as the P3. Outliers with respect to amplitude and latency were identified and their data visually screened to determine whether outlyingness was due to problems necessitating subject exclusion. In addition, we identified multivariate outliers using the three parietal electrodes and a robust version of Mahalanobis distance from the robustbase package (Rousseeuw et al., 2011) in the statistical computing environment (R Development Core Team, 2010). The corresponding data were examined visually, and the algorithm’s selection overridden if necessary.
