The N100 of Averaged ERPs Predicts LDA Classifier Success on an Individual Basis

We examine the success of a common LDA classification algorithm for P300 Speller BCI systems on an individual basis. Experiments performed on 16 subjects (7 with severe motor impairment, 9 with no motor impairment) indicate that the P300 Speller should, on average, work in both client and neurotypical populations. Here, we find that the N100 of an averaged ERP, a measure associated with selective attention, has a significant relationship with LDA classifier results, and may account for a large portion of variability we see in individual success in operating P300 Spellers.