A challenge imposed by the concept of vigilance is its varied meanings and applications across different fields. In neurophysiology or psychiatry, the meaning of vigilance is more tied in with physiological, either healthy or pathological, fluctuations of arousal. Neurophysiologists place vigilance as an intermediate state within the sleep-wake cycle, which can range from hypervigilance (over-excited), to vigilant (relaxed awake state), to a drowsy or hypo-vigilant, and a sub-vigilant state that transitions into sleep (Klösch et al., 2022; Oken et al., 2006). Psychiatrists, instead, refer to abnormal states of vigilance. On the one hand, they consider hypervigilance as a heightened attentiveness and response toward the environment, that may lead to perceiving innocuous stimuli as threats and is often observed as a clinical symptom of post-traumatic stress disorder (American Psychiatric Association, 2013; Oken et al., 2006). On the other extreme, hypo-vigilance is considered as a dampened responsiveness toward the environment, observable in depression (Weinberg and Harper, 1993). This shows how diversely the term “vigilance” is defined and used across disciplines. However, even within cognitive psychology and neuroscience, the term lacks a clear, consistent definition and is often used interchangeably with related phenomena such as arousal, alertness, and sustained attention. This section aims to disentangle a less ambiguous definition of vigilance by more clearly separating it from these overlapping but relatively distinct phenomena.

A first broad distinction can be made in terms of the attentional component of direction or focus, i.e., cortical activity that is directed or focused toward a specific stimulus, task, or purpose (van Schie et al., 2021). In this sense, direction refers not just to readiness to respond but to the continuous selection and monitoring of specific, task-relevant inputs over time, which goes beyond the more global readiness state that may facilitate fast responses or anticipate the selection of motor responses. This distinction based on direction, as depicted in Figure 1A, permits to jointly categorize arousal and alertness as processes attributed to cortical activity without a specific direction or selectivity, and to differentiate them from processes that do require a direction, such as vigilance and sustained attention.

Regarding processes with no direction, both arousal and alertness refer to a more generalized readiness to react, that subserves and facilitates other more complex cognitive processes. However, the two phenomena can be distinguished based on the level at which this readiness to react manifests. Arousal can be understood as a general physiological state of being awake or reactive to the environment, more in line with Head's original concept of vigilance (Head, 1923). It encompasses different levels of consciousness, from drowsiness or hypo-arousal states (such as sleep) to the opposite extreme of hyper-arousal (Aston-Jones and Cohen, 2005; Klösch et al., 2022; Unsworth and Robison, 2017). Arousal has been considered a pre-requisite for adequate cognitive processing (Aston-Jones and Cohen, 2005), and specifically for vigilance or sustained attention performance (Esterman and Rothlein, 2019). The monitoring required by vigilance tasks requires a certain level of cortical activation that depends on arousal (Sarter et al., 2001). In fact, it is known that the effects of arousal on these attentional processes are modulated by the effects of norepinephrine (NE) released by the locus coeruleus (LC): as low or high locus coeruleus activity is associated with poor task performance, either due to low task engagement (hypo-arousal) or over-active but nonspecific task engagement (hyper-arousal), respectively (Esterman and Rothlein, 2019). Thus, as depicted in Figure 1B, a state of preparedness or readiness to react at the physiological level—when properly balanced—can be considered as a filter that allows the input of adequate levels of task-relevant and task-irrelevant information.

