*3.2. Electrophysiological Data*

Grand averages of the ERPs in the three CPT conditions are presented in Figure 1. The respective descriptive statistics for the P3 amplitudes and latencies are given in Table 3. To examine differences between P3 latencies in the three CPT conditions for significance, a one-way ANOVA was computed, with P3 latencies in the three CPT conditions as three levels of a repeated-measures factor. Due to a violation of sphericity, the Greenhouse–Geisser correction was used with *ε* = 0.740. The main effect yielded statistical significance, *F*(1.480, 161.269) = 122.794, *p* < 0.001, *<sup>η</sup>p*<sup>2</sup> = 0.530. Planned comparisons revealed significantly shorter latencies in the CPT0 than in the CPT1 condition, *t*(109) = 10.385, *p* < 0.001, *d* = 0.990, as well as shorter latencies in the CPT1 than in the CPT2 condition, *t*(109) = 5.493, *p* < 0.001, *d* = 0.524.

**Figure 1.** Grand average waves for the event-related potentials (ERPs) in the three CPT task conditions at the PZ electrode site. The zero point of the time scale refers to the onset. Negative is plotted upwards.

**Table 3.** Descriptive statistics of P3 latencies and P3 amplitudes in the three CPT conditions in 110 participants.


M = Means, SD = standard deviation, Min = minimum, Max = maximum.

The same analysis was computed for the P3 amplitude as a dependent variable to probe differences between P3 amplitude in the three CPT conditions. Again, sphericity was violated, so that the Greenhouse–Geisser adjustment was used with *ε* = 0.927. The main effect on the P3 amplitude was significant, *F*(1.854,202.091) = 185.224, *p* < 0.001, *<sup>η</sup>p*<sup>2</sup> = 0.630. Planned comparisons revealed that P3 amplitude was significantly larger in the CPT1 than in the CPT0 condition, *t*(109) = 14.741, *p* < 0.001, *d* = 1.405, as well as in the CPT2 than in the CPT1 condition, *t*(109) = 3.786, *p* < 0.001, *d* = 0.361. Thus, P3 amplitudes and latencies were sensitive to the experimental manipulation and increased monotonically from the CPT0 to the CPT2 condition.

Correlations between MA, P3 latencies, and P3 amplitudes are presented in Table 2. P3 amplitudes were not associated with MA in the three CPT conditions. The same was true for the P3 latency in the least demanding task condition. In both more demanding task conditions, P3 latency was negatively and significantly related to MA.

In a next step, we submitted the P3 latencies from the three task conditions to stepwise regression analyses to predict MA. In the first model with P3 latency in the CPT0 condition as a predictor of MA, neither the *β* coefficient, *β* = −0.073, *p* = 0.446, nor the amount of explained variance, adjusted *R*<sup>2</sup> = 0.000, were statistically significant. The second model with P3 latencies from the CPT0 and the CPT1 condition as predictors of MA led to a higher amount of explained variance, adjusted *R*<sup>2</sup> = 0.043, with the *R*<sup>2</sup> change being significant, *F*(1,107) = 6.234, *p* = 0.014. The *β* coefficient of P3 latency in the CPT1, *β* = −0.234, *p* = 0.014, but not the *β* coefficient of P3 latency in the CPT0 condition, *β* = −0.067, *p* = 0.479, yielded statistical significance. Comparing these first two models, P3 latency in the CPT1 condition explained a significantly larger portion of variance in MA compared to P3 latency in the CPT0 condition. Finally, with the third model, a statistically significant additional increase in the amount of explained variance was obtained (adjusted *R*<sup>2</sup> = 0.101) compared to the second model, *F*(1,106) = 7.994, *p* = 0.006. Only the *β* coefficient of the P3 latency in the CPT2 condition, *β* = −0.297, *p* = 0.006, but not the *β* coefficients of P3 latency in the CPT1, *β* = −0.085, *p* = 0.421, and in the CPT0 condition, *β* = −0.085, *p* = 0.355, were statistically significant.

Also given in Table 2 are correlations among RT, P3 latency, and P3 amplitude. P3 latencies in the CPT1 and in the CPT2 conditions were significantly correlated with each other but not with P3 latency in the CPT0 condition. This result indicated a functional difference of P3 latency in the CPT0 compared to the CPT1 and CPT2 conditions. P3 latencies in the three CPT conditions were positively correlated with RTs, but only four out of the nine correlation coefficients yielded statistical significance. Correlations between P3 latencies and P3 amplitudes were all negative, but only four out of nine coefficients reached statistical significance. Finally, the P3 amplitudes in the three CPT conditions correlated significantly positively with each other, but negatively with RTs, except for the correlation between the P3 amplitude in the CPT0 condition and RT in the CPT2 condition (see Table 2).
