**1. Introduction**

The human brain continuously receives sensory and cognitive inputs, including even unimportant and unnecessary information for thriving and survival. Therefore, our attentional system must carry out a selection of the most relevant information to be thoroughly processed. Two different neural systems are involved in carrying out this selective function: An endogenous or voluntary system and an exogenous or automatic system [1,2]. Both hemodynamic and event-related potential (ERP) source neuroimaging studies identified in the frontal and in the temporo-parietal lobes the visual-spatial attentional endogenous and exogenous control areas [3,4], respectively. Corbetta and Shulman [3] and Shulman et al. [5] identified the endogenous attention system in the dorsal frontal-parietal areas. Indeed, both processes would primarily engage higher-level cortical circuits including, for endogenous orienting, frontal, parietal and temporal regions, and particularly the frontal eye fields (FEF) and the intraparietal sulcus (IPS) [6,7], and, for exogenous orienting, the right temporo-parietal junction (TPJ) [5,8]. Interestingly, the left frontal areas were more active than the parietal areas, but only during endogenous attention. In alerting tasks, endogenous and exogenous attention interact, and there is an increase both in global perceptual sensitivity and in the perception of exogenous spatial cues. Recently, Han and coworkers [9] showed that anterior insula (AI) is the key structure in endogenously reorienting of attention after reflexive attention is captured by a salient distractor. In tasks in which the participants know that a target will appear after a cue, brain responses will be optimized, and endogenous attention will be deployed [10]. There is an increased activation of the dorsal system when a stimulus appears at the same point in space of the cue. Top-down signals originate in the dorsal system and modulate sensory areas activity, detecting the current position of the presented stimuli [11]. Stimuli presented outside the attentional focus automatically catch attention and activate exogenous attention control in consequence of which ventral parietal areas and frontal cortex are activated. These stimuli activate the right-sided TPJ and the left-sided inferior frontal gyrus (IFG) [5]. Compared to exogenous spatial cues, vigilance tasks induce changes in endogenous and exogenous attention interactions and perceptual sensitivity. The sensitivity is reduced when stimulus onset is predicted, because an exogenous cue and the prediction time prepare participants to take over next stimulus [9]. Neurocognitive attentional processes related to the environmental input generate brain waves and electrical potentials with specific frequencies and amplitudes. In attentional processes, alpha band frequency, ranging between 7–14 Hz, plays a dominant role. Alpha electroencephalogram (EEG) activity is renown, like a typical response of the brain during relaxed wakefulness, for example, when the eyes are closed and/or attention is not heeded to any incoming information. Overall, this EEG frequency band has an occipito-parietal scalp distribution. It has been proposed that alpha oscillations provide pulsed inhibition for gamma activity and thereby dynamically route cortical information flow [12]. Kelly and colleagues [13,14] showed that alpha band activity was synchronized during visual selective or sustained attention tasks based on the presentation of visuospatial cues. The increase of this activity showed to be related to the inhibition of active processing of irrelevant and unexpected stimuli and contributed to the sustaining of top-down attentional control. Magnetoencephalography (MEG) studies (e.g., [15]) on visual attention perception also showed a pre-stimulus alpha activity. Most interestingly, a synchronization of alpha power in the parietal-occipital sulcus showed to be related to a functional modulation of information discrimination ability as undergirded by an overall inhibition of both the posterior occipital areas and the dorsal visual stream. This inhibition blocked out irrelevant information and avoided interference in working memory. Additionally, Foxe et al. [16] used an endogenous cueing task and found a different modulation of alpha rhythm at parietal-occipital areas. Furthermore, they also found that an alpha band desynchronization—or decrease—related to target information active processing possibly reflected a preparatory process [17], and that, during the valid visuospatial cueing task, an asymmetric, contralateral alpha deactivation in visual areas optimized orienting of visuospatial attention [18]. When a cue appeared prior of a target and the task required a greater attentional effort, Foxe and coauthors [16] also showed that alpha activity inhibited part of the visual space field in which interfering flankers were presented. Interestingly, alpha could be localized to the occipital-parietal areas of both hemispheres, and was active before the presentation of a stimulus, as if it had a role in the attentional preparatory mechanisms [18]. Indeed, while occipital-parietal areas integrate cue-related sensory information from multiple sensory modalities for programing the next engagemen<sup>t</sup> of visual attention, the inferior parietal cortex (IPC) would take care of maintaining attention orienting.