Alertness, on the other hand, supported by arousal on the physiological level, refers to a psychological dimension of this state of readiness to react and respond to the environment—more in line with Mackworth's (1948) above-mentioned definition of vigilance. Alertness depends on an optimal level of arousal, allowing adequate sensitivity to incoming stimuli (Posner, 2008). Alertness has additionally been subdivided into a tonic component, referring to slow changes associated with circadian rhythms, where the level of cortical activity allowing responsiveness to the environment is sustained for longer periods and experiences slow fluctuations over time (Posner, 2008; Sturm and Willmes, 2001); and a phasic component, which alludes to quicker or momentaneous switches into this state of readiness, that occur in response to an external cue or stimulus or self-initiated due to the expectation of a stimulus (Petersen and Posner, 2012; Sturm and Willmes, 2001). Although alertness is understood as a general, non-specific state of readiness, its expression within task contexts can sometimes appear directional. For instance, when tasks include specific cues or known response mappings, they may produce preparatory neural activity, such as electrophysiological contingent negative variation (CVN; Walter et al., 1964), which reflects task-induced anticipation (Van Boxtel and Böcker, 2004; Walter et al., 1964), and thus heightened cortical activity for a specific sensory modality or motor response. However, this directionality should not be attributed to alertness itself, but rather to more voluntary mechanisms such as temporal orienting (Correa et al., 2006; Nobre et al., 2007), or task-induced constraints that shape how alertness is expressed at the behavioral and neurophysiological level (Sarter et al., 2001). Furthermore, classic models of attention state that alertness remains fundamentally non-selective and diffuse, facilitating more targeted attentional processes such as orienting or vigilance (Petersen and Posner, 2012; Posner, 2008; Sturm and Willmes, 2001).

Thus, while tonic alertness has sometimes been equated to vigilance (Posner, 2008) and sustained attention, they would differ in the lack of direction or selective focus associated with vigilance, still reflecting a more general and diffuse state of preparation. Yet, despite this fundamental non-selectivity of alertness, recent accounts have argued that in certain situations, such as under increased urgency or emotional saliency, high (phasic) alertness may temporarily bias or override the selection of actions or responses. For instance, Poth (2021) and Krause and Poth (2025) describe how external stimuli can sometimes override top-down control, triggering stimulus-driven actions that conflict with current intentions, or temporal expectancies (Cappucci et al., 2018). These phenomena suggest that under specific conditions, arousal and alertness may modulate the readiness of certain motor plans, leading to a context-dependent directionality in behavior. However, we propose that this directionality arises not from an inherent selectivity of arousal or alertness themselves, but rather from automatic interactions between these global readiness states and environmental affordances or task demands. In other words, arousal and alertness remain fundamentally non-specific states (Weinbach and Henik, 2012); although their behavioral expression can become more directional under certain contextual conditions, that override the current task goal with a more relevant one in the moment (e.g., survival).

The two processes with direction—vigilance and sustained attention—are often used interchangeably (Klösch et al., 2022; Oken et al., 2006; Sarter et al., 2001), as both require the focus of directed attention on a task over a prolonged period. However, one can distinguish between the two in terms of the intensity of information processing that is required (van Zomeren and Brouwer, 1994; Zimmermann and Leclercq, 2002): whereas vigilance would refer to the detection of infrequent (and potentially harder to detect) changes in the environment, sustained attention would require more active and ongoing processing toward a broader set of stimuli (Singh-Curry and Husain, 2009), as schematically depicted in Figure 1C. For example, vigilance, on the lower end of the intensity continuum, might involve driving down a long, straight highway with minimal traffic, where responding to external stimuli is rare (e.g., adjusting speed in accordance with a speed-limit change or braking when noticing that cars ahead are doing so). On the opposite end of the intensity continuum, sustained attention, exemplified by driving through city traffic at rush hour, requires constant attention to a rapidly changing and stimulating environment with potentially several different foci to be attended simultaneously (e.g., traffic lights, pedestrians about to cross the street, other cars, etc.). Importantly, intensity is not only dependent on the frequency of target stimuli, as it may also interact with the saliency of targets or the processing they require; where higher target saliency may facilitate detection (Helton and Warm, 2008; Smallwood, 2013). In cases of lower saliency and low task demands (e.g., simple detection), participants must sustain control settings for selective attention over long intervals with little immediate reward or reinforcement, a situation that may promote disengagement or exploratory shifts in attention unless sufficient task utility is maintained (Aston-Jones and Cohen, 2005).

