**5. Discussion**

The present study had the manifold aims of investigating the role of alpha synchronization and/or desynchronization in visuospatial attention orienting, and, more specifically, in the exogenous and endogenous modes of attention-orienting. Additionally, we aimed to investigate the relationships of alpha synchronization and desynchronization with the functional activation of brain alerting, orienting, and executive attentional neural networks. We wished to assess whether the separate attentional networks might have been independently or interactively affected by a reduced, non-pathogenic brain oxygenation state—or hypoxia—and, in case of interactions, which of them would have interacted. Last but not the least, we inquired into possible influences of induced hypoxia on overt motor performance.

From the behavioral point of view, our findings indicated that both error rates and RTs showed to be affected by hypoxia, in that, during this respiratory condition, the participants showed a tout court lower performance accuracy and speed than in air—or normoxia—regardless of cueing mode and target/flanker congruency.

However, error rate differences between cueing conditions also showed to be strongly affected by the respiratory factor regardless of target/flanker congruency. In normoxia, in fact, the lowest error rate was observed for both the spatially cued (LC) and the alerting but spatially unpredictable, centrally cued (CC) targets, without any difference between them (see Figure 3). A higher errors rate was obtained for uncued targets (NC), and a highest rate for spatially cued, motor-choice overloaded

(LCmot) targets. Conversely, with the exception of uncued targets (NC), hypoxia markedly a ffected performance accuracy in that the latter progressively decreased—i.e., participants committed a greater number of errors—as a function of both the alerting but spatially unpredictable, centrally cued (CC) and spatially cued targets, regardless of their demand of a single or a double-choice motor response (i.e., LC and LCmot). All in all, these overt motor-accuracy findings suggested that hypoxia a ffected the original ANT-related attention alerting, orienting, and conflict networks independently from one another. Moreover, our data suggested that hypoxia also strongly a ffected the motor-charged attention orienting cueing mode.

Unlike for error rates, overt responses (i.e., RTs) showed a distinct pattern of significant changes as a function of cueing conditions in close interaction with target/flanker conflict level, regardless of the respiratory condition. Shortest RTs were observed for validly spatially cued (LC) targets, intermediate RTs for the spatially unpredictable, centrally cued (CC) targets, longer RTs for uncued targets (NC) and longest RTs for spatially cued, motor double-choice demanding (LCmot) targets (see Figure 4). The relative benefits of original ANT alerting and orienting [40] amounted to about 25 ms and 50 ms, respectively, while for the motor double-choice-related orienting condition of our ANT-SR, a heavy cost of about 200 ms, rather than a benefit, was found. Response conflict due to target/flankers incongruency led to an increase of overt motor response latency with a greater mean RTs cost of about ~40 ms for the spatially unpredictable, centrally cued (CC) targets than for both the uncued (NC; ~30 ms) and the validly spatially cued targets (LC; ~20ms), and no cost for LCmot cueing condition.

At least for what concerned the behavioral findings in response to the original ANT-related cueing conditions in normoxia, the aforementioned pattern of results is in good accordance with those that were published originally by Fan et al. [40], Neuhaus et al. [55], and by our group (e.g., Zani and Proverbio [43]), in that the use of either a spatially informative cue (LC) or of a no-cueing condition (NC) reduced costs for incongruent vs. congruen<sup>t</sup> targets processing as compared to both the 2C and CC cueing modes. 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 NC is a relatively low-alertness condition and resulted in longer RTs and relatively lower errors. In Fan's et al. [40] words, "it is possible that the longer time to produce a response due to low alertness can provide additional time for executive attention processes in the conflict condition, thus reducing executive costs." Following the above-mentioned lines of reasoning, it would seem highly plausible that these same neural mechanisms could be adopted for explaining the lack of any di fferences between the speed of motor response to the congruen<sup>t</sup> and incongruent target/flanker patterns for the motor-charged, LCmot condition (see Figure 4 again).

Generally speaking, we believe that the consistency of our behavioral findings with those obtained by the studies quoted above strongly supports the views that, despite the relatively exiguous number of participants analyzed in the present study, the obtained data may be considered sound and reliable, besides being truly dependent on the manipulation of experimental variables.

As far as the electrophysiological data are concerned, our findings showed to be quite in accordance with those obtained for behavioral data. Indeed, alpha measures separately changed as a function of cueing condition and respiratory condition factors, either in interaction with the hemisphere and/or the hemisphere and electrode factors. Unlike behavioral data, however, electrophysiological data were not apparently a ffected by the executive neural processing, neither per se or in interaction with other factors.