Experimental evidences have been provided showing task-related changes both in alpha power and scalp hemispheric lateralization or dominance. Indeed, a decrease in alpha power (i.e., alpha desynchronization) was found at occipital-parietal areas in attention orienting tasks, while an increase in this index was found during sustaining of attention [19]. Moreover, Li et al. [20] also found that task di fficulty and visual stimulus handling could change alpha band amplitudes in di fferent ways and at di fferent points in time during endogenous processing. The participants to the study had to decide whether an object was a car or a face; alpha activity showed a higher amplitude at the right-sided occipital-parietal regions when the decisional task was more di fficult. Volberg et al. [21] studied changes in alpha oscillatory activity in tasks of visual attention globally or locally directed by a cue. In this experiment, the subjects were prompted with a sound associated to the appearance of the target at a certain point of the screen, and with di fferent sounds associated to the presence of the stimulus at any point of the screen. They saw that, in a task demanding a certain speed of response and in which the target was presented at a certain point of the screen, the right occipital-parietal area was activated for processing of global information, and, consequently, the left homologous region had a greater presence of alpha band, being inhibited. Conversely, the left hemisphere became more active in response to a target stimulus fallen at a specific location and the processing of its features at local level, so that the right hemisphere was in this case inhibited. Interestingly, alpha band was prevalent before the presentation of the stimulus, as if it had a role in the attention preparatory mechanisms. Alpha rhythm power measured at frontal-parietal areas during top-down processing was functionally interpreted as if the brain dealt with participants' expectations towards the task. Most interestingly, alpha modulation in the posterior scalp areas is governed by the anterior, prefrontal areas, which have an important role in attentional selection: this might be a neural signature of executive control, in relation to which prefrontal and posterior occipital-parietal areas form an attention-related network. Wavelet-based EEG studies have also found an interaction between top-down and bottom-up processing hinting at the view that alpha activity preceding a stimulus would reflect a top-down preparatory mechanism modulating both the response timing and the performance in an attentional task [22].

In selective attention, alpha rhythm is higher on parietal-occipital sites. Furthermore, it has been proved that prefrontal cortex rules alpha power in posterior area during top-down processing. Wavelet Analysis studies also measured EEG oscillation bands in terms of spatio-temporal frequency. A desynchronization of alpha oscillations was related to a perceptual di fficulty in identifying a visual target in a discrimination task (target/mask). Larger pre-stimulus desynchronizations showed to be closely related to a better performance and to anticipatory attention [23].

As far as brain frequency-specific oscillatory activity in relation to cognitive functions is concerned, it is an acknowledged assumption that alpha power is a ffected by its oxygen supply. Indeed, from an inadequate blood oxygen delivery to the brain, it may in fact derive hypoxia, which is a common feature in many clinical disorders, including severe anemia, respiratory diseases (e.g., serious asthma and sleep apnea), and ischemic brain lesions, with or without a coma. This condition often also occurs in healthy individuals faced with extreme operational environments, such as high-altitude (HA) and/or low air pressure milieus (e.g., acknowledged alpinists and climbers as well as HA natives), often in relation with very low temperatures.

Altered EEG recordings, considered by clinicians as indicators of cerebral metabolism and useful tools for evaluating hypoxia severity—i.e., hypoxia with or without ischemia—in individuals a ffected by such a condition (e.g., [24–26]), have generally been reported in many electrophysiological studies based on di fferent types of both acute and chronic hypoxia. In human EEG studies of transient hypoxia, for instance, the latter was induced by low oxygen gas mixtures or hypoxic normobaric hypoxia (e.g., [27]), by simulated HA in hypobaric pressurized chambers (e.g., [28–31]), during rapid ascent to HA and lowland reoxygenation (e.g., [32–34]), and in HA natives [35]. Overall, a slowing activity of EEG in hypoxia with respect to normoxia has been generally advanced by all these studies.

As for the specific e ffects of acute hypoxia—regardless of the method of its induction—on spectral power density in the alpha frequency band, somehow inconsistent findings can be found in the literature. For example, 19 min of hypobaric hypoxia resulted in a significant decrease of alpha power (i.e., 8.9–11.8 Hz) in the spontaneous EEG of healthy volunteers as measured at the P4-O2 bipolar derivation [28]. Schellart and Reits [27] found that alpha content of brain spontaneous EEG during systemic hypoxia was strongly a ffected by the "eyes open/closed" volunteers' condition. In contrast to the eyes closed condition, a transient increase of alpha synchronization occurred during the eyes-open condition. An alpha (i.e., 10–11 Hz) spectral power selective desynchronization at a reduced air pressure of 3000 m, and a meaningful gradual desynchronization of this frequency with increasing altitude to 4000, 5000, and 6000 m, respectively, was found in the posterior scalp areas by Ozaki et al. [29], in volunteers who did not perform any active task during EEG recordings. Conversely, Papadelis and coauthors [30] showed an absolute significant increase of alpha power during acute hypoxia as induced by decreasing the pressure of a barometric chamber, compared to 100% oxygen condition, in right-handed volunteers who performed a simple computerized flight-simulation task in which participants had to keep a continuously moving aircraft-target as much closer to the center of a personal computer (PC) screen denoted by a rectangle, in order to avoid the drifting of the target towards the edges of the screen if no control was applied [30]. Additionally, a significant increase of alpha synchronization at posterior scalp sites after an HA chronic acclimatization of 30 days, but not after a sojourn of seven days only, with respect to previous HA EEG baseline, together with a cogen<sup>t</sup> desynchronization after re-descending to lowland, have instead been described by Zhao and coworkers [34], in the spontaneous EEG of a group of young soldiers who did not perform any tasks during the recording sessions.

