ERP analyses
An examination of the scalp topo-maps of the grand-averaged data revealed that the N2 component was right-lateralized on no-go trials, consistent with previous studies with children and adults (e.g., Bokura et al., 2001; Todd et al., 2008), but also that N2 amplitudes were larger for go trials than for no-go trials over the left frontal part of the scalp. Therefore, rather than create no-go minus go difference waveforms, we compared N2 waveforms for go and no-go trials as a function of laterality. Data were averaged across a right-hemisphere fronto-central cluster of electrodes that included HydroCel electrodes 112, 111, 118, 117, and 124 (encompassing F4 in the 10–20 System), and across a left-hemisphere fronto-central cluster that included electrodes 13, 29, 20, 28, and 24 (encompassing F3); see Figure 2.
Figure 2  Hydrocel electrode sites contributing to right- and left-lateralized N2 waveforms and data.

N2 results
N2 amplitudes and latency were analyzed using an ANOVA with cultural group as a between-subjects variable and go/no-go condition (go trials, no-go trials) and laterality (right, left) as within-subjects variables. Post hoc pairwise comparisons with Bonferroni adjustments were carried out for all significant interactions. Figure 3 presents the grand-averaged waveforms of the N2 component at the corresponding pair of left-right electrode sites that best illustrate the effects reported below: electrode site 112 (on the right) for the no-go condition and electrode site 13 (on the left) for the go condition. Waveforms are also shown for all other sites in the montage.
Figure 3  Stimulus-locked grand-averaged ERP waveforms for the two cultural groups. (A) Data from fronto-central electrode site 112 (right-hemisphere) for the no-go condition, accompanied by waveforms from electrode sites 111, 117, 118, 124 (right-hemisphere) for the no-go condition. (B) Data from fronto-central electrode site 13 (left-hemisphere) for the go condition, accompanied by waveforms from electrode sites 24, 28, 20, 29 (left-hemisphere) for the go condition. No significant effects or interactions were found for N2 latencies (ηp2<0.07 in all cases). The ANOVA on N2 amplitudes revealed a marginally significant main effect for cultural group, F(1, 35) = 3.78, p < 0.06, ηp2=0.10 (Chinese M = −6.01 μV, SE = 0.89; European M = −3.67 μV, SE = 0.82), and a significant effect of laterality, F(1, 35) = 15.05, p < 0.001, ηp2=0.30 (Right M = −5.72 μV, SE = 0.65; Left M = −3.95 μV, SE = 0.64). These effects were qualified by a significant Go/no-go condition × Laterality interaction, F(1, 35) = 110.64, p < 0.001, ηp2=0.76, and a significant Cultural group × Go/no-go condition × Laterality interaction, F(1, 35) = 5.40, p < 0.05, ηp2=0.13. Post hoc tests examining the effect of cultural group revealed that Chinese-Canadian children showed larger (i.e., more negative) N2 amplitudes than European-Canadian children on no-go trials on the right side of the scalp, p < 0.05 (Chinese M = −9.13 μV, SE = 1.14; European M = −5.79 μV, SE = 1.05; Figure 4A), as well as on go trials on the left side of the scalp, p < 0.05 (Chinese M = −7.66 μV, SE = 0.96; European M = −4.59 μV, SE = 0.89; Figure 4B), but that there were no group differences on no-go trials on the left or on go trials on the right. The three-way interaction reflected the fact that Chinese-Canadian children showed a stronger effect of laterality on no-go trials (ηp2=0.59) than did European-Canadian children on go trials (ηp2=0.16).
Figure 4  Cultural group differences in N2 amplitudes (μV) by go/no-go condition. (A) Mean amplitudes for the right-hemisphere fronto-central cluster of electrodes (Hydrocel electrodes 112, 111, 118, 117, and 124). (B) Mean amplitudes for the left-hemisphere fronto-central cluster of electrodes (Hydrocel electrodes 13, 29, 20, 28, and 24).