It would seem rather plausible that this slight inconsistency between overt motor responses and covert alpha processing data depended on the fact that the former were timely closer to the executive conflict resolution processes than the latter because alpha power computed by FFT starting from the presentation of cue type, at time 0, up until 1500 ms after cue, with the target falling at 500 ms, included, according to the experimental conditions, time spans of either alpha synchrony or desynchrony occurring after the emission of the overt motor response. Due to this, it is somehow conceivable that alpha power processing occurring during these time spans may have been less closely related to the executive conflict resolution processes.

As for the specific trend of our electrophysiological data, it is worth of note that, notwithstanding the small size of our sample of participants, our findings are in line with those of previous studies, thus confirming the view that a bout of increased alpha amplitude or synchronization recorded at posterior occipito-parietal areas during a visuospatial attention task reflects an inhibition of attention orienting towards information presented at an irrelevant point in space [56]. Indeed, brain areas involved in the processing of task-irrelevant space may be actively inhibited during an increase of alpha-oscillations synchronization to refrain from processing of distracting information [19,57]. More in details, the data showed that alpha power changed not only as a function of endogenous and exogenous attention orienting modes, but also of the di fferent informative content transmitted by the various cue-target combinations within the former and the latter modes of attention-orienting, regardless of brain oxygenation condition. They also indicated that attention-related alpha frequency desynchronization is more prominent in the LH than in the RH, thus suggesting asymmetries that might be regarded as a sign of cerebral lateralization or hemispheric dominance for this neurocognitive function.

In agreemen<sup>t</sup> with these viewpoints, alpha showed the lowest power (i.e., stronger desynchronization) during the no cue (NC) mode, in which, most likely, brain tonic alerting and utterly exogenous orienting of attention responses to targets sequentially presented in random order at one out of the two relevant space locations without being primed by any cue, took place (see EEG alpha-power maps drawn in Figure 5a, and Figure 6 for mean power values for this condition). At the opposite, alpha showed the highest synchronization in the ANT condition in which a spatially uninformative, but alerting cue was delivered to the center of the screen, followed later on by the presentation of a target at one out of the two spatially relevant locations (i.e., CC condition; see EEG alpha-power maps drawn in Figure 5b, and Figure 6 for mean power values for this condition). Very likely, this finding may have arisen because, in this condition, a phasic alerting together with a transient suppression of attention-orienting might have been elicited by the warning CC, followed by a reflexive orienting of attention to the stimulated space location as triggered by the target delivery.

As for the endogenous attention-orienting mode conditions, the alpha power recorded for LC task, where the cue was both alerting and validly informative of the spatial point at which the target would follow, showed to be somewhat lower than that elicited by CC condition over the LH, but not the RH, and, in general, pretty higher than that computed by FFT for the NC mode over both hemispheres. It seems highly plausible that alpha desynchronization found over the left hemisphere in response to LC condition may have undergirded a fast orienting of attention to the location indicated by the valid cue. A possibility strongly supported by Posner's theses [44] who considered the left lateralized processes as more likely involved with higher temporal (phasic) mechanisms, such as orienting of attention, and right lateralized processes often more involved with slower (tonic) alerting e ffects. Consistent with these views, LC showed a same alpha power as CC over the RH, but a significant higher desynchronization (i.e., a significant lower alpha power) over the LH (see EEG alpha-power maps drawn in Figure 5c, and Figure 6 for mean power values for this condition). Interestingly, a right sided asymmetry in alpha scalp distribution has been shown by several previous studies [21,58].

Most interestingly, the data also indicated that introducing a neurocognitive perceptual (e.g., a target-type discrimination) and psychomotor overload (e.g., a double-choice motor task) as in the LCmot condition further increased alpha desynchronization over the LH, but not the RH. This left-sided increase in alpha desynchronization in LCmot with respect to both CC and LC showed to be of such a degree that the alpha for this cueing mode did show any di fferences from that for the exogenous NC task over the LH as compared to the RH. Conversely, LCmot elicited a same alpha power level as LC over the RH. As a result of the afore-mentioned changes, a prominent right-sided hemispheric asymmetry of the alpha power occurred for this motor-charged cueing-mode only (see EEG alpha-power maps drawn in Figure 5d, and Figure 6 for mean power values for this condition).

As for the functional meaning of a left-sided asymmetry of alpha desynchronization and a right-sided asymmetry of alpha synchronization found for LCmot cueing condition, it might be possible that, in line with Posner's proposals [44], during this high-motor workload task the LH may have been engaged in the orienting of attention to the location indicated by the valid cue, as reflected by an increase of alpha desynchronization, but the demands of the target-type discrimination and of a decision relative to the double-choice overt response may have required to sustain alerting and/or attention over time, thus recruiting the slower, tonic processing by the RH, reflected at the scalp by an increase of alpha synchronization.