Inasmuch, both alpha oscillatory synchronization or desynchronization have been associated with attention, attention-orienting inhibition, long-range synchronization, memory performance, and inhibition of interfering visual memories (e.g., [36,37]); these inconsistencies represent a nuisance for a more specific definition of both the functional significance of this EEG oscillatory activity in relation to attentional functions and, most importantly, of their possible alterations with hypoxia.

However, because attention is not a unitary function, it is important to relate EEG alpha oscillatory activity to the three specialized networks sub-serving the human attentional system as originally conceptualized by Posner and Petersen [38]. After the reviewing of many neuroanatomical and neuropsychological studies, in fact, Posner's group [39], advanced the view that human attention system is undergirded by three di fferent networks regulating di fferent sub-processes, which is, alerting, orienting, and executive control. The alerting system would take care of achieving and maintaining an alert state; it would be focalized in frontal and parietal regions of the right hemisphere (RH). The orienting system, driving attentional focus on a specific point of space, would be localized in the frontal-parietal areas. Executive control, allowing to solve conflicting cognitive situations and psychomotor responses, would be associated with medial frontal regions and lateral prefrontal cortex.

To analyze the functional activation and interdependence/independence of these networks, Fan et al. [40] devised a so called Attention Network Test (ANT) based on the presentation of four di fferent cueing conditions, that is, NC (no cue), CC (central cue), 2C (two vertically lateralized spatial cues at the same time above and below the fixation point), and LC (one vertically lateralized spatial cue above or below the fixation point). Each of these cueing conditions was randomly followed by directionally congruen<sup>t</sup> and incongruent strings of five arrows, of which the central one, which might point either toward the left or to the right, posed as a target, and the peripheral ones as low-level (congruent condition) or high-level interfering (incongruent condition) flankers. A third target condition made up of a central arrow-target pointing either leftward or rightward surrounded by four simple straight-lines was originally included. Due to some inessential experimental findings and to the theoretical interpretation of the latter in several behavioral studies, Posner's group later hemodynamic (i.e., functional magnetic resonance imaging (fMRI) [41]) and electrophysiological imaging (i.e., EEG [42]) studies omitted the 2C cue type and the target-arrow-surrounded by straight-line flankers condition, and used three cueing types, namely the NC, CC, and LC conditions, and two target-arrow vs. flanker-arrows conditions only.

#### *1.1. ANT and Functional Meaning of its Cueing Conditions*

The ANT has shown to be a useful and simple measure of attentional e fficiency so that it can be used with adults, children, monkeys and patients with various abnormalities of attention [40]. As the cueing conditions are concerned, both neutral (CC) and valid spatial cues (LC) serve as a form of alerting cue, but only the latter provides always valid, predictive spatial information that allows participants to begin orienting attention to the appropriate location before the target is presented. Conversely, the central cue (CC) tends to keep reflexive attention focused on one location, which is the central location at the fixation cross, before the target exogenously attract orienting of attention at its appearance location. Unlike for these cueing conditions, under NC, there are neither alerting nor spatial cues and participants see only the fixation cross and the sequence of randomly presented targets above or below it. According to Fan et al. [40], the NC condition is a relatively low-alertness tonic orienting condition in which attention tends to remain di ffused across the two potential target locations.

Since the three sources of attention, alerting, orienting, and executive attention appear to engage separate brain mechanisms, Fan et al. [40] investigated the independence and/or interaction of these networks and concluded that the latter were relatively independent mechanisms. Indeed, while phasic alerting enhanced flanker interference costs, both LC and NC conditions reduced them. According to Fan et al. [40], this occurred, in the first case, because participants directed attention to the target ahead of time, and, in the second case, because the NC condition is a relatively low-alertness condition and resulted in longer RTs and relatively lower errors. Consistent behavioral and electrophysiological findings were more recently found by Zani and Proverbio [43], in that both RTs and mean latency measures of a so-called ERPs conflict negativity (CN) component showed a decreased flanker conflict interference for both NC and LC cueing conditions as compared to both CC and 2C conditions. Overall, these findings support the view that, far from being independent from each other, the attentional system networks somehow directly interact one another.