Based on this differentiation from other phenomena, we can propose that vigilance may be defined as the ability to monitor the environment and detect rare but critical stimuli. This definition accurately reflects the directionality (cortical activity directed toward a specific critical stimulus or task) and the low intensity (rare appearance of the critical stimulus in the environment) of vigilance. It also aligns with the working definition used across several research projects authored or co-authored by the current authors (Cásedas et al., 2022; Hemmerich et al., 2023, 2024; Luna et al., 2021a, 2018, 2020, 2021c, 2022a,b; Román-Caballero et al., 2021; Sanchis et al., 2020). In line with this definition, the decrement of vigilance with time-on-task can be understood, not as a decline in arousal or alertness (although these processes subserve, and therefore, influence vigilance performance), but rather due to a loss of maintaining a continuous directional focus over a selected stimulus over time. Section 2 will cover theories that have been established to explain why the vigilance decrement takes place.

Finally, it is important to clarify the functional role we attribute to the concept of vigilance in our framework. While vigilance is often operationalized in terms of behavioral outcomes (e.g., target detection rates, reaction time variability, or performance decrements over time), our definition aims to go beyond the behavior itself, presenting vigilance as an explanatory construct, i.e., a latent cognitive state characterized by sustained and directed monitoring for infrequent but critical stimuli. This state is inferred from empirical observations but not reducible to them. Making this distinction is important for interpreting results: behavioral indicators are necessary to measure vigilance, but our goal is to understand the underlying process that gives rise to those observable patterns. The following section will briefly cover aspects related to the empirical observation of the construct, to highlight how the vigilance decrement manifests over time, and which tasks and specific measures have been used to identify it.

Common paradigms used to assess vigilance have been summarized in Table 1, so as to give a better overview of how it is usually operationalized in research. Across the studies that employ these tasks, a certain overreliance on hit rates, accuracy measures, and simple reaction times (RTs) is noted. Some paradigms allow for the extraction of Signal Detection Theory (SDT) measures (Stanislaw and Todorov, 1999), that offer a more nuanced interpretation of performance. Specifically, the vigilance decrement may arise from a shift toward a more conservative response criterion (e.g., β) or a decline in perceptual sensitivity (e.g., d′). Recent research has highlighted the importance of disentangling these components. Thomson et al. (2016) argue that much of the observed vigilance decrement may be better accounted for by strategic shifts in response bias rather than by a true loss of perceptual sensitivity. In their view, motivational and effort-related factors, rather than purely sensory fatigue, drive participants to adopt a more conservative criterion to minimize errors. Luna et al. (2022b) further stress that the use of SDT measures can uncover hidden dynamics in attentional control: in some cases, sensitivity may retain stable while response bias fluctuates, revealing changes in cognitive control, motivation, or perceived task demands rather than perceptual degradation per se. Despite these insights, SDT metrics remain underutilized in many vigilance studies, limiting our ability to specify the cognitive processes underlying performance changes. Recent evidence inspecting psychometric curves have allowed identifying which SDT measures contribute most to the vigilance decrement. McCarley and Yamani (2021) observed that a shift toward a more conservative response criterion, decreased sensitivity, and increased attentional lapses were associated with the vigilance decrement. However, subsequent studies using the same approach have identified only the response bias and lapses as robust predictors of the vigilance decrement (Gyles et al., 2023; Román-Caballero et al., 2023).

Most of the research that has attempted to establish a time-course of the vigilance decrement converges on the findings that rather than a steady linear decline with time-on-task, performance declines steeply in the early phases of a task and then plateaus or presents a less steep decline. This pattern was already evident in early experimental work, such as Mackworth's (1948) Clock Test, where declines in signal detection were most prominent within the first 30 min, followed by a steadier decline in the remainder of the full 2 h of the task. Teichner's (1974) review on vigilance studies also observes that performance in vigilance tasks generally declines early on in the task, highlighting the need to include fine-grained measures in the time-domain to adequately characterize the vigilance decrement. Further work has supported this general pattern: modeling approaches have frequently used an exponential function to characterize the time course of the performance decline (Helton et al., 2007; Parasuraman and Jiang, 2012; Warm et al., 2008). However, Parasuraman and Jiang (2012) also point out that, critically, the established patterns are often only observed when the data is analyzed at a group level, but that individual vigilance trajectories over time do rarely adjust well to such (e.g., exponential) fits, suggesting that this pattern may be a useful heuristic only at the group level. However, it does not necessarily capture the full heterogeneity in how vigilance fluctuates within and across individuals, as well as across tasks, task conditions, or assessed dependent variables. Surprisingly, the temporal aspect of vigilance is severely under-reported, with many studies presenting only mean aggregated values across the task, while not investigating how performance or SDT measures may fluctuate over time (e.g., at the block or trial-level), further obscuring an adequate understanding of a precise time-course of vigilance.

