EEG decoding of finger numeral configurations with machine learning

In this study, we used multivariate decoding methods to study processing differences between canonical (montring and count) noncanonical finger numeral configurations (FNCs). While previous research investigated these using behavioral event-related potentials (ERP) methods, conventional univariate ERP analyses focus on specific time intervals electrode sites fail capture broader scalp distribution EEG frequency patterns. To address issue a supervised learning classifier—support vector machines (SVM)—was decode distributions alpha-band power for montring, counting, FNCs (for integers 1 4). The SVM was test whether the numerical information presented in can be decoded from data. Differences magnitude timing of accuracy rates were compare three types FNCs. Overall, algorithm able predict beyond random chance level accuracy, with higher than alpha-power. Montring had lower peak compared counting configurations, likely due automaticity montring leading less distinct four magnitudes (1 Paralleling response data, earlier (472 ms), (577 ms) (604 ms). results provide support being processed automatically, somewhat similar number symbols, additional insights across different forms This also highlights strengths EEG/ERP cognition.