#### *1.2. Rationale for the Present Study*

Purpose of the present study was to identify the most prominent features of EEG alpha power in the classification of attention modes and response control as related to the ANT neural networks with the hope of providing alpha-based reliable markers of their functionality. Moreover, we wanted to assess the possible impact of respiratory hypoxia on these three attention networks because not ye<sup>t</sup> investigated and renown up to now. More specifically, we wanted to tap possible di fferences in this impact between conditions of lower or greater motor and cognitive workload. To achieve these aims we used the afore-mentioned ANT-SR (ANT-Slightly Redesigned) version and measured the decrease and/or increase of alpha power synchronization as a function of the possibly independent and/or partly interdependent activation of the three attention networks. We wondered whether hypoxia would have affected the three attentional networks independently from one another or in interaction.

Considered that ANT paradigm is an extremely simple task so that "The instructions to the subjects only require they know how to press a left key for a leftward-pointing arrow and a right key for a rightward-pointing arrow [40]", we slightly redesigned it with the goal to carry out a behavioral and electrophysiological study aimed to find possible di fferences in the e ffects of hypoxia on a simple with respect to a di fficult, psycho-motor overloaded attention orienting task. In details, we devised one cue-target condition in which a valid, spatially informative local cue preceded the delivery of one out of two di fferent types of arrow-strings which had to be discriminated to provide a dual motor-choice overt response with the index or medium fingers of the left or right hand, depending on the type and direction of the arrowhead, no matter the congruency of the flankers. In this di fficult motor variant, defined by us "LCmot", the participants had to use one out of four fingers to give their response. The goal was to investigate the mechanisms of visuospatial attention, and, more specifically, of attentional conflict control situations during the execution of a relatively di fficult motor task. In order to compare the e ffects of hypoxia on this condition with those possibly occurring in a simple orienting of attention condition, besides on alerting and executive conditions, we added to our ANT-SR the no Cue (NC), the Central Cue (CC), and the Local Cue (LC) conditions used in the original version of the ANT paradigm by Fan and colleagues [40], and in Fan's et al. [41,42] later studies. However, to make the four cue conditions comparable from the stimulus-information point of view, the two types of arrows were also presented in the three afore-mentioned conditions despite the fact that the volunteers had not to discriminate between their types and gave an overt motor response using alternatively only the index fingers of the two hands depending on the target arrowhead orientation.

The ANT-SR [40] was used to investigate the effects of hypoxia on endogenous and exogenous visuospatial attention orienting modes. Capitalizing on the different set of alerting and attention-orienting cue-target conditions of the redesigned ANT-SR test [40,44], and assuming that an increase of alpha synchronization represented an inhibition of attention orienting, we hypothesized that:


#### **2. Materials and Methods**

#### *2.1. Study Compliance with Ethical Standards and Participants*

The study was approved by the ethics committee of the Italian National Research Council (CNR) and was conducted in the Electro-Functional Brain Imaging unit (EFBIu) of the CNR-IBFM Institute in accordance with American Psychological Association (APA) ethical standards for the treatment of human experimental participants (APA, Monitor Staff, 2003, vol. 34, n. 1). Furthermore, the experiments were conducted with the understanding and the written consent of each participant in compliance with the indications of the 2018 Declaration of Helsinki ethical principles for medical research involving human subjects by the World Medical Association (WMA Declaration of Helsinki, 9 July, 2018, PDF file).

Ten (10) healthy volunteers (4 females and 6 males), with age ranging from 19 to 27 years (Mean = 24, SD = 2.7) were selected for participating in randomized order in two EEG recording sessions during which they breathed either ambient air (also indicated as "normoxia") or a 12.5% O2-impoverished air mixture (also indicated as "hypoxia"). Besides suffering or having suffered of any neurological and psychological syndromes or the intake of any psychopharmacological substances, criteria of exclusion from the study were cigarettes smoking, arterial hypertension, and cardiovascular or respiratory diseases. In addition, to minimize confounding effects, no participant had to have sojourned at a higher altitude than 300–400 m in the 4 weeks preceding the study nor had to have been regularly and intensively engaged in any physical training program. Again, volunteers were required to refrain from any strenuous physical activity and from unlimited consumption of alcohol, caffeine, and theophylline containing beverages in the 24 h prior to the experimental sessions of the study. All selected participants had normal or corrected–to–normal vision and right-eye as well as right-hand dominance and none of them had any left-handed relatives as assessed by the Edinburgh Inventory. Unfortunately, the data of two volunteers had to be excluded from the statistical analyses either because of excessive eye—and body—movement artifacts during EEG recordings or for not completing the two recording sessions cycle.
