Reappraisal of filter effects on P300 voltage and latency.

The selection of which high-pass filter cutoff to use in P300 studies continues to be a serious methodological consideration. To determine whether there was an optimal range of bandpass widths-a range in which P300 voltage and latency would not show statistically significant differences-the authors recorded P300 responses to the auditory "oddball" paradigm from Cz and Pz electrodes in a group of eight healthy males. The authors used high-pass filter cutoffs of 0.01, 0.1, 0.3, and 1.0 Hz with low-pass filter cutoffs of 30 and 100 Hz and measured both P300 peak voltages and P300 integrated mean voltages at 12 bandpass settings. There were statistically significant differences in 7 out of 12 bandpass comparisons for P300 peak voltages and in 7 out of 12 bandwidth comparisons for P300 integrated mean voltages. For P300 latencies, there were statistically significant differences in 9 out of 12 bandwidth comparisons. Based on these results, the best recommendation, therefore, is that the high-pass filter be set no higher than 0.3 Hz.