Lastly, more critical accounts of the vigilance decrement have suggested that it may, in fact, be an iatrogenic phenomenon (Hancock, 2013, 2017). Hancock argues that the decrement arises due to the artificial imposition of the vigil itself, claiming it is more a product of laboratory design than a phenomenon with real-world validity. This critique also raises the issue of distinguishing empirical observations of a phenomenon from their mechanistic explanation. In the case of vigilance, however, there is a certain overlap in the mechanistic or theoretical definition of the construct and its operationalization. In this sense, direction and intensity are both core components of the mechanisms defining vigilance, as well as aspects elicited by vigilance tasks. While critical points such as those raised by Hancock (2013) are valuable reminders to continuously interrogate the assumptions behind experimental paradigms, we argue that the core function of sustained environmental monitoring remains essential in many real-life settings. Indeed, current empirical evidence supports that vigilance-like performance failures do occur outside of controlled laboratory conditions. For example, studies on automated driving have consistently shown that drivers' ability to detect critical events declines over time, even in realistic, safety-critical contexts (Biondi et al., 2024; McWilliams and Ward, 2021). Similar decrements have been observed in surveillance and military monitoring tasks (Wohleber et al., 2019), supporting the view that the vigilance decrement reflects an actual impediment to directing the attentional focus on critical targets in real-life scenarios. That said, Hancock's (2013) critique also raises an important point about the ecological validity of experimental findings. Enhancing ecological validity would not only clarify the generalizability of lab-based results but also support the development of protocols better suited to real-world and clinical applications.

Several accounts integrate both underload and overload across a continuum, wherein a middle ground for optimal performance can be achieved. These accounts often explain that vigilance performance depends on the degree of arousal (Esterman and Rothlein, 2019) or cognitive load (McWilliams and Ward, 2021), following the reverse U-shaped function that Yerkes and Dodson (1908) used to relate stress and cognitive performance. In this regard, underload would lead to what is coined as passive fatigue, whereas overload would lead to active fatigue (McWilliams and Ward, 2021; Saxby et al., 2013). A recent study has observed that tasks with both low and high cognitive load led to a pronounced vigilance decrement, whereas performance did not decay during a task with an intermediate cognitive load (Luna et al., 2022a), although this effect is not always replicated (Hemmerich et al., 2024).

Thomson et al. (2015a) highlighted gaps within the underload and overload theories, proposing the resource control account. This model operates on the notion that: (i) the amount of cognitive resources we have available is constant (i.e., resources are not depleted as time progresses), (ii) the default state of the mind is mind-wandering, and (iii) with time-on-task our ability to exert executive control in order to maintain attention focused on the task at hand decreases. This decline in executive control would progressively hamper the ability to allocate mental resources toward the task at hand, as they gradually shift to support other task-unrelated thoughts, i.e., mind-wandering (Cunningham et al., 2000; McVay and Kane, 2012; Thomson et al., 2014), leading to the observed vigilance decrement in the task. While this theory has received some initial empirical support demonstrating the role of declining executive control as an explanation for the vigilance decrement (Luna et al., 2022b), it has not been extensively tested. Furthermore, the model posits that executive control decreases with time-on-task, not due to a lack of resources—since these are assumed to remain constant—but rather due to the adoption of less effortful processing strategies. This shift is believed to occur as participants adjust their performance to the low probability of encountering the critical target (Thomson et al., 2015a). However, this explanation for the decline in executive control remains somewhat abstract and theoretical. The following two models provide a more detailed examination of potential factors driving the adoption of different processing strategies as time-on-task progresses.