N2 amplitude lateralization scores
To assess the degree of lateralization of the N2, right-hemisphere N2 amplitudes were subtracted from left-hemisphere N2 amplitudes, and then these difference scores were analyzed using an ANOVA with cultural group as a between-subjects variable and go/no-go condition as a within-subjects variable. There was a main effect of go/no-go condition, F(1, 35) = 110.64, p < 0.001, ηp2=0.76, and a Cultural group × Go/no-go condition interaction, F(1, 35) = 5.4, p < 0.05, ηp2=0.13. Examination of means revealed that for both cultural groups, left-minus-right difference scores were positive for no-go trials (Chinese M = 6.98, SE = 0.98; European M = 4.39, SE = 0.91), reflecting larger (i.e., more negative) amplitudes on the right, and negative for go trials (Chinese M = −2.57, SE = 0.72; European M = −1.72, SE = 0.67), reflecting larger (i.e., more negative) amplitudes on the left. Pairwise comparisons indicated that the effect of go/no-go condition was significant for both cultural groups, but that the size of this effect was larger for the Chinese-Canadian participants, ηp2=0.69 than for the European-Canadian participants, ηp2=0.51. These findings suggest that the N2 was more strongly lateralized (to the left for go trials and to the right for no-go trials) for the Chinese-Canadian participants.

N2 amplitudes: relations with behavioral performance
A total of eight Pearson correlations were conducted to examine relations among mean N2 amplitude for the left- and right-hemisphere electrode clusters, median RTs, and no-go accuracy cost. Significant correlations were found between median RTs on the go/no-go task and the following: right-hemisphere N2 amplitudes on no-go trials, r = 0.30, p < 0.05, left-hemisphere N2 amplitudes on go trials, r = 0.28, p < 0.05, and left-hemisphere N2 amplitude on no-go trials, r = 0.35, p < 0.05. Larger (i.e., more negative) N2 amplitudes were associated with faster responding on go trials. Correlations involving no-go accuracy cost were not significant (−0.13 < r < 0.11). These correlations indicate that larger N2 amplitudes are related to better EF performance.

Source analysis
We used a minimum norm method with the local autoregressive average (LAURA) to create a source model of the grand-averaged scalp data. Modeled source activation was then examined using GeoSource (EGI) for latencies between 250 and 500 ms post-stimulus. Given that relatively little is known about the sources underlying the N2 in young children, an underdetermined model was used (Michel et al., 2004; Luck, 2005). Regions of interest (ROIs) were defined functionally around the voxels of peak activation in the model, and also in light of source analyses of the N2 in studies conducted with North American children (Lamm et al., 2006; Lewis et al., 2006, 2008; Todd et al., 2008). Activation was then averaged across all voxels in each ROI for each participant in each go/no-go condition, and a single-source waveform was extracted for each ROI. This process yielded source activation waveforms for four hypothetical generators of scalp activation located generally in ventromedial prefrontal cortex, right ventrolateral prefrontal cortex, left ventrolateral prefrontal cortex, and dorsomedial prefrontal cortex. Figure 5 shows means and standard deviations of activation levels (in nA) extracted from these modeled sources for the two cultural groups for the peak N2 interval between 300 and 350 ms post-stimulus. To investigate cultural group differences in extracted activation levels for each of the four modeled sources, separate ANOVAs were carried out with cultural group as a between-subjects variable, and go/no-go condition and 50-ms interval as within-subjects variables.
Figure 5  Modeled source activations (in nA) and ROIs displayed using the Montreal Neurological Institute (MNI) average adult MRI scan for the peak N2 interval of 300–350 ms in ventromedial prefrontal cortex (VMPFC; −3, 38, −20), right ventrolateral prefrontal cortex (VLPFC: 32, 17, −20), left VLPFC (−31, 17, −20), and dorsomedial prefrontal cortex (DMPFC: −3, 24, 36), as a function of cultural group and go/no-go condition. GeoSource voxel locations (x, y, z) in Talairach space are provided for the peak activation in each ROI. For the VMPFC modeled source, there were significant main effects of cultural group, F(1, 35) = 23.79, p < 0.001, ηp2=0.41 (Chinese M = 0.45, SE = 0.027; European M = 0.28, SE = 0.025) and go/no-go condition, F(1, 35) = 48.23, p < 0.001, ηp2=0.58 (go M = 0.28, SE = 0.015; no-go M = 0.45, SE = 0.027), with greater overall activation on no-go trials and for Chinese children. There were no interactions.
