*2.3. Procedure*

The selected participants took part in randomized order in two 4 h lasting EEG recording sessions, one week apart from each other. During these two experimental sessions they breathed either ambient air (or normoxia) or a 12.5% O2-impoverished air mixture (simulating respiration conditions at an high altitude of ~4200 m (~13,780 feet) at sea level), which may be assimilated to an acute bout of tolerable and far from pathogenic normobaric hypoxia while performing in four different cue-target visuo-spatial attention conditions taken and readapted from Fan's et al. [40] and Posner's [44] ANT.

Independent of the respiratory session with which the volunteers started their participation in the study, while being prepared for the EEG recording during their first report to the lab, they underwent a general psychophysiological screening that included the chronicling of their chronobiological daily habitual activity patterns (Morningness Eveningness Questionnaire—MEQ [45]). In addition, we also measured their subjective mood and alert levels by means of so-called paper-and-pencil psychophysiological Likert-like scales and three minutes speeded-up letter-cancellation tasks. As for the hypoxia session, just starting from their arrival to the laboratory the volunteers breathed a normobaric hypoxic mixture obtained removing a controlled amount of oxygen from air (i.e., 12.5% O2) by means of a MAG-10 hypoxicator apparatus (Higher Peak LLC, Winchester, MA, USA). The mixture was delivered through a facemask at 30 L/min. Excess air flow was diverted outside the mask to prevent inspired oxygen pressure from increasing above 90 Torr (see Figure 2 for a visual exemplification).

**Figure 2.** A picture of a participant wearing the mask connected to the hypoxicator apparatus which provided air impoverished of a 12.5% of oxygen during the electroencephalogram (EEG) recording session.

Since there are influential data in the physiological literature indicating that in humans the most relevant cardiorespiratory and blood plasmatic e ffects were found in between 2–4 h of serious hypoxic respiration (see, e.g., [46]), all our volunteers were systematically submitted to the EEG recording session starting after 2 h of such a respiratory condition, which was spent in the application and jellying of a 128 electrodes electrocap as well as in the filling of the mood and alert paper-and-pencil scales, besides in an alert letter-cancellation-based task. For experimental standardization and session comparison-sake, the volunteers started their tasks-related EEG recordings after 2 h during the normoxic session too, a span of time overall spent in the activities mentioned above besides the electrocap application.

At the end of the preparation time-span, participants were invited to take a seat in a comfortable easy chair with a high backrest within an electrically and magnetically shielded cubicle (Faraday cage) in front of a CRT screen with a small white fixation cross (FC) in the center of its black background placed at a distance of 114 cm (or 3.34 feet) from them. During the EEG recordings, sequences of the white arrow-strings were randomly presented above or below the fixation-cross at the center of the visual screen. The central target-arrow of each string fell just in correspondence of the fixation cross with a vertical eccentricity of ±1.25 degrees of visual angle from the latter. Apart for the NC condition, an asterisk-cue always preceded the arrow-strings in the CC, LC, and LC mot conditions.

To familiarize participants with the task to be performed in each cue-target condition, before recording was started volunteers read detailed written instructions and performed a practice run of 30 stimulus pairs. For each of the four cue-target condition, there were four separate blocks of trials, each containing 128 trials grouped in a di fferently randomized order and lasting approximately 3.5 min. To avoid any confounding and systematic interacting e ffects of practice, fatigue and hypoxia, we also randomized the order of administration of the respiratory conditions, of the cue-target conditions, of the type of target-arrows, of the falling of the target in the upper or lower visual field, and of blocks presentation across participants.

