In previous studies, after artifact attenuation the EEG data were typically used for ERP analysis, based on electromagnetic source localizations at different latencies (Goldstein et al., 2002). Accordingly, the single epochs were averaged, and therefore their variability was not taken into account. In our study, this analysis revealed an important contribution of the P300 component to the ERP, in particular at the centro-parietal electrodes, whereas we generally observed weak P1 and N1 responses, probably due to our specific experimental paradigm. Since we observed a SNR sufficient for single-trial analysis of the P300 component, we measured its trial-by-trial response fluctuations. This information, considered to indirectly allow off-line monitoring of target detection performance (Debener et al., 2005), was particularly valuable. Other authors reported that the P300 amplitude variability was affected by inter-stimulus interval (Sambeth et al., 2004), target-to-target interval (Gonsalvez and Polich, 2002), or target probability (Croft et al., 2003). However, no clear indication about the cerebral mechanisms accounting for the P300 response fluctuations observed during the oddball task could be provided in these studies. We analyzed the trial-by-trial responses recorded by