Another relevant model that also considers that individuals flexibly adapt their performance during a task is the opportunity-cost model. While it has been defined more broadly for overall cognitive control (Kurzban, 2016; Kurzban et al., 2013), it can add an additional relevant perspective to explaining the vigilance decrement. The opportunity-cost model also considers that we operate with a limited but constant set of cognitive resources. However, with the ongoing performance of a task, we unconsciously (i.e., without these evaluations necessarily raising to awareness) weigh the benefit of continuing with this performance against the cost of losing the opportunity to perform other, potentially more rewarding or engaging tasks (Kurzban et al., 2013). The relevance of this model lies in the fact that the vigilance decrement can be considered not merely in terms of the loss of an ability, expended resources, or loss of sensitivity, but rather as a process that is tied in a more complex manner to emotional and motivational factors (Kurzban, 2016). Boksem and Tops (2008) offer an interesting brain-based account of this cost-benefit model explaining performance declines, expanding the role of dopamine beyond reactivity to reward into a basic component for motivation-guided behavior. Furthermore, dynamic shifts in performance driven by motivational factors can also be tied in with Aston-Jones and Cohen's (2005) Adaptive Gain Theory, which proposes that shifts in task engagement are regulated by changes in tonic and phasic norepinephrine activity. From this perspective, the vigilance decrement may reflect a neurobiologically-driven shift from task-focused exploitation to exploratory behavior when the perceived utility of sustaining performance on the same task declines.

Many of the reviewed theoretical models that refer to “cognitive resources,” be it in the context of overload or allocation, often treat them as a fairly abstract concept (Grier et al., 2003; Thomson et al., 2015a; Warm et al., 2008). Christie and Schrater (2015) posit that a decline in cognitive performance with time-on-task can be accounted for by a depletion of the amount of glucose that is available to active neurons. However, they argue that cognitive performance is not solely dictated by raw resource availability. Instead, performance fluctuations can also reflect strategic allocation of metabolic resources based on reward contingencies. A key mechanism proposed in their model is the role of glycogen stored in astrocytes. Glycogen could act as a metabolic buffer, mobilized when tasks demand short bursts of elevated neural activity beyond the steady-state glucose supply. This allows for momentary performance recovery, even after initial depletion (Christie and Schrater, 2015). This offers an interesting integration of resource-depletion models and cost-benefit models, as the 2-fold expenditure of resources would explain different behavioral patterns based on cost-benefit analysis performed by the individual (i.e., tapping into glycogen stores to ensure continued performance, only if remaining in the current task is deemed a greater benefit or opportunity than switching to a different task). This could potentially explain the reports of null effects of hypoglycemia on sustained attention tasks (McAulay et al., 2001), if glycogen reserves are factored in as a putative compensatory mechanism. On the other hand, other accounts posit that declines in performance, and thus, indirectly the vigilance decrement, could be explained in terms of a “protective” neural mechanism against the potential damage of exerting extended high control over extended periods, which is experienced as cognitive effort (Holroyd, 2024). In this view, cognitive effort is tied not only to resource use but to mechanisms that limit prolonged high activation to preserve long-term neural health. Overall, while the idea of energy-budget-based decision making adds depth to models of vigilance, more empirical work is needed to determine whether glucose and glycogen dynamics could play a causal role in the vigilance decrement.