#### *2.4. EEG Recording and Analysis*

For EEG recordings we used scalp electrodes mounted in an ANT elastic Waveguard 128-electrodes electrocap. The electrodes were densely spaced all over the frontal, central, temporal, parietal, and occipital scalp-sites as proposed by the 5% system [47] devised for high-spatial resolution EEG/ERP recordings. Two electrodes placed below and above the right eye recorded vertical eye movements, whereas two further electrodes placed at the outer canthi of the eyes recorded horizontal eye movements. Linked ears served as the reference lead, whereas a frontal electrode served as a ground site. Electrode impedance was below 5 k Ω. Both EEG and electrooculogram (EOG) continuous signals were acquired using directional-current (DC) amplifiers and a digitization rate of 512 samples/s. O ffline, automated rejection of electrical artifacts was performed before EEG averaging to discard epochs in which eye movements, blinks, or excessive body muscle potentials occurred. The artifact rejection criterion was a peak–to–peak amplitude exceeding ± 70 μV for EEG signal or ± 100 μV for EOG signal, and, in line with the trials stimulus events timing, went from 100 ms before the cue-type to 1500 ms after it for each trial in sequence, until the routine detected EEG values falling within the indicated window-values. Overall, the rejection rate was ~6.0 %. Although we used the indicated rejection method implemented in our EEG signals analysis applications package, we are aware that other independent component analyses would have quite e fficiently worked [48,49]. Trials associated with missing (i.e., overt motor responses falling after 1000 ms from the target) were also discarded, which resulted in a rejection rate of less the 2%. Overall, then, a global 92% out of the 100% trials administered were accepted for averaging. Trials associated with motor-response errors (i.e., on the one hand, responses given with the wrong hand with respect to the left/right target-arrow pointing directions, no matter the arrow types, for the NC, CC, and LCmot conditions, and, on the other hand, the wrong finger with respect to the standard/star arrow types, besides with the wrong hand, for LCmot) were also computed separately as a function of respiratory conditions and discarded from EEG averaging. So, after rejection procedures and the due mean computations over the sample of participants, the mean percentages of correct trials averaged as a function of NC, CC, LC, and LCmot conditions in normoxia amounted to 98%, 99%, 99%, and 94% of the 100% of total trials administered, respectively, and to 98%, 95%, 94%, and 89%, respectively, in hypoxia. More specifically, in normoxia the percentages of congruen<sup>t</sup> vs. incongruent trials averaged for the former congruency level of NC, CC, LC, and LCmot conditions were 50.41%, 49.79%, 50.69%, and 47.29% of the 100% total trials administered, respectively, and 48.35%, 48.83%, 49.15%, and 46.65%, respectively, for the latter congruency level. In hypoxia the percentages of congruen<sup>t</sup> vs. incongruent trials averaged for the congruen<sup>t</sup> target trials of the four cueing conditions were 49.90%, 48.61% 47.50%, and 45.54% of the 100% total trials administered, respectively, and 48.32%, 47.23%, 46.70%, and 43.42%, respectively, for the incongruent target trials.

Before selective averaging, accepted EEG signal chunks were digitally filtered with a half–amplitude band-pass of 0.016–70 Hz. As for the rejected epochs, accepted EEG epochs in the averages were synchronized with the onset of cue presentation (CC, LC, and LCmot conditions) or omission (NC condition) and went from 100 ms before the cue-type to 1500 ms after it, the target-related EEG being elicited 500 ms after cue-onset. In order to compare the averaged EEG epochs relative to the single-choice motor-workload conditions (i.e., NC, CC, and LC) with those generated in the double-choice motor-workload condition (LCmot), EEG responses to the di fferent types of arrow-targets (i.e., standard and star arrows) were collapsed together in all cueing conditions. Furthermore, in order to increase the signal-to-noise ratio EEG trials related to targets delivered in the upper and lower visuospatial hemifields were also collapsed together in all cueing conditions.

Hence, for each subject distinct EEG average waveforms were obtained according to respiratory condition (i.e., ambient-air or hypoxia), cueing-task condition (CC, LC, NC, LCmot), and target-congruency (i.e., congruent, incongruent) condition. Besides average EEG waveforms for each single participant, grand-average EEG signals were also computed for the participants' sample.

#### *2.5. Alpha Power Analysis*

To obtain the alpha power as a function of the di fferent experimental conditions, the Fast Fourier Transform (FFT; [50]) computation was used. The reference point for the FFT analysis or Start Time was the time of cue presentation, at 0 ms latency, while the End Time was the 1500 ms latency. The time span of the EEG waveforms was divided in three (i.e., 3) blocks, each with a length of 500 ms, with a sample count which had the power of two (i.e., 512 Hz). Alpha oscillations content and, more specifically, alpha amplitude in the range of 7.5–12.5 Hz was computed and averaged over the three blocks. To allow a better comparison of FFT results over experimental conditions, a normalization of the computed alpha power in the aforementioned band with respect to the summed power over all frequency values of the channels with the highest power was carried out, and, as a whole, reported in μV2. Additionally, the baselines of the spectral analyses were corrected across conditions before FFT computation. These procedures were automatically applied to each site of the electrodes montage so to obtain a topographic distribution of the alpha power spectra over the scalp as a function of the experimental variables. Topographical voltage maps of alpha were obtained by plotting color-coded alpha power values derived by interpolating frequency content values between scalp electrodes as a function of respiratory and cueing conditions. Eye-balls inspection of the maps indicated that EEG alpha power reached higher amplitude values at posterior than anterior scalp regions regardless of respiratory and attention cueing conditions, and that the topographic distribution of alpha amplitude at posterior scalp regions changed as a function of experimental conditions considered. In order to test the statistical significance of these apparent data changes, taking into account previous EEG findings in the literature, alpha power amplitude was measured at four posteriorly-anteriorly distributed couples of homologous electrode sites: O1 and O2, mesial-occipital electrodes; PO7 and PO8, lateral parietal-occipital electrodes; TPP7h and TPP8h, temporo-parietal-parietal electrodes; and F5 and F6, pre-frontal lateralized electrodes.