For the right VLPFC source, there were significant main effects of cultural group, F(1, 35) = 13.96, p < 0.001, ηp2=0.29 (Chinese M = 0.27, SE = 0.016; European M = 0.19, SE = 0.014), go/no-go condition, F(1, 35) = 197.52, p < 0.001, ηp2=0.90 (go M = 0.12, SE = 0.007; no-go M = 0.33, SE = 0.017), and interval, F(4, 140) = 21.08, p < 0.001, ηp2=0.38 (250–300 ms M = 0.19, SE = 0.011; 300–350 ms M = 0.23, SE = 0.012; 350–400 ms M = 0.24, SE = 0.011; 400–450 ms M = 0.23, SE = 0.010; 450–500 ms M = 0.24, SE = 0.010) with greater activation for Chinese children; no-go trials; and the interval of 250–300 ms vs. all others. These main effects were qualified by a Cultural group × Go/no-go condition interaction, F(1, 35) = 6.91, p < 0.05, ηp2=0.17, a Go/no-go condition × Interval interaction, F(4, 140) = 25.28, p < 0.001, ηp2=0.42, and a Cultural group × Go/no-go condition × Interval interaction, F(4, 132) = 3.10, p < 0.05, ηp2=0.09. Post hoc tests revealed that the difference between go and no-go trials (i.e., no-go > go) for this right VLPFC source was larger for Chinese-Canadian children than for European-Canadian children, that the difference between go and no-go trials was largest between 350 and 400 ms, and that the Cultural group × Go/no-go condition interaction was especially pronounced for the two intervals between 300 and 400 ms.
For the left VLPFC source, there were main effects of cultural group, F(1, 35) = 22.49, p < 0.001, ηp2=0.39 (Chinese M = 0.26, SE = 0.015; European M = 0.17, SE = 0.013), go/no-go condition, F(1, 35) = 22.39, p < 0.001, ηp2=0.39 (go M = 0.24, SE = 0.012; no-go M = 0.18, SE = 0.012), and interval, F(4, 140) = 10.70, p < 0.001, ηp2=0.23 (250–300 ms M = 0.19, SE = 0.011; 300–350 ms M = 0.21, SE = 0.011; 350–400 ms M = 0.22, SE = 0.011; 400–450 ms M = 0.22, SE = 0.010; 450–500 ms M = 0.23, SE = 0.010) with greater activation for Chinese children and for go trials, and differences between 250–300 ms vs. all other intervals, and between 400–450 ms and 450–500 ms). These were qualified by a Go/no-go condition × Interval interaction, F(4, 140) = 42.50, p < 0.001, ηp2=0.55, and a Cultural group × Go/no-go condition × Interval interaction, F(4, 140) = 2.63, p < 0.05, ηp2=0.07, reflecting the fact that the difference between go and no-go trials (i.e., go > no-go) for this left VLPFC source was larger for Chinese-Canadian children, and that this two-way interaction was especially pronounced for the two intervals between 300 and 400 ms.
Finally, for the DMPFC source, there were main effects of cultural group, F(1, 35) = 14.73, p < 0.001, ηp2=0.30 (Chinese M = 0.34, SE = 0.020; European M = 0.24, SE = 0.019), go/no-go condition, F(1, 35) = 7.72, p < 0.01, ηp2=0.18 (go M = 0.27, SE = 0.013; no-go M = 0.31, SE = 0.019), and interval, F(4, 140) = 6.38, p < 0.001, ηp2=0.15 (250–300 ms M = 0.27, SE = 0.014; 300–350 ms M = 0.30, SE = 0.015; 350–400 ms M = 0.30, SE = 0.016; 400–450 ms M = 0.29, SE = 0.014; 450–500 ms M = 0.28, SE = 0.014) with greater activation for Chinese children and for no-go trials, and a difference between 250–300 ms vs. both 300–350 ms, 350–400 ms. These were qualified by a Go/no-go condition × Interval interaction, F(4, 140) = 4.40, p < 0.01, ηp2=0.11, and a Cultural group × Interval interaction, F(4, 140) = 4.87, p < 0.001, ηp2=0.12. Post hoc tests revealed that Chinese-Canadians showed more activation than European-Canadians especially between 250 and 350 ms, and that the differences between go and no-go trials (i.e., no-go > go) for this dorsomedial prefrontal region of interest was largest between 250 and 300 ms.