Given the above-outlined overlap of vigilance with other attentional functions and its interaction with other cognitive processes, it is to be expected that it cannot be circumscribed to one specific neural location. In fact, it has been established by a coordinate-based meta-analysis on functional Magnetic Resonance Imaging (fMRI) and Positron Emission Tomography (PET) data that vigilance is related to neural activity distributed across different neural networks or clusters, many of which are lateralized toward the right hemisphere (Langner and Eickhoff, 2013). While Langner and Eickhoff (2013) considered a noticeably low duration criterion (>10 s into the task) to include studies, within the aforementioned identified areas, a further right-lateralization was observed when looking at foci of brain activity correlating with longer task durations (see Figure 3A). In line with these results, the right-lateralization of vigilance has also been reported from lesion studies. Patients who had suffered a lesion to right frontal regions, presented a larger vigilance decrement than patients with left frontal or other lesion sites (Koski and Petrides, 2001; Molenberghs et al., 2009; Rueckert and Grafman, 1996). A more recent study has additionally shown that patients with lesions to the right-hemisphere also present steeper within-block vigilance decrements compared to healthy controls (Brosnan et al., 2022). Further evidence of this lateralization has also been gathered from neuroimaging studies with healthy participants. An earlier study showed that right frontal and parietal areas show activation during vigilance tasks in PET imaging (Pardo et al., 1991). On the other hand, perfusion fMRI data has shown that blood flow in the frontoparietal network is reduced from pre- to post-task, and this reduction in blood flow was associated with a vigilance decrement (Lim et al., 2010). In line with this right-lateralization of vigilance, several studies report improved vigilance with tDCS applied over the right frontal cortex (Brosnan et al., 2018), as well as both tDCS (Luna et al., 2020) and tRNS (Harty and Cohen Kadosh, 2019) applied over right frontoparietal regions. In contrast, absent effects on vigilance have been frequently reported with anodal tDCS protocols over left frontal regions (Coulborn and Fernández-Espejo, 2022; Dai et al., 2022; Filmer et al., 2019; Hussey et al., 2020; Martínez-Pérez et al., 2023). However, some beneficial effects have also been reported with anodal tDCS over left frontal regions (Gan et al., 2022; McIntire et al., 2014), and also with bilateral targeting of right and left frontal regions with opposing polarities (Nelson et al., 2015, 2014). These findings support the dominant role of right-hemisphere structures in vigilance but suggest that left-hemisphere involvement, possibly through compensatory mechanisms, should not be overlooked. It is possible that the vigilance decrement also reflects an unbalanced inter-hemispheric synchronization, which may be mitigated by external stimulation that helps restore this balance.

The regions identified by Langner and Eickhoff (2013) show an overlap with networks identified in other attentional models, such as the dorsal top-down stream and the ventral bottom-up stream identified by Corbetta and Shulman (Corbetta et al., 2008; Corbetta and Shulman, 2002). These two streams make up the orienting network identified by Posner and Petersen (1990) as depicted in Figure 3B, which regulates goal- and stimulus-driven allocation of attentional resources to relevant stimuli (Petersen and Posner, 2012; Posner and Petersen, 1990). Furthermore, some overlap can also be observed with the executive control network (Petersen and Posner, 2012; Posner and Petersen, 1990), which encompasses what Dosenbach et al. (2007, 2008) characterized as a frontoparietal network, associated with initiating and adjusting control over ongoing performance; and the cingulo-opercular network, associated with a stable maintenance of task-goals across longer time periods (see Figure 3C). As depicted in Figure 3D, most of these regions are reached by the alerting network, composed of the cortical projections of the LC. Further potential contributions for adequate vigilance functioning can be found in relation again to the cingulo-opercular network, this time identified as part of the salience network (SN), which has been proposed to assist in balancing exogenous or task-driven activity in the central executive network (CEN) and the more endogenous or self-referential activity of the default mode network (DMN), as shown in Figure 3E (Menon, 2011; Menon and Uddin, 2010). Achieving this balance could play a key role in regulating neural activity for task-related thoughts and neural activity for self-referential thoughts or mind-wandering. Lastly, in a more directional model, Unsworth and Robison (2017) propose that the inhibitory effect of the frontoparietal network (FPN) on the DMN is aided by the SN, driven by the projections of the LC, as shown in Figure 3F. However, in relation to this last model, a recent study using taVNS, showed effects on the LC norepinephrine system with active stimulation, but no effects on vigilance performance (Luna et al., 2025).

In line with the prior established roles of attentional networks and the DMN, it must be noted that the notion of the DMN as task-negative, or the attribution of its activity with degraded performance has been challenged by findings from Esterman et al. (2013), indicating that instead, a push-pull relationship between the DMN and the dorsal attention network (DAN) subserves different attentional states. An “in the zone,” more stable and automatic processing that can arise in less challenging tasks, is characterized by higher DMN activity, and permits less effortful processing at the expense of risking errors if DMN activity increases past a certain threshold. During more demanding tasks, a more effortful “out of the zone” state emerges, characterized by higher activity in the DAN, where errors are more likely to occur if insufficient control is exerted by the DAN (Esterman et al., 2013).

