**4. Results**

#### *4.1. Behavioral Results*

## 4.1.1. Error-Percentage Rates

The ANOVA carried out on motor response errors proved the significance of the main "respiratory condition" factor (F (1,7) = 7.276: *p* < 0.025)), indicating that in air participants showed a 2.51% of errors, while during hypoxia this rate increased on average to 6.12%. In addition, error percentage-rates for the various cueing conditions significantly changed as a function of the respiratory condition (F (2.63, 18.45) = 31.25; ε = 0.879; adjusted *p* value < 0.000024; η2*p* = 0.61). Post-hoc comparisons indicated that, in air, error rates for NC were higher than for both CC (*p* < 0.0001) and LC (*p* < 0.0001), but lower than for LCmot (*p* < 0.0002). In turn, CC did not differ from LC, but it showed a lower errors rate than LCmot (*p* < 0.0001). Again, LC obtained a lower errors rate than LCmot (*p* < 0.0001; see Figure 3 for these findings.)

In hypoxia, instead, the post-hoc contrasts indicated that error percentage rates progressively and significantly increased from NC to LCmot cueing conditions: NC vs. CC, *p* < 0.0002; NC vs. LC, *p* < 0.0001; NC vs. LCmot, *p* < 0.0001, respectively. Moreover, CC was also different from LC (*p* < 0.0002) and from LCmot (*p* < 0.00001), and, in turn, LC was different from LCmot (*p* < 0.0001). Due to the aforementioned increases, with the exception of NC, the CC, LC and LCmot cueing conditions showed significantly higher error rates in hypoxia than in air (i.e., *p* < 0.001 for CC; *p* < 0.0001 for LC; *p* < 0.000025 for LCmot; see Figure 3 again).

**Figure 3.** Mean percentages (along with standard errors (SE)) of errors committed by participants as a function of cueing condition and respiratory condition regardless of the target strings congruency.

#### 4.1.2. Reaction Times (RTs)

The ANOVA carried out on mean RTs showed that the measures obtained for the two levels of the respiratory factor (i.e., Air: Mean = 458.60 ms, standard error (SE) = 13.68; Hypoxia: Mean = 474.21 ms, SE = 15.01) were significantly different (F (1,7) = 4.799; ε = 1; *p* < 0.05, η2*p* = 0.51). The ANOVA also yielded significant effects of cue-type (F (2, 16; 15, 15) = 176.96; ε = 0.72; adjusted *p* value < 0.00001, η2*p* = 0.96) and target flanker-type (F (1, 7) = 167.23; *p* < 0.000004) and a significant interaction between the two (F (2,58; 18,08) = 39.96; ε = 0.86; adjusted *p* value < 0.000002, η2*p* = 0.85).

Post-hoc comparisons showed that, except for the LCmot cueing type, response times were much faster for all the other cueing conditions when target-strings were congruen<sup>t</sup> than incongruent (i.e., NC (*p* < 0.00002), CC (*p* < 0.0002) and LC (*p* < 0.0002), respectively; see Figure 4). Post-hoc contrasts also indicated that RTs were slower for NC than for both CC (Cong: *p* < 0.0002; Incong: *p* < 0.0002) and LC (Cong: *p* < 0.0002; Incong: *p* < 0.0002), and in turn for CC than LC for both target flanker types (Cong: *p* < 0.0002; Incong: *p* < 0.0002). Further contrast analyses proved that LCmot obtained much slower RTs than the other three cueing conditions, namely NC, CC, and LC, for both target congruency conditions (Cong: LCmot vs. NC, *p* < 0.0002; LCmot vs. CC, *p* < 0.0002; LCmot vs LC, *p* < 0.0002; Incong: LCmot vs. NC, *p* < 0002; LCmot vs. CC, *p* < 0.0002; LCmot vs. LC, *p* < 0.0002; see Figure 4 again).

**Figure 4.** Mean RT and Standard Error (SE) values obtained as a function of the cueing condition and target congruency over the sample of participants. Note that RTs were measured starting from the target delivery time, that is 500 ms later than the cue omission or administration according to the cueing condition.