As far as hypoxia is concerned, our alpha measures indicated that this deficient-oxygenation condition mostly increased alpha power at right-sided brain posterior districts, and more in details striate and extrastriate occipital areas as well as parietal areas (Figure 7). However, some of our findings also indicated that hypoxia selectively enhanced alpha power at the left-sided anterior prefrontal areas (Figure 8). As for the effects of this respiratory condition on brain attentional processes, independent of brain areas involved, our results showed increases of alpha synchronization no matter the endogenous and exogenous nature of visuospatial attention orienting. Moreover, they also hinted at the view that hypoxia had diffused influences on all attentional processing networks (i.e., alerting, orienting and executive control networks), in that, no matter the network activated, hypoxia resulted in a general increase of alpha synchronization.

At least for the increase of alpha power in response to hypoxic hypoxia obtained at posterior occipital-parietal areas, our discoveries fully mesh with the findings reported by some previous studies (i.e., [27,30]). For truth sake, however, it must be said that these discoveries are only partially consistent with the results of another relatively recent study [32], and clearly in contrast with the results of some dated investigations (e.g., [28,29]). We believe that rather than to methodological and procedural differences in (1) the induction methods of hypoxia in neurologically healthy volunteers, (2) the different alpha frequency band ranges taken into account, and (3) the number of scalp sites at which the alpha power was computed, the different results obtained by the present investigation as compared to several previous studies (e.g., [28,29,34]) may be closely due to the participants' neurocognitive functional state during EEG recordings. Indeed, while the latter studies analyzed the alpha power in the spontaneous EEG of volunteers who were not submitted to any active psycho-motor or attentional tasks, the investigations showing fully consistent findings with our own (i.e., [27,30]) measured alpha power in EEG recorded either during a presumably more environmentally-aware or attentional-driven condition—though not directly analyzed or accounted for—such as the eyes-open state, compared to the eyes-closed one [27] or during a simple, though somehow effortful, computerized visuospatial attentional task in which participants had to keep a moving target as much closer to the center of a screen denoted by a rectangle, in order to avoid the drifting of the target towards the edges of the screen [30].

Despite the lack of any report of the participants' performance scores by these authors [30], it may be assumed that the aforementioned task might have been perceived as easier or less effortful in normoxia than in hypoxia, and that the generalized increases in alpha synchrony during the latter condition might have represented an alteration or, up-to a certain degree, a suppression of brain attentional orienting capacity. In support of this view, Zhao et al. [34] found changes of EEG during hypoxia/reoxygenation confined in posterior right cortices, with increased alpha synchrony related to hypoxia and increased beta desynchrony related to reoxygenation. Further support derived from our current data, which showed increased error rates and motor response times as well as of alpha power in hypoxia with respect to normoxia.

#### **6. General Conclusions**

All in all, the modulation of alpha power induced by the cueing conditions used in our study lends reliable and robust support to the view advanced by several previous studies e.g., [19,56,57] that the desynchronization versus the synchronization of this oscillatory-band reflects an active orienting versus an inhibition of visuospatial attention, respectively. Following this line of reasoning, the generalized right-sided alpha band synchronization observed in hypoxia as compared to normoxia would reflect the e ffort to sustain attention-alerting over time, as mediated by the RH, in order to cope with a transient impairment or, up to a certain degree, suppression of brain attention-orienting capacity, mediated by the LH. The lack of any e ffects of the di fferent attentional cueing and executive conflict conditions on alpha synchronization induced by hypoxia may also indicate that this enhancement possibly reflected an overall transient alteration or tout-court suppression of attention alerting, orienting, and executive networks functionality independently from one another.

#### **7. Study Limits and Wishes**

A possible limitation of the present study was the relatively small sample size (*n* = 8.) The latter, in fact, was somewhat small and therefore conceivably statistically underpowered. However, the consistency of our general behavioral and electrophysiological findings with those of previous studies in the literature lends strong support to their soundness as also championed by partial eta squared values of computed size e ffects here provided. Furthermore, the sample size was large enough to assess the expected pattern of attentional e ffects predicted by available literature. Another edge for our data might be our use of parametric statistical tests which require the a-priori hypothesis of variables with normal distribution, here somehow not assured because of the lack of linearity of FFT values and of the small sample size. Nevertheless, because of the potential relevance of our findings, we think it is worth pursuing further research aimed at their sound replication by enlarging the sample size and the use of non-parametric statistical tests.

**Author Contributions:** A.Z. theoretically devised the study. A.Z., A.M.P., and C.T. collected the data. C.T. carried out computations of alpha power by means of the FFT for the single subjects' and the sample's grand-average EEG. A.Z., and A.M.P. wrote the manuscript. A.Z. and A.M.P. critically processed the final version of the latter and the theoretical interpretation and discussion of the experimental findings of the study. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by CNR grants to A.Z. that were used for purchasing EEG recording supplies and materials required for data collection and analysis.

**Conflicts of Interest:** The authors declare no conflict of interest.