While sustaining the idea that there is no unique location that subserves vigilance, the right posterior parietal cortex (rPPC) may play a fundamental role in permitting adequate vigilance performance. The rPPC—integrated by the superior parietal lobule (SPL) and the inferior parietal lobule (IPL)—plays a crucial role in spatial attention, given that the IPL is the main lesioned area in hemispatial neglect (Malhotra et al., 2009; Molenberghs et al., 2009). However, neglect patients often present additional deficits in vigilance/sustained attention (Malhotra et al., 2009), which highlights the involvement of this region in vigilance functioning. Furthermore, the rPPC shows a heightened hemodynamic response to the presentation of infrequent (Stevens et al., 2005) and novel (internal and external) stimuli (Singh-Curry and Husain, 2009). Additionally, it has also been associated with the active maintenance of task goals (Singh-Curry and Husain, 2009). This has led some authors to establish the rPPC as a “convergence node” between the ventral attention network and the DMN: thus considering its relevant role in maintaining task goals active, whilst flexibly reacting toward novel or salient stimuli and relaying between task-relevant and task-irrelevant regions (Giacometti Giordani et al., 2023). This role can be feasible on a structural level due to the densely interconnected core that has been observed in this region, with further dense connections to other neural regions (Hagmann et al., 2008). Conceptually, the rPPC could play a relevant role as a relay switch in the complex interplay of forces that lead to the vigilance decrement: resources, mind-wandering, executive control, motivation, and cost-benefit analyses. Further support for the potential involvement of the rPPC in vigilance, specifically executive vigilance, comes from findings of a mitigated EV decrement when this area was targeted with tDCS, showing promising behavioral and neurophysiological effects (Hemmerich et al., 2023; Luna et al., 2020).

As a counterpoint, some accounts suggest that the right-lateralization of vigilance is observed only in simpler, less demanding tasks, whereas in more complex tasks a bilateral hemispheric activation is observed (Helton et al., 2010). This observation highlights the fact that despite the above-discussed relevance of the rPPC for vigilance, the importance of broad networks in supporting the adequate functioning of vigilance must be considered. In line with this, Rosenberg et al. (2016) have established a connectome-based predictive model that can predict individual differences in sustained attention functioning from task-based as well as resting-state functional connectivity data. This model can predict attentional fluctuations within and between task blocks and sessions, as well as responsiveness to external modulations of attention, such as the administration of sedatives (Rosenberg et al., 2016). The model has also proven to effectively predict attention-deficit symptom severity in an independent sample (Rosenberg et al., 2020). Interestingly, this model includes regions beyond the canonical regions associated with attention (salience, frontoparietal, or default mode networks), and implicates other regions such as the cerebellum (Rosenberg et al., 2016). Although, as already mentioned, an intervention with tDCS targeting the cerebellum has not shown significant effects on vigilance functioning (Erdogan et al., 2023); which could suggest that these non-canonical regions have less principal roles, and participate more at the network level to promote adequate levels of vigilance.

