, and reproducibility that allows the exploration of the involvement of anatomical circuits in specific cognitive functions (the functionnectome, Nozais et al., 2021). Based on previous studies, we hypothesized the main fronto-parietal association tract, the SLF, would be involved in all three attentional networks (Carretié et al., 2012; Chica et al., 2018; Crespi et al., 2018; Ge et al., 2013; Sasson et al., 2012, 2013; Smolker et al., 2018; Thiebaut de Schotten et al., 2011). We also expected a left lateralization of the alerting network (Fan et al., 2005; Yanaka et al., 2010) and a right lateralization of the orienting network (Thiebaut de Schotten et al., 2011). Additionally, we expected that the alerting network would recruit projection tracts, given that the anterior alerting system depends on cortico-subcortical interactions (Clemens et al., 2011; Sturm et al., 1999; Sturm & Willmes, 2001). Finally, we predicted that the executive attention network would chiefly rely on white matter tracts connecting the frontal lobe with the rest of the brain.