Diffusion Weighted Imaging (DWI) offers insight into structural anatomical features that are highly relevant for the adaptive signal transmission required by attentional processes, by, for example, linking the integrity of white matter tracts to attentional functioning. Considering the above-reviewed evidence, pathways connecting frontoparietal areas, such as the branches of the superior longitudinal fasciculus (SLF) could be of special interest (Thiebaut De Schotten et al., 2011). The SLF connects frontal, temporal, parietal, and occipital regions and has been subdivided into three branches: SLF I (dorsal), SLF II (medial), and SLF III (ventral) (Janelle et al., 2022). As a case in point, a higher fractional anisotropy (FA) in the SLF in typically developing children has been associated with better sustained attention performance (Klarborg et al., 2013). Moreover, adolescents with ADHD show a strong relationship between reported inattentive symptomatology and alterations in the right SLF (Chiang et al., 2015). Furthermore, this link has also been established in healthy adults, where higher fiber density (FD, an estimate of axon density) of the SLF I was associated with fewer attentional lapses during a global-local task (Clemente et al., 2021). Furthermore, Luna et al. (2021b) observed that higher white matter integrity of the SLF I in healthy adults was associated with faster response times for correct responses in an EV sub-task. However, no significant associations were observed with other more reliable or direct indicators of the vigilance decrement (such as the decrement of hits or sensitivity with time-on-task) (Luna et al., 2021b). Theoretically, the SLF I is thought to mediate goal-directed attentional processes within the dorsal attention network. This activity may be modulated by the SLF II, which acts as a communication bridge between the dorsal and ventral attention networks, enabling the redirection of attention toward salient stimuli identified by the SLF III (Thiebaut De Schotten et al., 2011).

Another tract that may be relevant for vigilance is the right cingulate fasciculus (or cingulum), which runs around the corpus callosum (Catani and Thiebaut de Schotten, 2008). Higher FA in the cingulum has been associated with higher sensitivity to infrequent targets in an EV-like task (CPT, Takahashi et al., 2010); which might be related to this structure's role as a mediator between the FPN and DMN, balancing task-related and self-referential processes (Menon, 2011). The dorsolateral prefrontal-caudate tract has been associated with vigilance performance (Chiang et al., 2015), while the right inferior fronto-occipital fasciculus (IFOF) may play a role in response inhibition (Pironti et al., 2014), potentially helping to suppress task-irrelevant stimuli during vigilance tasks. The IFOF may also exert fast top-down control from frontal regions over visual areas, supporting attentional modulation of perception (Bartolomeo and Seidel Malkinson, 2019). Additionally, considering the detrimental effect of sleep deprivation on the vigilance decrement (Lim and Dinges, 2008), it is worth noting that DWI data has also been used to predict individual vulnerability to sleep deprivation. Wang et al. (2022) found that the integrity of the SLF, posterior corona radiata, anterior limb of the internal capsule, as well as the body and genu of the corpus callosum, best predicted vulnerability to sleep deprivation, suggesting that inter-individual differences in white matter structure may underlie resilience to attentional decline in extremer conditions. Moreover, Niogi et al. (2010) identified positive correlations between white matter FA and the functioning of Posner and Petersen's (1990) three attentional networks: the alerting network was linked to the posterior limb of the internal capsule (PLIC), the orienting network to the splenium of the corpus callosum, and the executive control network to the left anterior corona radiata. These associations suggest a distributed anatomical basis for distinct attentional components.

Importantly, examining how the effects of NIBS spread across the brain may offer deeper insights into the white matter structures that support vigilance-related functions. By stimulating a specific cortical region that serves as a node within a broader network, it is possible to observe network-level effects at both neural and behavioral levels (Hartwigsen and Silvanto, 2023; Momi et al., 2021). As a case in point, stimulating nodes of the DMN with TMS has shown to produce wider effects within white-matter wiring of the DMN (Esposito et al., 2022; Momi et al., 2021). Similarly, by stimulating nodes of the FPN with TMS, the involvement of the SLF has been prominently related to broader attentional and cognitive functioning (Botta et al., 2021; Martín-Arévalo et al., 2019; Martín-Signes et al., 2019, 2021, 2024; Quentin et al., 2016). Furthermore, the successful modulation of vigilance via transcranial electrical stimulation over right frontal, parietal, and frontoparietal regions (Brosnan et al., 2018; Hemmerich et al., 2023; Luna et al., 2020) mentioned in prior sections, could also be linked to reaching nodes of the FPN that then distribute across the whole structure. The associations with vigilance and deeper-set white matter structures might be more difficult to establish, as fewer nodes are available to be reached via NIBS. Nonetheless, the fact that these different networks subserve vigilance functioning also permits thinking about network effects and the distribution of effects from brain stimulation at a more indirect level. However, the more direct links that have been established with TMS, in predicting stimulation outcomes, remain underexplored in applications specific to vigilance and in relation to other NIBS techniques.